Compare commits

...

21 Commits

Author SHA1 Message Date
admin e415bf5ba5 fix: color_xrange 范围限定 d_max + color_driver + color 范围改为 [0,1] + bug 修复 2026-06-24 16:09:39 +08:00
admin b50c55d0a3 docs: examples/Readme.html v2.1 更新引擎 DLL 架构说明 2026-06-21 17:05:19 +08:00
admin 75d8aae1f2 fix: 修复 DLL 文件被 git CRLF 转换损坏
将 DLL 标记为 binary,禁止行尾转换
2026-06-21 08:17:08 +08:00
admin dd6fbdfccf chore: .gitattributes 标记 DLL 为二进制文件 2026-06-21 08:16:43 +08:00
admin eea62de479 fix: case05 atom 1 z=0(起点拉回链平面) 2026-06-21 08:13:23 +08:00
admin 1261398be2 fix: case05 启用 step_simulate=1 2026-06-21 08:11:34 +08:00
admin 369b840cf2 fix: Makefile Windows 检测兼容 Msys2/MINGW 环境
uname -s 在 Msys2 上返回 MINGW64_NT-* 而非 Windows,
添加 findstring 匹配确保静态链接选项生效
2026-06-21 05:58:37 +08:00
admin 7fb2b730e9 refactor: 源码迁至 engines/src/,删除旧的 engines/{c,cpp,fortran}/
- 删除 engines/{c,cpp,fortran}/ 目录(源码和 Makefile 已移至 src/)
- engines/src/{c,cpp,fortran}/: 清理 main.*/bak 等无用文件
- Makefile 改为直接编译 DLL 到 engines/release/
- .gitignore: 更新路径指向 engines/src/*/build/
- engine_dll.py: 更新注释中的编译命令路径
2026-06-21 05:42:29 +08:00
admin e371fa8db1 refactor: 移除旧子进程模式的 main.* 源文件,Makefile 只编译 DLL
- 删除 engines/c/main.c, engines/cpp/main.cpp, engines/fortran/main.f90
- Makefile 移除 all/exe 构建目标,默认只编译 DLL
2026-06-21 05:33:52 +08:00
admin 7468a2d458 chore: 清理引擎目录旧校准缓存 _calib_cache.json
旧子进程模式的预校准缓存,切换到 DLL 路径后已无代码引用
2026-06-21 05:28:24 +08:00
admin 76a26d888e chore: 清理旧的引擎参数/校准 JSON 文件
- c.json / fortran.json: 旧子进程模式参数,DLL 路径不再需要
- _calib_c.json / _calib_fortran.json: 旧预校准缓存,DLL 路径实时跟踪无需预跑
2026-06-21 05:25:31 +08:00
admin 8183cbc22e chore: .gitignore 去掉全局 *.dll 限制,release DLL 可自由提交 2026-06-21 05:21:22 +08:00
admin 41c454eeab refactor: 引擎只走 DLL 路径,release 目录清理 exe
- dynamics.py: 去掉 exe 子进程回退,只走 DLL 路径
- .gitignore: 白名单 engines/release/dynamics_*.dll
- engines/release/: 移除 dynamics_*.exe,新增 3 个引擎 DLL
2026-06-21 05:19:33 +08:00
admin 6c06c00f4b fix: Fortran engine use_marker/alpha header passthrough + case04 Moon orbit fix
- engines/fortran/main.f90:
  - 新增 use_marker 和 alpha_str 的读取、传递、写入 display.txt header
  - 新增 json_get_alpha 函数兼容 JSON 单值或 6 元数组
- engines/release/: 引擎校准缓存 & fortran.json 参数
- examples/case04/: 修正月球位置 (11.5→10.5) 和速度 (526→647),
  地月距缩至希尔半径以内 (0.5 < 1.0), 新增希尔半径验证说明
- examples/Readme.html: case04/case09 描述更新
2026-06-19 08:05:05 +08:00
admin c8d12e7f45 docs: case04 Readme.html 新增月球绕地轨道交互图
- 月球轨道图使用绿色配色,与地球轨道图(蓝色)区分
- 默认偏心率 e=0.0549(月球真实值)
- 标注半长轴 a / 半短轴 b / 偏心距 c
- 标注近地点/远地点距离
- 地球在左焦点,月球在椭圆轨道运行
- 独立滑块控制月球偏心率
2026-06-18 23:06:55 +08:00
admin 7078cbc744 fix: 轨道示意图太阳移至左焦点,默认 e=0.017(地球真实 0.0167)
- 椭圆中心改为 cx+f,使左焦点=cx 为太阳
- 偏心距标注从椭圆中心指向左焦点
- a/b 标注、近日点/远日点位置同步修正
- 滑块映射改为 e=value/1000(精度 0.001)
- 滑块默认值 17 → e=0.017 ≈ 地球真实 0.0167
2026-06-18 23:05:23 +08:00
admin d4ce23a07e feat: case04 以月球质量=1 为基准更新质量比和质量缩放
- coord.txt: 太阳 27000 / 地球 81 / 月球 1(月球=1 质量单位)
- 地球绕日轨道速度 v=520(v = sqrt(g_strength * M_sun / r))
- 月球绕地速度 v=526(地球速度 + 6 绕地附加速度)
- box_a: 20 → 30(容纳更大轨道半径)
- Readme.html 同步更新质量表和初始构型表
- 已知局限说明太阳质量缩至 1/1000 的原因
2026-06-18 23:03:36 +08:00
admin 84aed813df docs: case04 Readme.html 新增交互式轨道示意图
使用 HTML5 Canvas + JavaScript 绘制椭圆轨道图,
支持滑块调整偏心率 e(0~0.85),实时更新:
- 椭圆形状变化
- a/b/c 参数标注
- 近日点/远日点数值
- 速度矢量方向
- 底部信息面板
2026-06-18 23:01:04 +08:00
admin f6c468702f docs: case04 新增 Readme.html 含 MathJax 公式
涵盖天体质量/轨道参数/偏心率/开普勒定律/近日点速度
公式等全部天体力学内容,使用 MathJax 3 (SVG) 渲染。
2026-06-18 22:59:07 +08:00
admin 8b82d3c4e4 fix: leapfrog 模式缺少边界反弹导致粒子穿模 2026-06-18 14:28:56 +08:00
admin 80b7467421 fix: leapfrog 模式缺少边界反弹,粒子穿透地板
问题:leapfrog 模式的主循环直接调用 leapfrog_staggered_step,
未经过 apply_motion_update,因此 Limit_in_box(边界反弹)
从未被执行。粒子超出 z<-20 后被 wrap_position 回绕到顶端,
导致物理行为完全错误。

修复:
1. 在预热和记录两个循环中,leapfrog 路径后显式调用
   Limit_in_box(与 apply_motion_update 内的非 leapfrog 路径一致)
2. 移除 apply_motion_update 末尾重复的 Limit_in_box 调用
   (统一到主循环中执行)
2026-06-18 09:28:38 +08:00
119 changed files with 361409 additions and 8377 deletions
+8 -6
View File
@@ -1,7 +1,9 @@
# 全部使用 LF 换行,仓库内外一致,不随系统自动转换
* text eol=lf
# 引擎 DLL 是二进制文件,禁止行尾转换
engines/release/*.dll binary
# 二进制文件不转换
*.png binary
*.jpg binary
*.ico binary
# Python 源码使用 LF
*.py text eol=lf
# Makefile 使用 LF
Makefile text eol=lf
*.mk text eol=lf
+4 -6
View File
@@ -18,10 +18,9 @@ pip-wheel-metadata/
venv/
ENV/
# ── C / C++ 编译产物 ─────────────────────────────────────────
# Makefile 构建输出(engines/c/build/
engines/c/build/
engines/cpp/build/
# ── C / C++ / Fortran 编译产物 ────────────────────────
# 源码目录 engines/src/*/ 中可能产生的构建输出
engines/src/*/build/
# CMake 构建目录(根目录或自定义 build 目录)
CMakeCache.txt
@@ -42,12 +41,11 @@ build_*/
*.so
*.so.*
*.dylib
*.dll
# 可执行文件(保留源码,排除编译出的二进制)
# 注意:Windows 下 .exe 后缀的可执行文件
*.exe
# engines/c/Makefile 里指定了 build/ 目录,已由上面覆盖
# 可在 engines/src/*/ 中用 make dll 编译引擎 DLL
# 运行时生成的引擎参数文件(每次运行都会覆盖)
engines/*/param.json
+18 -4
View File
@@ -668,12 +668,20 @@ def load_driver_file(driver_path, atom_ids):
return None
print(f"[compute] 已加载驱动力: {len(drivers)} 条定义")
if len(drivers) <= 20:
for d in drivers:
print(f" 原子 {d['atom_id']}: "
f"A=({d['amp'][0]},{d['amp'][1]},{d['amp'][2]}), "
f"f=({d['freq'][0]},{d['freq'][1]},{d['freq'][2]}), "
f"φ=({phi_deg[d['amp'].tolist().index(max(d['amp']))]}° 等), "
f"period={d['period_str']}")
else:
_first = drivers[0]
_last = drivers[-1]
print(f" 原子 {_first['atom_id']}~{_last['atom_id']}: "
f"A=({_first['amp'][0]},{_first['amp'][1]},{_first['amp'][2]}), "
f"f=({_first['freq'][0]},{_first['freq'][1]},{_first['freq'][2]}), "
f"period={_first['period_str']}{len(drivers)}")
return drivers
@@ -1070,6 +1078,10 @@ def run_engine_dll(engine, output_dir, config):
"number_of_particles": str(n_atoms),
# draw.py 需要的渲染参数
"use_marker": str(use_marker),
"display_color": json.dumps(config.get("display_color",
{"x":[0,[1.0,0.0,0.0]],"y":[0,[0.0,1.0,0.0]],"z":[0,[0.0,0.0,1.0]],
"xy":[0,[1.0,1.0,0.0]],"yz":[0,[0.0,1.0,1.0]],"zx":[0,[1.0,0.0,1.0]],
"xyz":[1,[1.0,1.0,1.0]]})),
"ball_radius": str(config.get("ball_radius", float(ATOM_RADII[0]) if ATOM_RADII is not None else 0.5)),
"ball_color_r": str(config.get("ball_color_r", 0.9)),
"ball_color_g": str(config.get("ball_color_g", 0.2)),
@@ -1798,10 +1810,6 @@ def apply_motion_update(x, y, z, vx, vy, vz, dt, m, g, b):
x, y, z, vx, vy, vz = Leapfrog_Method(x, y, z, vx, vy, vz, dt, m, g, b)
else:
raise ValueError(f"未知算法: {METHOD}")
x, vx = Limit_in_box(x, X_MIN, X_MAX, vx)
y, vy = Limit_in_box(y, Y_MIN, Y_MAX, vy)
z, vz = Limit_in_box(z, Z_MIN, Z_MAX, vz)
return x, y, z, vx, vy, vz
@@ -1870,6 +1878,9 @@ def run_simulation(save_trajectory=0):
x, y, z, vx, vy, vz, DT, ATOM_MASSES, G, B)
else:
x, y, z, vx, vy, vz = apply_motion_update(x, y, z, vx, vy, vz, DT, ATOM_MASSES, G, B)
x, vx = Limit_in_box(x, X_MIN, X_MAX, vx)
y, vy = Limit_in_box(y, Y_MIN, Y_MAX, vy)
z, vz = Limit_in_box(z, Z_MIN, Z_MAX, vz)
x, y, z = wrap_position(x, y, z)
x, y, z, vx, vy, vz = apply_fixed_constraints(x, y, z, vx, vy, vz)
print(
@@ -1924,6 +1935,9 @@ def run_simulation(save_trajectory=0):
x, y, z, vx, vy, vz, DT, ATOM_MASSES, G, B)
else:
x, y, z, vx, vy, vz = apply_motion_update(x, y, z, vx, vy, vz, DT, ATOM_MASSES, G, B)
x, vx = Limit_in_box(x, X_MIN, X_MAX, vx)
y, vy = Limit_in_box(y, Y_MIN, Y_MAX, vy)
z, vz = Limit_in_box(z, Z_MIN, Z_MAX, vz)
x, y, z = wrap_position(x, y, z)
x, y, z, vx, vy, vz = apply_fixed_constraints(x, y, z, vx, vy, vz)
+269 -49
View File
@@ -45,7 +45,65 @@ if os.path.exists(npz_path):
disp_data = compute.load_display_npz(npz_path)
else:
disp_data = compute.load_display_txt(disp_path)
h = disp_data["header_fields"]
# ── 从 input.txt 读取参数(替代 display.npz 中的 meta)──
try:
import yaml
_have_yaml = True
except ImportError:
_have_yaml = False
input_dir = os.path.join(os.path.dirname(output_dir), "input")
input_path = os.path.join(input_dir, "input.txt")
if _have_yaml and os.path.exists(input_path):
try:
with open(input_path, "r", encoding="utf-8") as f:
config = yaml.safe_load(f)
except Exception:
config = {}
else:
config = {}
# 兼容旧版:若 input.txt 不存在或解析失败,降级到 display.npz 的 meta
if not config:
config = disp_data.get("header_fields", {})
# ── 从 coord.txt 读取平衡位置 ─────────────────
# 注意: config 中的 *_file 可能带 "input/" 前缀,但 input_dir 已指向 input/ 目录
_coord_file_raw = config.get("coord_file", "coord.txt")
coord_path = os.path.join(input_dir, os.path.basename(_coord_file_raw))
if os.path.exists(coord_path):
try:
_ids, _masses, _radii, _pos, _vel, _fixed = compute.load_coord_file(coord_path)
EQ_POS = _pos # (n_atoms, 3)
ATOM_FIXED = _fixed # (n_atoms, 3)
except Exception:
EQ_POS = None
ATOM_FIXED = None
else:
EQ_POS = None
ATOM_FIXED = None
# ── 从 connection.txt 读取成键信息(若 meta 中没有)──
BOND_PAIRS = disp_data.get("bond_pairs", [])
if not BOND_PAIRS and 'bond_pairs' not in disp_data:
conn_path = os.path.join(input_dir, os.path.basename(config.get("connection_file", "connection.txt")))
if os.path.exists(conn_path):
try:
_pairs = []
with open(conn_path, "r", encoding="utf-8") as _f:
_f.readline() # skip header
for _line in _f:
_line = _line.strip()
if not _line or _line.startswith("#"):
continue
_parts = _line.split()
if len(_parts) >= 2:
_pairs.append([int(_parts[0]) - 1, int(_parts[1]) - 1])
BOND_PAIRS = np.array(_pairs, dtype=np.int32) if _pairs else []
except Exception:
BOND_PAIRS = []
BOND_PAIRS = BOND_PAIRS.tolist() if hasattr(BOND_PAIRS, 'tolist') else BOND_PAIRS
# 全原子帧数据
DISP_ALL_X = disp_data["frames_x"] # (n_frames, n_atoms)
@@ -66,64 +124,220 @@ DISP_VZ = DISP_ALL_VZ[:, 0]
N_FRAMES = DISP_ALL_X.shape[0]
NT = int(disp_data["n_total_frames"])
N_ATOMS = int(disp_data["n_total_particles"])
DT = float(h.get("DT", 0.001))
DT = float(config.get("DT", 0.001))
# 视觉位移放大:display_amp: [ax, ay, az],对偏离第0帧的位移乘以倍数
_damp_raw = h.get("display_amp", "")
if _damp_raw.strip():
# 视觉位移放大:display_amp: [ax, ay, az],对偏离平衡位置的位移乘以倍数
_damp_raw = config.get("display_amp", "")
if isinstance(_damp_raw, (list, tuple)):
_damp = np.array(_damp_raw, dtype=np.float64)
elif isinstance(_damp_raw, str) and _damp_raw.strip():
import ast as _ast
_damp_vals = _ast.literal_eval(_damp_raw.strip())
_damp = np.array(_damp_vals, dtype=np.float64)
if _damp.shape == (3,) and not np.allclose(_damp, 1.0):
_damp = np.array(_ast.literal_eval(_damp_raw.strip()), dtype=np.float64)
else:
_damp = None
if _damp is not None and _damp.shape == (3,) and not np.allclose(_damp, 1.0):
if EQ_POS is not None:
_eq_x = EQ_POS[None, :, 0] # coord.txt 平衡位置
_eq_y = EQ_POS[None, :, 1]
_eq_z = EQ_POS[None, :, 2]
else:
_eq_x = DISP_ALL_X[0:1, :] # 第0帧作为平衡位置参考
_eq_y = DISP_ALL_Y[0:1, :]
_eq_z = DISP_ALL_Z[0:1, :]
DISP_ALL_X = _eq_x + (DISP_ALL_X - _eq_x) * _damp[0]
DISP_ALL_Y = _eq_y + (DISP_ALL_Y - _eq_y) * _damp[1]
DISP_ALL_Z = _eq_z + (DISP_ALL_Z - _eq_z) * _damp[2]
NSTEP = int(h.get("NSTEP", 1))
# ── 位移颜色映射 ──────────────────────────────
# display_color: { mode: [enabled, [R,G,B]], ... }
# mode: x, y, z, xy, yz, zx, xyz
# enabled: 0=off, 1=on
# [R,G,B]: 最大位移时的颜色(渐变起点为白色)
# 例:xyz: [1, [1.0,1.0,1.0]] → 三方向合成位移 → 白色渐变
FRAME_COLORS = None
_DC_MODE = None
_DC_COLOR = None
_raw_dc = config.get("display_color", "")
if _raw_dc:
try:
dc = json.loads(_raw_dc) if isinstance(_raw_dc, str) else _raw_dc
if isinstance(dc, dict):
for mode, (enabled, color) in dc.items():
if int(enabled) and mode in ("x","y","z","xy","yz","zx","xyz"):
_DC_MODE = mode
_DC_COLOR = np.array(color, dtype=np.float32)
break
except Exception:
pass
if _DC_MODE is not None and _DC_COLOR is not None:
if EQ_POS is not None:
eq_x = EQ_POS[None, :, 0] # coord.txt 平衡位置
eq_y = EQ_POS[None, :, 1]
eq_z = EQ_POS[None, :, 2]
else:
eq_x = DISP_ALL_X[0:1, :] # 第0帧
eq_y = DISP_ALL_Y[0:1, :]
eq_z = DISP_ALL_Z[0:1, :]
dx = DISP_ALL_X - eq_x
dy = DISP_ALL_Y - eq_y
dz = DISP_ALL_Z - eq_z
if _DC_MODE == "x":
disp_mag = np.abs(dx)
elif _DC_MODE == "y":
disp_mag = np.abs(dy)
elif _DC_MODE == "z":
disp_mag = np.abs(dz)
elif _DC_MODE == "xy":
disp_mag = np.sqrt(dx**2 + dy**2)
elif _DC_MODE == "yz":
disp_mag = np.sqrt(dy**2 + dz**2)
elif _DC_MODE == "zx":
disp_mag = np.sqrt(dz**2 + dx**2)
else: # xyz
disp_mag = np.sqrt(dx**2 + dy**2 + dz**2)
# ── color_xrange: 限定归一化基准的原子范围 ───────────
# 格式: [['min'|'mid'|'max'|数值, 'min'|'mid'|'max'|数值], ...] 对应 x,y,z
# 在此范围内的原子计算最大位移 d_max,所有原子以此基准归一化着色
_cxr = config.get("color_xrange", None)
_d_max_source = disp_mag # 默认:所有原子
_range_label = "全部原子"
if _cxr and isinstance(_cxr, (list, tuple)) and len(_cxr) == 3:
try:
# 获取坐标极值
if EQ_POS is not None:
_eq_all = EQ_POS
else:
_eq_all = np.column_stack([
DISP_ALL_X[0], DISP_ALL_Y[0], DISP_ALL_Z[0]])
_cmin = _eq_all.min(axis=0)
_cmax = _eq_all.max(axis=0)
_cmid = (_cmin + _cmax) / 2
_key_map = {"min": _cmin, "mid": _cmid, "max": _cmax}
_range_lo = np.zeros(3, dtype=np.float64)
_range_hi = np.zeros(3, dtype=np.float64)
for _i in range(3):
_lo = _cxr[_i][0]
_hi = _cxr[_i][1]
_range_lo[_i] = _key_map[_lo][_i] if _lo in _key_map else float(_lo)
_range_hi[_i] = _key_map[_hi][_i] if _hi in _key_map else float(_hi)
_in_x = (_eq_all[:, 0] >= _range_lo[0]) & (_eq_all[:, 0] <= _range_hi[0])
_in_y = (_eq_all[:, 1] >= _range_lo[1]) & (_eq_all[:, 1] <= _range_hi[1])
_in_z = (_eq_all[:, 2] >= _range_lo[2]) & (_eq_all[:, 2] <= _range_hi[2])
_color_mask = _in_x & _in_y & _in_z
_n_in_range = _color_mask.sum()
if _n_in_range > 0:
_d_max_source = disp_mag[:, _color_mask] # 仅在范围内找最大位移
_range_label = (f"x[{_range_lo[0]:.1f},{_range_hi[0]:.1f}] "
f"y[{_range_lo[1]:.1f},{_range_hi[1]:.1f}] "
f"z[{_range_lo[2]:.1f},{_range_hi[2]:.1f}]")
except Exception as _e:
print(f"[draw] color_xrange 解析失败: {_e}")
# 用(范围限定的)最大位移归一化,所有原子统一着色
d_max = _d_max_source.max()
if d_max > 1e-12:
disp_norm = disp_mag / d_max
else:
disp_norm = disp_mag
FRAME_COLORS = np.ones((N_FRAMES, N_ATOMS, 4), dtype=np.float32)
t = disp_norm[..., None]
FRAME_COLORS[..., :3] = 1.0 + (_DC_COLOR - 1.0) * t
print(f"[draw] 位移颜色映射: mode={_DC_MODE}, color={_DC_COLOR.tolist()}, "
f"d_max={d_max:.4f}, 范围: {_range_label}")
# ── color_driver: 驱动原子 → 统一颜色 ──
_cd_raw = config.get("color_driver", None)
if _cd_raw is not None:
try:
_cd = np.array(_cd_raw, dtype=np.float32)
# 读取 driver.txt 获取驱动原子 ID
_driver_file = config.get("driver_file", "driver.txt")
_driver_path = os.path.join(input_dir, os.path.basename(_driver_file))
if os.path.exists(_driver_path):
_driver_ids = []
with open(_driver_path, "r", encoding="utf-8") as _df:
_df.readline() # skip header
for _line in _df:
_line = _line.strip()
if not _line or _line.startswith("#"):
continue
_parts = _line.split()
if _parts:
_driver_ids.append(int(_parts[0]) - 1) # 1-based → 0-based
if _driver_ids:
FRAME_COLORS[:, _driver_ids, :3] = _cd
print(f"[draw] color_driver: {len(_driver_ids)} 驱动原子 → RGB{_cd_raw}")
except Exception as _e:
print(f"[draw] color_driver 解析失败: {_e}")
# ── color_fix: 全固定原子 (fix_x=fix_y=fix_z=1) → 统一颜色 ──
_cf_raw = config.get("color_fix", None)
if _cf_raw is not None and ATOM_FIXED is not None:
try:
_cf = np.array(_cf_raw, dtype=np.float32)
_full_fixed = (ATOM_FIXED[:, 0] == 1) & (ATOM_FIXED[:, 1] == 1) & (ATOM_FIXED[:, 2] == 1)
_n_fix = _full_fixed.sum()
if _n_fix > 0:
FRAME_COLORS[:, _full_fixed, :3] = _cf
print(f"[draw] color_fix: {_n_fix} 全固定原子 → RGB{_cf_raw}")
except Exception as _e:
print(f"[draw] color_fix 解析失败: {_e}")
NSTEP = int(config.get("NSTEP", 1))
DISP_STEP = np.arange(N_FRAMES) * NSTEP
DISP_T = DISP_STEP * DT
# 原子信息
ATOM_IDS = disp_data["atom_ids"]
# 优先使用 per-atom 半径,否则用统一的 ball_radius
_raw_radii = h.get("atom_radii", "")
_raw_radii = config.get("atom_radii", "")
if _raw_radii.strip():
ATOM_RADII = np.array([float(x) for x in _raw_radii.split(",")])
else:
ATOM_RADII = np.full(N_ATOMS, float(h.get("ball_radius", 0.5)))
ATOM_RADII = np.full(N_ATOMS, float(config.get("ball_radius", 0.5)))
PLOT_ATOM_ROW = 0
PLOT_ATOM_ID = int(ATOM_IDS[0])
BOND_PAIRS = [] # display 格式不含成键信息,从原始数据加载
# 成键信息已在上面从 connection.txt 加载
# 渲染方式:0=Sphere(网格球体), 1=Marker(GPU点精灵)
USE_MARKER = int(h.get("use_marker", 0))
USE_MARKER = int(config.get("use_marker", 0))
if N_FRAMES <= 0:
raise ValueError(
"output/display.txt 中没有可播放的帧,请检查 sample_start/sample_end/NSTEP 配置。")
# 保留模拟边界常量(用于场景缩放、相机等),从 output/display.txt 中读取
X_MIN = float(h.get("X_MIN", -10)); X_MAX = float(h.get("X_MAX", 10))
Y_MIN = float(h.get("Y_MIN", -10)); Y_MAX = float(h.get("Y_MAX", 10))
Z_MIN = float(h.get("Z_MIN", -10)); Z_MAX = float(h.get("Z_MAX", 10))
raw_alpha = h.get("alpha", "0.2")
# 模拟边界(从 input.txt 的 box_a 计算)
_box_a = float(config.get("box_a", 10.0))
X_MIN = -_box_a; X_MAX = _box_a
Y_MIN = -_box_a; Y_MAX = _box_a
Z_MIN = -_box_a; Z_MAX = _box_a
raw_alpha = config.get("alpha", "0.2")
if isinstance(raw_alpha, (list, tuple)):
alpha_list = [float(x) for x in raw_alpha]
else:
try:
alpha_list = [float(x) for x in raw_alpha.split(",")]
if len(alpha_list) != 6:
alpha_list = alpha_list * 6
except (ValueError, AttributeError):
alpha_list = [float(raw_alpha)] * 6
if len(alpha_list) != 6:
alpha_list = (alpha_list * 6)[:6]
# 绘图参数
ball_radius = float(h.get("ball_radius", 0.5))
ball_color_r = float(h.get("ball_color_r", 0.9))
ball_color_g = float(h.get("ball_color_g", 0.2))
ball_color_b = float(h.get("ball_color_b", 0.2))
box_color_r = float(h.get("box_color_r", 0.8))
box_color_g = float(h.get("box_color_g", 0.8))
box_color_b = float(h.get("box_color_b", 0.85))
ball_radius = float(config.get("ball_radius", 0.5))
ball_color_r = float(config.get("ball_color_r", 0.9))
ball_color_g = float(config.get("ball_color_g", 0.2))
ball_color_b = float(config.get("ball_color_b", 0.2))
box_color_r = float(config.get("box_color_r", 0.8))
box_color_g = float(config.get("box_color_g", 0.8))
box_color_b = float(config.get("box_color_b", 0.85))
# ===========================================================================
@@ -135,23 +349,23 @@ axis_length = 10.0
import math as _math_cam
_cx = float(h.get("camera_center_x", 0.0))
_cy = float(h.get("camera_center_y", 0.0))
_cz = float(h.get("camera_center_z", 0.0))
_cx = float(config.get("camera_center_x", 0.0))
_cy = float(config.get("camera_center_y", 0.0))
_cz = float(config.get("camera_center_z", 0.0))
# 若 input.txt 指定了摄像机自身坐标,则由坐标反推 distance/elevation/azimuth
if h.get("camera_pos_x") is not None:
_px = float(h["camera_pos_x"])
_py = float(h["camera_pos_y"])
_pz = float(h["camera_pos_z"])
if config.get("camera_pos_x") is not None:
_px = float(config.get("camera_pos_x"))
_py = float(config.get("camera_pos_y"))
_pz = float(config.get("camera_pos_z"))
_dx, _dy, _dz = _px - _cx, _py - _cy, _pz - _cz
_dist = _math_cam.sqrt(_dx*_dx + _dy*_dy + _dz*_dz) or 1.0
_elev = _math_cam.degrees(_math_cam.asin(max(-1.0, min(1.0, _dy / _dist))))
_azim = _math_cam.degrees(_math_cam.atan2(_dx, _dz))
else:
_dist = float(h.get("camera_distance", 40.0))
_elev = float(h.get("camera_elevation", 0))
_azim = float(h.get("camera_azimuth", 0))
_dist = float(config.get("camera_distance", 40.0))
_elev = float(config.get("camera_elevation", 0))
_azim = float(config.get("camera_azimuth", 0))
initial_camera = {
"distance": _dist,
@@ -212,11 +426,11 @@ axes_group.append(scene.visuals.Arrow(
parent=view.scene,
))
axes_group.append(scene.visuals.Text(text="x", color=(1.0, 0.2, 0.2, 1.0), font_size=18,
axes_group.append(scene.visuals.Text(text="x", color=(1.0, 0.2, 0.2, 1.0), font_size=14,
pos=(axis_length + 0.2, 0, 0), anchor_x="left", anchor_y="center", parent=view.scene))
axes_group.append(scene.visuals.Text(text="y", color=(0.2, 1.0, 0.2, 1.0), font_size=18,
axes_group.append(scene.visuals.Text(text="y", color=(0.2, 1.0, 0.2, 1.0), font_size=14,
pos=(0, axis_length + 0.2, 0), anchor_x="left", anchor_y="bottom", parent=view.scene))
axes_group.append(scene.visuals.Text(text="z", color=(0.3, 0.6, 1.0, 1.0), font_size=18,
axes_group.append(scene.visuals.Text(text="z", color=(0.3, 0.6, 1.0, 1.0), font_size=14,
pos=(0, 0, axis_length + 0.2), anchor_x="left", anchor_y="bottom", parent=view.scene))
# ── 原子渲染 ──────────────────────────────────
@@ -235,8 +449,11 @@ TAB10_RGB = np.array([
[0.7373, 0.7412, 0.1333], # 黄绿
[0.0902, 0.7451, 0.8118], # 青
])
# 每个原子的颜色(循环使用 tab10 色板)
# 每个原子的颜色(循环使用 tab10 色板,或按位移着色
atom_colors = np.zeros((N_ATOMS, 4), dtype=np.float32)
if FRAME_COLORS is not None:
atom_colors[:] = FRAME_COLORS[0] # 初始帧颜色
else:
for i in range(N_ATOMS):
r, g, b = TAB10_RGB[i % len(TAB10_RGB)]
atom_colors[i] = [r, g, b, 1.0]
@@ -295,12 +512,12 @@ for f_idx, (pos, direction) in enumerate(faces):
# 右上角:相机信息
camera_info = scene.visuals.Text(
text="", color="white", font_size=14,
text="", color="white", font_size=12,
pos=(0, 0), anchor_x="right", anchor_y="top", parent=canvas.scene)
# 左上角:小球信息
ball_info = scene.visuals.Text(
text="", color=(0.2, 1.0, 0.2, 1.0), font_size=18,
text="", color=(0.2, 1.0, 0.2, 1.0), font_size=14,
pos=(0, 0), anchor_x="left", anchor_y="top",
face="黑体", bold=True, parent=canvas.scene)
@@ -312,7 +529,7 @@ reset_button = scene.visuals.Rectangle(
radius=6, color=(0.18, 0.35, 0.65, 0.85),
border_color="white", parent=canvas.scene)
reset_button_label = scene.visuals.Text(
text="reset", color="white", font_size=16,
text="reset", color="white", font_size=13,
pos=(reset_btn_size[0] / 2 + 8, reset_btn_size[1] / 2 + 8),
anchor_x="center", anchor_y="center",
bold=True, parent=canvas.scene)
@@ -326,7 +543,7 @@ info_button = scene.visuals.Rectangle(
radius=6, color=(0.9, 0.3, 0.3, 0.9),
border_color="white", parent=canvas.scene)
info_button_label = scene.visuals.Text(
text="info", color="white", font_size=16,
text="info", color="white", font_size=13,
pos=(info_btn_size[0] / 2 + 8, info_btn_size[1] / 2 + 8),
anchor_x="center", anchor_y="center",
bold=True, parent=canvas.scene)
@@ -346,7 +563,7 @@ axes_button = scene.visuals.Rectangle(
radius=6, color=(0.3, 0.7, 0.3, 0.9),
border_color="white", parent=canvas.scene)
axes_button_label = scene.visuals.Text(
text="axes", color="white", font_size=16,
text="axes", color="white", font_size=13,
pos=(axes_btn_size[0] / 2 + 8, axes_btn_size[1] / 2 + 8),
anchor_x="center", anchor_y="center",
bold=True, parent=canvas.scene)
@@ -551,11 +768,14 @@ def handle_mouse_press(event):
# ===========================================================================
def _update_atom_positions(f_idx):
"""更新所有原子到第 f_idx 帧的位置。"""
"""更新所有原子到第 f_idx 帧的位置,必要时更新颜色"""
if USE_MARKER:
marker_pos[:, 0] = DISP_ALL_X[f_idx]
marker_pos[:, 1] = DISP_ALL_Y[f_idx]
marker_pos[:, 2] = DISP_ALL_Z[f_idx]
if FRAME_COLORS is not None:
balls.set_data(pos=marker_pos, face_color=FRAME_COLORS[f_idx])
else:
balls.set_data(pos=marker_pos)
else:
for i in range(N_ATOMS):
@@ -630,10 +850,10 @@ def _load_move_camera_txt():
# 先试 move_camera.txt 直读,没有则用 display.txt 缓存
# header 中 camera_keyframes 为空字符串表示 move_camera=0(开关关闭),跳过文件加载
_camera_motion_enabled = bool(h.get("camera_keyframes", ""))
_camera_motion_enabled = bool(config.get("camera_keyframes", ""))
_CAM_MOTION = _load_move_camera_txt() if _camera_motion_enabled else None
if not _CAM_MOTION:
_CAM_MOTION = json.loads(h.get("camera_keyframes", "null")) if h.get("camera_keyframes") else None
_CAM_MOTION = json.loads(config.get("camera_keyframes", "null")) if config.get("camera_keyframes") else None
if _CAM_MOTION:
_cam_center = [0.0, 0.0, 0.0]
_cam_elev = initial_camera["elevation"]
+9 -20
View File
@@ -187,12 +187,11 @@ def run_case(config_path, runtime_base, input_dir="input", output_dir="output",
disp_path = os.path.join(output_dir_abs, "display.txt")
# ── 自动缓存检测 ───────────────────────────────────────
# force_calc=1: 强制重新计算,忽略缓存
# force_calc=0: 尊重 step_simulate 设置,不自动覆盖
# force_calc=1: 强制重新计算,忽略缓存(仅在 step_simulate=1 时生效)
# force_calc=0: 尊重 step_simulate 设置
force_calc = int(config.get("force_calc", 0))
if force_calc:
if force_calc and config.get("step_simulate", 1):
print(f"[run] force_calc=1,跳过缓存,强制重新计算")
config["step_simulate"] = 1
config["step_sample"] = 1
elif config.get("step_simulate", 1):
# step_simulate=1 且 force_calc=0 → 按用户要求执行计算
@@ -225,12 +224,12 @@ def run_case(config_path, runtime_base, input_dir="input", output_dir="output",
print(f"[run] 没有可用的缓存输出,但 step_simulate=0,将跳过模拟")
# 2. 运行物理模拟 → output/trajectory.txt
if config.get("step_simulate", 1):
_engine_aliases = {"c++": "cpp", "f90": "fortran", "f": "fortran"}
engine = _engine_aliases.get(
str(config.get("engine", "python")).lower(),
str(config.get("engine", "python")).lower()
)
if config.get("step_simulate", 1):
total_steps = config["NT"]
record_steps = total_steps - (config.get("warmup_steps") or 0)
print(f"[run] 开始计算 总步数={total_steps} 记录步数={record_steps} DT={config['DT']}")
@@ -248,23 +247,13 @@ def run_case(config_path, runtime_base, input_dir="input", output_dir="output",
input_dir_abs = str(input_dir_path.resolve())
output_dir_abs = str(output_dir_path.resolve())
# ── 优先尝试 DLL 路径(无文件 I/O,直接输出 display.npz)──
_dll_used = False
try:
# ── DLL 路径(无文件 I/O,直接输出 display.npz)──
from engines.engine_dll import is_dll_available
if is_dll_available(engine):
if not is_dll_available(engine):
raise FileNotFoundError(
f"DLL 未找到(引擎 {engine})。"
f"请先编译:cd engines/{engine} && make dll")
compute.run_engine_dll(engine, output_dir_abs, config)
_dll_used = True
print(f"[run] DLL 路径成功")
except Exception as _dll_err:
print(f"[run] DLL 路径不可用 ({_dll_err}),回退到子进程模式")
if not _dll_used:
# 回退:子进程模式(读写 display.txt / trajectory.txt
traj_x, traj_y, traj_z, traj_vx, traj_vy, traj_vz = compute.run_engine(
engine, input_dir_abs, output_dir_abs, config)
if int(config.get("save_trajectory", 0)):
compute.save_trajectory_txt(traj_x, traj_y, traj_z, traj_vx, traj_vy, traj_vz, str(runtime_base))
_elapsed = _time.time() - _t0
print(f"[run] 引擎: {engine} 计算完成: {record_steps}{_elapsed:.3f} s")
-82
View File
@@ -1,82 +0,0 @@
# engines/c/Makefile
# 跨平台编译:make → 本地系统编译
# make linux → Linux 交叉编译(需 x86_64-linux-gnu-gcc
# make windows → Windows 交叉编译(需 x86_64-w64-mingw32-gcc
# make macos → macOS 交叉编译(需 osxcross 工具链)
CC = gcc
CFLAGS = -O3 -march=native -Wall -Wextra
LDFLAGS = -lm
SRCS = main.c
LIB_SRC = dynamics_lib.c
# 自动检测系统
UNAME_S := $(shell uname -s 2>/dev/null || echo Windows)
# 目标文件名:统一使用 .exe 后缀(方便 Python 跨平台调用)
TARGET = build/dynamics_c.exe
# DLL 目标(平台自动选择后缀)
ifeq ($(UNAME_S),Linux)
DLL_TARGET = build/dynamics_c.so
DLL_FLAGS = -shared -fPIC
else ifeq ($(UNAME_S),Darwin)
DLL_TARGET = build/dynamics_c.dylib
DLL_FLAGS = -dynamiclib
else
DLL_TARGET = build/dynamics_c.dll
DLL_FLAGS = -shared
endif
# ── 本地编译 ─────────────────────────────────
.PHONY: all dll clean linux windows macos
all: $(TARGET)
dll: $(DLL_TARGET)
$(TARGET): $(SRCS) | build
$(CC) $(CFLAGS) -o $@ $(SRCS) $(LDFLAGS)
@echo " === C engine built: $@ ==="
$(DLL_TARGET): $(LIB_SRC) | build
$(CC) $(CFLAGS) $(DLL_FLAGS) -o $@ $(LIB_SRC) $(LDFLAGS)
@echo " === C DLL built: $@ ==="
build:
mkdir -p build
# ── 交叉编译 ─────────────────────────────────
# Linux → Linux (x86_64)
linux: CROSS_PREFIX = x86_64-linux-gnu-
linux: CC = $(CROSS_PREFIX)gcc
linux: CFLAGS = -O3 -march=x86-64 -Wall -Wextra
linux: $(SRCS) | build
$(CC) $(CFLAGS) -o build/dynamics_c_linux.exe $(SRCS) $(LDFLAGS)
@echo " === Linux binary: build/dynamics_c_linux.exe ==="
# 任意平台 → Windows (x86_64)
# 需要安装 MinGW 交叉编译器:
# apt install mingw-w64 (Debian/Ubuntu)
# brew install mingw-w64 (macOS)
windows: CROSS_PREFIX = x86_64-w64-mingw32-
windows: CC = $(CROSS_PREFIX)gcc
windows: CFLAGS = -O3 -march=x86-64 -Wall -Wextra
windows: $(SRCS) | build
$(CC) $(CFLAGS) -o build/dynamics_c_win.exe $(SRCS) $(LDFLAGS)
@echo " === Windows binary: build/dynamics_c_win.exe ==="
# 任意平台 → macOS (x86_64)
# 需要安装 osxcross 工具链
macos: CROSS_PREFIX = x86_64-apple-darwin-
macos: CC = $(CROSS_PREFIX)gcc
macos: CFLAGS = -O3 -march=x86-64 -Wall -Wextra
macos: $(SRCS) | build
$(CC) $(CFLAGS) -o build/dynamics_c_mac.exe $(SRCS) $(LDFLAGS)
@echo " === macOS binary: build/dynamics_c_mac.exe ==="
# ── 编译所有平台 ──────────────────────────────
all-platforms: linux windows macos
clean:
rm -rf build *.o
-1
View File
@@ -1 +0,0 @@
{"n_atoms": 40, "nt": 200000, "step_time": 2.5352442264556887e-05}
-554
View File
@@ -1,554 +0,0 @@
/**
* engines/c/dynamics_lib.c
* -------------------------
* 纯计算 DLL:无文件 I/O,所有数据由 Python 以 NumPy 数组传入,
* 结果直接写入 Python 预分配的输出数组。
* 算法与 main.c 和 compute.py 保持完全一致。
*
* 编译(Windows DLL:
* gcc -O3 -march=native -shared -o build/dynamics_c.dll dynamics_lib.c -lm
* 编译(Linux .so:
* gcc -O3 -march=native -shared -fPIC -o build/dynamics_c.so dynamics_lib.c -lm
* 编译(macOS .dylib:
* gcc -O3 -march=native -dynamiclib -o build/dynamics_c.dylib dynamics_lib.c -lm
*/
#ifdef _WIN32
# define EXPORT __declspec(dllexport)
#else
# define EXPORT __attribute__((visibility("default")))
#endif
#include <math.h>
#include <stdlib.h>
#include <string.h>
#include <stdio.h>
/* ── 驱动力结构体 ─────────────────────────────────────────── */
typedef struct {
int n_drivers;
const int *idx; /* [n_drivers] 0-based local atom index */
const double *amp; /* [n_drivers*3] (ax,ay,az) interleaved */
const double *freq; /* [n_drivers*3] */
const double *phi; /* [n_drivers*3] radians */
const double *eq; /* [n_drivers*3] equilibrium positions */
const double *ncycles; /* [n_drivers] 0=unlimited */
const int *has_period; /* [n_drivers] */
/* mutable freeze positions (allocated internally) */
double *freeze; /* [n_drivers*3] */
} Drivers;
/* ── 加速度:保守力(弹簧键 + 均匀重力场)────────────────── */
static void accel_conservative(
int n, const double *x, const double *y, const double *z,
const double *m,
double Gx, double Gy, double Gz,
int gravity_field, int elastic_force,
int n_bonds, const int *bond_pairs,
const double *bond_k, const double *bond_r0,
double *ax, double *ay, double *az)
{
for (int i = 0; i < n; i++) {
ax[i] = gravity_field ? Gx : 0.0;
ay[i] = gravity_field ? Gy : 0.0;
az[i] = gravity_field ? Gz : 0.0;
}
if (!elastic_force || n_bonds == 0) return;
for (int b = 0; b < n_bonds; b++) {
int ii = bond_pairs[b*2];
int jj = bond_pairs[b*2+1];
double dx = x[jj] - x[ii];
double dy = y[jj] - y[ii];
double dz = z[jj] - z[ii];
double dist = sqrt(dx*dx + dy*dy + dz*dz);
if (dist < 1e-12) continue;
double k = bond_k[b];
double r0 = bond_r0[b];
double fac = k * (dist - r0) / dist;
double fx = fac * dx, fy = fac * dy, fz_b = fac * dz;
ax[ii] += fx / m[ii]; ay[ii] += fy / m[ii]; az[ii] += fz_b / m[ii];
ax[jj] -= fx / m[jj]; ay[jj] -= fy / m[jj]; az[jj] -= fz_b / m[jj];
}
}
/* ── 完整加速度(含阻尼)────────────────────────────────── */
static void accel_full(
int n, const double *x, const double *y, const double *z,
const double *vx, const double *vy, const double *vz,
const double *m,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bond_pairs,
const double *bond_k, const double *bond_r0,
double *ax, double *ay, double *az)
{
accel_conservative(n, x, y, z, m, Gx, Gy, Gz,
gravity_field, elastic_force,
n_bonds, bond_pairs, bond_k, bond_r0,
ax, ay, az);
if (damping_force) {
for (int i = 0; i < n; i++) {
ax[i] -= Bx * vx[i] / m[i];
ay[i] -= By * vy[i] / m[i];
az[i] -= Bz * vz[i] / m[i];
}
}
}
/* ── 边界:反弹(与 main.c limit_in_box 一致)────────────── */
static inline void _limit1(double *p, double *v, double lo, double hi) {
if (*p > hi) { *p = hi; *v = -fabs(*v); }
if (*p < lo) { *p = lo; *v = fabs(*v); }
}
/* ── 边界:回绕(与 main.c wrap_position 一致)──────────── */
static inline void _wrap1(double *p, double lo, double hi) {
if (*p > hi) *p = lo;
if (*p < lo) *p = hi;
}
/* ── 边界 + 固定约束(与 main.c apply_step 末尾一致)──────── */
static void apply_boundary_and_constraints(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const int *fixed, const double *pos_init,
double box_a)
{
double lo = -box_a, hi = box_a;
/* 反弹 */
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
_limit1(&x[i], &vx[i], lo, hi);
_limit1(&y[i], &vy[i], lo, hi);
_limit1(&z[i], &vz[i], lo, hi);
}
/* 回绕 */
for (int i = 0; i < n; i++) {
_wrap1(&x[i], lo, hi);
_wrap1(&y[i], lo, hi);
_wrap1(&z[i], lo, hi);
}
/* 逐自由度固定约束:与 main.c 和 Python apply_fixed_constraints 一致 */
for (int i = 0; i < n; i++) {
if (fixed[i*3+0]) { x[i] = pos_init[i*3+0]; vx[i] = 0.0; }
if (fixed[i*3+1]) { y[i] = pos_init[i*3+1]; vy[i] = 0.0; }
if (fixed[i*3+2]) { z[i] = pos_init[i*3+2]; vz[i] = 0.0; }
}
}
/* ══════════════════════════════════════════════════════════
* 蛙跳法(与 main.c leapfrog_step 完全一致)
* x(t), v(t-dt/2) → x(t+dt), v(t+dt/2)
* 无阻尼:纯辛积分。有阻尼:半隐式处理 α = B·dt/(2m)
* ══════════════════════════════════════════════════════════ */
static void leapfrog_step(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const double *m, const int *fixed,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bp, const double *bk, const double *br0,
double dt)
{
double *ax = (double*)alloca(n*sizeof(double)*3);
double *ay = ax+n; double *az = ay+n;
accel_conservative(n, x, y, z, m, Gx, Gy, Gz,
gravity_field, elastic_force,
n_bonds, bp, bk, br0, ax, ay, az);
int has_damp = damping_force && (Bx != 0.0 || By != 0.0 || Bz != 0.0);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
if (has_damp) {
double ax_ = Bx*dt/(2.0*m[i]);
double ay_ = By*dt/(2.0*m[i]);
double az_ = Bz*dt/(2.0*m[i]);
vx[i] = (vx[i]*(1.0-ax_) + ax[i]*dt) / (1.0+ax_);
vy[i] = (vy[i]*(1.0-ay_) + ay[i]*dt) / (1.0+ay_);
vz[i] = (vz[i]*(1.0-az_) + az[i]*dt) / (1.0+az_);
} else {
vx[i] += ax[i]*dt;
vy[i] += ay[i]*dt;
vz[i] += az[i]*dt;
}
x[i] += vx[i]*dt;
y[i] += vy[i]*dt;
z[i] += vz[i]*dt;
}
}
/* ══════════════════════════════════════════════════════════
* 显式欧拉法(与 main.c explicit_euler_step 一致)
* ══════════════════════════════════════════════════════════ */
static void euler_step(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const double *m, const int *fixed,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bp, const double *bk, const double *br0,
double dt)
{
double *ax = (double*)alloca(n*sizeof(double)*3);
double *ay = ax+n; double *az = ay+n;
accel_full(n, x, y, z, vx, vy, vz, m, Gx, Gy, Gz, Bx, By, Bz,
gravity_field, elastic_force, damping_force,
n_bonds, bp, bk, br0, ax, ay, az);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
x[i] += vx[i]*dt; y[i] += vy[i]*dt; z[i] += vz[i]*dt;
vx[i]+= ax[i]*dt; vy[i]+= ay[i]*dt; vz[i]+= az[i]*dt;
}
}
/* ══════════════════════════════════════════════════════════
* 隐式欧拉法(与 main.c implicit_euler_step 完全一致)
*
* main.c 逻辑:
* 1. 用 v_next ≈ (v + G·dt)/(1 + γ·dt) 预测(只含重力+阻尼,不含弹簧)
* 2. 用 (x, v_next) 计算完整加速度 a_next
* 3. v += a_next·dt; x += v·dt
* ══════════════════════════════════════════════════════════ */
static void implicit_euler_step(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const double *m, const int *fixed,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bp, const double *bk, const double *br0,
double dt)
{
double *vxn = (double*)alloca(n*sizeof(double)*3);
double *vyn = vxn+n; double *vzn = vyn+n;
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) {
vxn[i] = vyn[i] = vzn[i] = 0.0; continue;
}
double gx = Bx / m[i], gy = By / m[i], gz = Bz / m[i];
vxn[i] = (vx[i] + Gx*dt) / (1.0 + gx*dt);
vyn[i] = (vy[i] + Gy*dt) / (1.0 + gy*dt);
vzn[i] = (vz[i] + Gz*dt) / (1.0 + gz*dt);
}
double *ax = (double*)alloca(n*sizeof(double)*3);
double *ay = ax+n; double *az = ay+n;
accel_full(n, x, y, z, vxn, vyn, vzn, m, Gx, Gy, Gz, Bx, By, Bz,
gravity_field, elastic_force, damping_force,
n_bonds, bp, bk, br0, ax, ay, az);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
vx[i] += ax[i]*dt;
vy[i] += ay[i]*dt;
vz[i] += az[i]*dt;
x[i] += vx[i]*dt;
y[i] += vy[i]*dt;
z[i] += vz[i]*dt;
}
}
/* ══════════════════════════════════════════════════════════
* 中点法(与 main.c midpoint_step 完全一致)
*
* main.c 逻辑:
* 1. a = accel(x, v)
* 2. xm = x + 0.5·v·dt; vm = v + 0.5·a·dt
* 3. x = x + vm·dt (位置更新用 vm,即中点速度)
* 4. am = accel(xm, vm)
* 5. v = v + am·dt
* ══════════════════════════════════════════════════════════ */
static void midpoint_step(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const double *m, const int *fixed,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bp, const double *bk, const double *br0,
double dt)
{
/* Allocate in one block for cache locality */
double *buf = (double*)alloca(n*sizeof(double)*9);
double *ax = buf;
double *ay = ax+n; double *az = ay+n;
double *xm = az+n; double *ym = xm+n; double *zm = ym+n;
double *vxm = zm+n; double *vym = vxm+n; double *vzm = vym+n;
accel_full(n, x, y, z, vx, vy, vz, m, Gx, Gy, Gz, Bx, By, Bz,
gravity_field, elastic_force, damping_force,
n_bonds, bp, bk, br0, ax, ay, az);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) {
xm[i]=x[i]; ym[i]=y[i]; zm[i]=z[i];
vxm[i]=vym[i]=vzm[i]=0.0; continue;
}
xm[i] = x[i] + 0.5*vx[i]*dt;
ym[i] = y[i] + 0.5*vy[i]*dt;
zm[i] = z[i] + 0.5*vz[i]*dt;
vxm[i] = vx[i] + 0.5*ax[i]*dt;
vym[i] = vy[i] + 0.5*ay[i]*dt;
vzm[i] = vz[i] + 0.5*az[i]*dt;
/* position updated with midpoint velocity (same as main.c) */
x[i] = x[i] + vxm[i]*dt;
y[i] = y[i] + vym[i]*dt;
z[i] = z[i] + vzm[i]*dt;
}
double *axm = (double*)alloca(n*sizeof(double)*3);
double *aym = axm+n; double *azm = aym+n;
accel_full(n, xm, ym, zm, vxm, vym, vzm, m, Gx, Gy, Gz, Bx, By, Bz,
gravity_field, elastic_force, damping_force,
n_bonds, bp, bk, br0, axm, aym, azm);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
vx[i] += axm[i]*dt;
vy[i] += aym[i]*dt;
vz[i] += azm[i]*dt;
}
}
/* ── 驱动力(与 main.c apply_driving_force 一致)────────── */
static void apply_driving(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
double t, int step, double dt, Drivers *drv)
{
(void)n;
if (!drv || drv->n_drivers == 0) return;
const double TWO_PI = 2.0 * 3.14159265358979323846;
for (int d = 0; d < drv->n_drivers; d++) {
int idx = drv->idx[d];
double fx = drv->freq[d*3+0];
double fy = drv->freq[d*3+1];
double fz = drv->freq[d*3+2];
if (drv->has_period[d]) {
double mf = fabs(fx) > fabs(fy) ? fabs(fx) : fabs(fy);
if (fabs(fz) > mf) mf = fabs(fz);
int period_steps = 0;
if (mf > 1e-12)
period_steps = (int)(drv->ncycles[d] / mf / dt);
if (step > period_steps) {
x[idx] = drv->freeze[d*3+0];
y[idx] = drv->freeze[d*3+1];
z[idx] = drv->freeze[d*3+2];
vx[idx] = vy[idx] = vz[idx] = 0.0;
continue;
}
double px = drv->eq[d*3+0] + drv->amp[d*3+0]*cos(TWO_PI*fx*t + drv->phi[d*3+0]);
double py = drv->eq[d*3+1] + drv->amp[d*3+1]*cos(TWO_PI*fy*t + drv->phi[d*3+1]);
double pz = drv->eq[d*3+2] + drv->amp[d*3+2]*cos(TWO_PI*fz*t + drv->phi[d*3+2]);
if (step == period_steps) {
drv->freeze[d*3+0] = px;
drv->freeze[d*3+1] = py;
drv->freeze[d*3+2] = pz;
}
}
x[idx] = drv->eq[d*3+0] + drv->amp[d*3+0]*cos(TWO_PI*fx*t + drv->phi[d*3+0]);
y[idx] = drv->eq[d*3+1] + drv->amp[d*3+1]*cos(TWO_PI*fy*t + drv->phi[d*3+1]);
z[idx] = drv->eq[d*3+2] + drv->amp[d*3+2]*cos(TWO_PI*fz*t + drv->phi[d*3+2]);
vx[idx] = -drv->amp[d*3+0]*TWO_PI*fx*sin(TWO_PI*fx*t + drv->phi[d*3+0]);
vy[idx] = -drv->amp[d*3+1]*TWO_PI*fy*sin(TWO_PI*fy*t + drv->phi[d*3+1]);
vz[idx] = -drv->amp[d*3+2]*TWO_PI*fz*sin(TWO_PI*fz*t + drv->phi[d*3+2]);
}
}
/* ══════════════════════════════════════════════════════════
* 导出函数:run_dynamics
*
* 与 main.c 的计算顺序完全一致:
* 1. leapfrog 初始化 v(-dt/2)
* 2. 初始驱动 t=0
* 3. 预热循环(不记录)
* 4. 记录循环:drive → record → step → boundary → constraints
*
* 参数说明(所有数组均为 C-contiguous 行优先 float64/int32):
* n_atoms 原子数
* pos_init 初始位置 [n_atoms*3] x0,y0,z0, x1,y1,z1, ...
* vel_init 初始速度 [n_atoms*3]
* masses 质量 [n_atoms]
* fixed 自由度约束 [n_atoms*3] int32, 1=固定
* n_bonds 键数
* bond_pairs 键对 [n_bonds*2] int32, 0-based local index
* bond_k 刚度 [n_bonds]
* bond_r0 平衡键长 [n_bonds]
* box_a 盒子半边长
* dt 时间步长
* NT 总步数(含预热)
* NSTEP 抽帧间隔
* warmup_steps 预热步数
* method_id 0=euler 1=implicit 2=midpoint 3=leapfrog
* Gx/Gy/Gz 均匀重力场加速度分量
* Bx/By/Bz 阻尼系数分量
* gravity_field / elastic_force / damping_force 力开关
* gravity_strength 原子间引力强度(暂未实现,留接口)
* n_drivers 驱动原子数
* drv_idx 驱动原子局部索引 [n_drivers] int32
* drv_amp 振幅 [n_drivers*3]
* drv_freq 频率 [n_drivers*3]
* drv_phi 初相(弧度)[n_drivers*3]
* drv_eq 平衡位置 [n_drivers*3]
* drv_ncycles 周期数 [n_drivers] 0=不限
* drv_has_period [n_drivers] int32
* n_frames 输出帧数(Python 预计算:(NT-warmup)/NSTEP 向上取整)
* out_x/y/z/vx/vy/vz 输出数组 [n_frames*n_atoms] 由 Python 预分配
* progress_cb 进度回调(可为 NULL
*
* 返回:0=成功,负数=错误
* ══════════════════════════════════════════════════════════ */
EXPORT int run_dynamics(
int n_atoms,
const double *pos_init,
const double *vel_init,
const double *masses,
const int *fixed,
int n_bonds,
const int *bond_pairs,
const double *bond_k,
const double *bond_r0,
double box_a, double dt,
int NT, int NSTEP, int warmup_steps, int method_id,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
double gravity_strength,
int n_drivers,
const int *drv_idx,
const double *drv_amp,
const double *drv_freq,
const double *drv_phi,
const double *drv_eq,
const double *drv_ncycles,
const int *drv_has_period,
int n_frames,
double *out_x, double *out_y, double *out_z,
double *out_vx, double *out_vy, double *out_vz,
void (*progress_cb)(int step, int total))
{
(void)gravity_strength; /* 原子间引力暂未实现 */
int n = n_atoms;
/* ── 工作数组 ── */
double *x = (double*)malloc(n*sizeof(double));
double *y = (double*)malloc(n*sizeof(double));
double *z = (double*)malloc(n*sizeof(double));
double *vx = (double*)malloc(n*sizeof(double));
double *vy = (double*)malloc(n*sizeof(double));
double *vz = (double*)malloc(n*sizeof(double));
if (!x||!y||!z||!vx||!vy||!vz) return -1;
for (int i = 0; i < n; i++) {
x[i]=pos_init[i*3+0]; y[i]=pos_init[i*3+1]; z[i]=pos_init[i*3+2];
vx[i]=vel_init[i*3+0]; vy[i]=vel_init[i*3+1]; vz[i]=vel_init[i*3+2];
}
/* ── 驱动结构 ── */
Drivers drv;
drv.n_drivers = n_drivers;
drv.idx = drv_idx;
drv.amp = drv_amp;
drv.freq = drv_freq;
drv.phi = drv_phi;
drv.eq = drv_eq;
drv.ncycles = drv_ncycles;
drv.has_period = drv_has_period;
drv.freeze = NULL;
if (n_drivers > 0) {
drv.freeze = (double*)calloc(n_drivers*3, sizeof(double));
if (!drv.freeze) { free(x);free(y);free(z);free(vx);free(vy);free(vz); return -2; }
}
/* ── 内联步进宏 ── */
#define DO_STEP() do { \
switch (method_id) { \
case 0: euler_step(n,x,y,z,vx,vy,vz,masses,fixed,Gx,Gy,Gz,Bx,By,Bz, \
gravity_field,elastic_force,damping_force, \
n_bonds,bond_pairs,bond_k,bond_r0,dt); break; \
case 1: implicit_euler_step(n,x,y,z,vx,vy,vz,masses,fixed,Gx,Gy,Gz,Bx,By,Bz, \
gravity_field,elastic_force,damping_force, \
n_bonds,bond_pairs,bond_k,bond_r0,dt); break; \
case 2: midpoint_step(n,x,y,z,vx,vy,vz,masses,fixed,Gx,Gy,Gz,Bx,By,Bz, \
gravity_field,elastic_force,damping_force, \
n_bonds,bond_pairs,bond_k,bond_r0,dt); break; \
default: leapfrog_step(n,x,y,z,vx,vy,vz,masses,fixed,Gx,Gy,Gz,Bx,By,Bz, \
gravity_field,elastic_force,damping_force, \
n_bonds,bond_pairs,bond_k,bond_r0,dt); break; \
} \
apply_boundary_and_constraints(n,x,y,z,vx,vy,vz,fixed,pos_init,box_a); \
} while(0)
/* ── 蛙跳法:初始化 v(-dt/2) = v(0) - 0.5·a_c(0)·dt ── */
if (method_id == 3) {
double *ax0 = (double*)alloca(n*sizeof(double)*3);
double *ay0 = ax0+n; double *az0 = ay0+n;
accel_conservative(n, x, y, z, masses, Gx, Gy, Gz,
gravity_field, elastic_force,
n_bonds, bond_pairs, bond_k, bond_r0,
ax0, ay0, az0);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
vx[i] -= 0.5*ax0[i]*dt;
vy[i] -= 0.5*ay0[i]*dt;
vz[i] -= 0.5*az0[i]*dt;
}
}
/* ── 初始驱动 t=0(与 main.c 一致:leapfrog init 之后施加)── */
if (n_drivers > 0) apply_driving(n, x, y, z, vx, vy, vz, 0.0, 0, dt, &drv);
/* ── 预热(不记录)── */
for (int s = 0; s < warmup_steps; s++) {
double tw = (s + 1) * dt;
if (n_drivers > 0) apply_driving(n, x, y, z, vx, vy, vz, tw, s, dt, &drv);
DO_STEP();
}
/* ── 记录循环 ── */
int record_steps = NT - warmup_steps;
int prog_interval = record_steps / 100;
if (prog_interval < 1) prog_interval = 1;
int frame_idx = 0;
for (int s = 0; s < record_steps; s++) {
if (progress_cb && s % prog_interval == 0 && s > 0)
progress_cb(s, record_steps);
double t = (s + warmup_steps) * dt;
if (n_drivers > 0) apply_driving(n, x, y, z, vx, vy, vz, t, s, dt, &drv);
/* 抽帧记录(drive 之后,step 之前,与 main.c 一致)*/
if (s % NSTEP == 0 && frame_idx < n_frames) {
int base = frame_idx * n;
for (int i = 0; i < n; i++) {
out_x [base+i] = x[i]; out_y [base+i] = y[i]; out_z [base+i] = z[i];
out_vx[base+i] = vx[i]; out_vy[base+i] = vy[i]; out_vz[base+i] = vz[i];
}
frame_idx++;
}
DO_STEP();
}
#undef DO_STEP
free(x); free(y); free(z);
free(vx); free(vy); free(vz);
if (drv.freeze) free(drv.freeze);
return 0;
}
-1114
View File
File diff suppressed because it is too large Load Diff
-49
View File
@@ -1,49 +0,0 @@
# engines/cpp/Makefile
CXX = g++
SRCS = main.cpp
LIB_SRC = dynamics_lib.cpp
UNAME_S := $(shell uname -s 2>/dev/null || echo Windows)
CXXFLAGS = -O3 -march=native -std=c++17 -Wall -Wextra -D_USE_MATH_DEFINES
# Windows 下静态链接运行时,避免 libstdc++-6.dll / libgcc_s_seh-1.dll 版本冲突
ifeq ($(UNAME_S),Windows)
STATIC_FLAGS = -static-libgcc -static-libstdc++
else
STATIC_FLAGS =
endif
TARGET = build/dynamics_cpp.exe
ifeq ($(UNAME_S),Linux)
DLL_TARGET = build/dynamics_cpp.so
DLL_FLAGS = -shared -fPIC
else ifeq ($(UNAME_S),Darwin)
DLL_TARGET = build/dynamics_cpp.dylib
DLL_FLAGS = -dynamiclib
else
DLL_TARGET = build/dynamics_cpp.dll
DLL_FLAGS = -shared
endif
.PHONY: all dll clean
all: $(TARGET)
dll: $(DLL_TARGET)
$(TARGET): $(SRCS) | build
$(CXX) $(CXXFLAGS) $(STATIC_FLAGS) -o $@ $(SRCS)
@echo " === C++ engine built: $@ ==="
$(DLL_TARGET): $(LIB_SRC) | build
$(CXX) $(CXXFLAGS) $(STATIC_FLAGS) $(DLL_FLAGS) -o $@ $(LIB_SRC)
@echo " === C++ DLL built: $@ ==="
build:
mkdir -p build
clean:
rm -rf build *.o
-1
View File
@@ -1 +0,0 @@
{"n_atoms": 40, "nt": 200000, "step_time": 0.0022148028612136842}
-450
View File
@@ -1,450 +0,0 @@
/**
* engines/cpp/dynamics_lib.cpp
* -----------------------------
* 纯计算 DLL(C++ 版):无文件 I/O,所有数据由 Python 以 NumPy 数组传入。
* 算法与 main.cpp / compute.py 保持完全一致。
*
* 编译(Windows:
* g++ -O3 -march=native -std=c++17 -shared -o build/dynamics_cpp.dll dynamics_lib.cpp
* 编译(Linux:
* g++ -O3 -march=native -std=c++17 -shared -fPIC -o build/dynamics_cpp.so dynamics_lib.cpp
* 编译(macOS:
* g++ -O3 -march=native -std=c++17 -dynamiclib -o build/dynamics_cpp.dylib dynamics_lib.cpp
*/
#ifdef _WIN32
# define EXPORT extern "C" __declspec(dllexport)
#else
# define EXPORT extern "C" __attribute__((visibility("default")))
#endif
#include <cmath>
#include <cstring>
#include <cstdlib>
#include <vector>
/* ── 驱动力结构体 ─────────────────────────────────────────── */
struct Drivers {
int n_drivers = 0;
const int *idx = nullptr;
const double *amp = nullptr;
const double *freq = nullptr;
const double *phi = nullptr;
const double *eq = nullptr;
const double *ncycles = nullptr;
const int *has_period = nullptr;
std::vector<double> freeze; /* [n_drivers*3] 冻结位置(period 结束时锁定)*/
};
/* ── 加速度:保守力(弹簧键 + 均匀重力场)────────────────── */
static void accel_conservative(
int n, const double *x, const double *y, const double *z,
const double *m,
double Gx, double Gy, double Gz,
int gravity_field, int elastic_force,
int n_bonds, const int *bond_pairs,
const double *bond_k, const double *bond_r0,
double *ax, double *ay, double *az)
{
for (int i = 0; i < n; i++) {
ax[i] = gravity_field ? Gx : 0.0;
ay[i] = gravity_field ? Gy : 0.0;
az[i] = gravity_field ? Gz : 0.0;
}
if (!elastic_force || n_bonds == 0) return;
for (int b = 0; b < n_bonds; b++) {
int ii = bond_pairs[b*2];
int jj = bond_pairs[b*2+1];
double dx = x[jj]-x[ii], dy = y[jj]-y[ii], dz = z[jj]-z[ii];
double dist = std::sqrt(dx*dx + dy*dy + dz*dz);
if (dist < 1e-12) continue;
double fac = bond_k[b] * (dist - bond_r0[b]) / dist;
double fx = fac*dx, fy = fac*dy, fz_b = fac*dz;
ax[ii] += fx/m[ii]; ay[ii] += fy/m[ii]; az[ii] += fz_b/m[ii];
ax[jj] -= fx/m[jj]; ay[jj] -= fy/m[jj]; az[jj] -= fz_b/m[jj];
}
}
/* ── 完整加速度(含阻尼)────────────────────────────────── */
static void accel_full(
int n, const double *x, const double *y, const double *z,
const double *vx, const double *vy, const double *vz,
const double *m,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bond_pairs,
const double *bond_k, const double *bond_r0,
double *ax, double *ay, double *az)
{
accel_conservative(n, x, y, z, m, Gx, Gy, Gz,
gravity_field, elastic_force,
n_bonds, bond_pairs, bond_k, bond_r0,
ax, ay, az);
if (damping_force) {
for (int i = 0; i < n; i++) {
ax[i] -= Bx * vx[i] / m[i];
ay[i] -= By * vy[i] / m[i];
az[i] -= Bz * vz[i] / m[i];
}
}
}
/* ── 边界:反弹 ──────────────────────────────────────────── */
static inline void _limit1(double &p, double &v, double lo, double hi) {
if (p > hi) { p = hi; v = -std::fabs(v); }
if (p < lo) { p = lo; v = std::fabs(v); }
}
/* ── 边界:回绕 ──────────────────────────────────────────── */
static inline void _wrap1(double &p, double lo, double hi) {
if (p > hi) p = lo;
if (p < lo) p = hi;
}
/* ── 边界 + 固定约束 ────────────────────────────────────── */
static void apply_boundary_and_constraints(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const int *fixed, const double *pos_init, double box_a)
{
double lo = -box_a, hi = box_a;
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
_limit1(x[i], vx[i], lo, hi);
_limit1(y[i], vy[i], lo, hi);
_limit1(z[i], vz[i], lo, hi);
}
for (int i = 0; i < n; i++) {
_wrap1(x[i], lo, hi);
_wrap1(y[i], lo, hi);
_wrap1(z[i], lo, hi);
}
for (int i = 0; i < n; i++) {
if (fixed[i*3+0]) { x[i] = pos_init[i*3+0]; vx[i] = 0.0; }
if (fixed[i*3+1]) { y[i] = pos_init[i*3+1]; vy[i] = 0.0; }
if (fixed[i*3+2]) { z[i] = pos_init[i*3+2]; vz[i] = 0.0; }
}
}
/* ══════════════════════════════════════════════════════════
* 蛙跳法(与 main.cpp leapfrog_step 完全一致)
* ══════════════════════════════════════════════════════════ */
static void leapfrog_step(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const double *m, const int *fixed,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bp, const double *bk, const double *br0, double dt)
{
std::vector<double> buf(n * 3);
double *ax = buf.data(), *ay = ax+n, *az = ay+n;
accel_conservative(n, x, y, z, m, Gx, Gy, Gz,
gravity_field, elastic_force,
n_bonds, bp, bk, br0, ax, ay, az);
bool has_damp = damping_force && (Bx != 0.0 || By != 0.0 || Bz != 0.0);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
if (has_damp) {
double ax_ = Bx*dt/(2.0*m[i]);
double ay_ = By*dt/(2.0*m[i]);
double az_ = Bz*dt/(2.0*m[i]);
vx[i] = (vx[i]*(1.0-ax_) + ax[i]*dt) / (1.0+ax_);
vy[i] = (vy[i]*(1.0-ay_) + ay[i]*dt) / (1.0+ay_);
vz[i] = (vz[i]*(1.0-az_) + az[i]*dt) / (1.0+az_);
} else {
vx[i] += ax[i]*dt;
vy[i] += ay[i]*dt;
vz[i] += az[i]*dt;
}
x[i] += vx[i]*dt;
y[i] += vy[i]*dt;
z[i] += vz[i]*dt;
}
}
/* ══════════════════════════════════════════════════════════
* 显式欧拉法
* ══════════════════════════════════════════════════════════ */
static void euler_step(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const double *m, const int *fixed,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bp, const double *bk, const double *br0, double dt)
{
std::vector<double> buf(n * 3);
double *ax = buf.data(), *ay = ax+n, *az = ay+n;
accel_full(n, x, y, z, vx, vy, vz, m, Gx, Gy, Gz, Bx, By, Bz,
gravity_field, elastic_force, damping_force,
n_bonds, bp, bk, br0, ax, ay, az);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
x[i] += vx[i]*dt; y[i] += vy[i]*dt; z[i] += vz[i]*dt;
vx[i]+= ax[i]*dt; vy[i]+= ay[i]*dt; vz[i]+= az[i]*dt;
}
}
/* ══════════════════════════════════════════════════════════
* 隐式欧拉法(与 main.cpp implicit_euler_step 完全一致)
* ══════════════════════════════════════════════════════════ */
static void implicit_euler_step(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const double *m, const int *fixed,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bp, const double *bk, const double *br0, double dt)
{
std::vector<double> vbuf(n * 3), abuf(n * 3);
double *vxn = vbuf.data(), *vyn = vxn+n, *vzn = vyn+n;
double *ax = abuf.data(), *ay = ax+n, *az = ay+n;
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) {
vxn[i] = vyn[i] = vzn[i] = 0.0; continue;
}
double gx = Bx/m[i], gy = By/m[i], gz = Bz/m[i];
vxn[i] = (vx[i] + Gx*dt) / (1.0 + gx*dt);
vyn[i] = (vy[i] + Gy*dt) / (1.0 + gy*dt);
vzn[i] = (vz[i] + Gz*dt) / (1.0 + gz*dt);
}
accel_full(n, x, y, z, vxn, vyn, vzn, m, Gx, Gy, Gz, Bx, By, Bz,
gravity_field, elastic_force, damping_force,
n_bonds, bp, bk, br0, ax, ay, az);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
vx[i] += ax[i]*dt; vy[i] += ay[i]*dt; vz[i] += az[i]*dt;
x[i] += vx[i]*dt; y[i] += vy[i]*dt; z[i] += vz[i]*dt;
}
}
/* ══════════════════════════════════════════════════════════
* 中点法(与 main.cpp midpoint_step 完全一致)
* ══════════════════════════════════════════════════════════ */
static void midpoint_step(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
const double *m, const int *fixed,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
int n_bonds, const int *bp, const double *bk, const double *br0, double dt)
{
std::vector<double> buf(n * 9);
double *ax = buf.data();
double *ay = ax+n; double *az = ay+n;
double *xm = az+n; double *ym = xm+n; double *zm = ym+n;
double *vxm = zm+n; double *vym = vxm+n; double *vzm = vym+n;
accel_full(n, x, y, z, vx, vy, vz, m, Gx, Gy, Gz, Bx, By, Bz,
gravity_field, elastic_force, damping_force,
n_bonds, bp, bk, br0, ax, ay, az);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) {
xm[i]=x[i]; ym[i]=y[i]; zm[i]=z[i];
vxm[i]=vym[i]=vzm[i]=0.0; continue;
}
xm[i] = x[i] + 0.5*vx[i]*dt;
ym[i] = y[i] + 0.5*vy[i]*dt;
zm[i] = z[i] + 0.5*vz[i]*dt;
vxm[i] = vx[i] + 0.5*ax[i]*dt;
vym[i] = vy[i] + 0.5*ay[i]*dt;
vzm[i] = vz[i] + 0.5*az[i]*dt;
x[i] = x[i] + vxm[i]*dt;
y[i] = y[i] + vym[i]*dt;
z[i] = z[i] + vzm[i]*dt;
}
std::vector<double> abuf(n * 3);
double *axm = abuf.data(), *aym = axm+n, *azm = aym+n;
accel_full(n, xm, ym, zm, vxm, vym, vzm, m, Gx, Gy, Gz, Bx, By, Bz,
gravity_field, elastic_force, damping_force,
n_bonds, bp, bk, br0, axm, aym, azm);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
vx[i] += axm[i]*dt;
vy[i] += aym[i]*dt;
vz[i] += azm[i]*dt;
}
}
/* ── 驱动力 ─────────────────────────────────────────────── */
static void apply_driving(
int n, double *x, double *y, double *z,
double *vx, double *vy, double *vz,
double t, int step, double dt, Drivers &drv)
{
(void)n;
if (drv.n_drivers == 0) return;
constexpr double TWO_PI = 2.0 * 3.14159265358979323846;
for (int d = 0; d < drv.n_drivers; d++) {
int idx = drv.idx[d];
double fx = drv.freq[d*3+0];
double fy = drv.freq[d*3+1];
double fz = drv.freq[d*3+2];
if (drv.has_period[d]) {
double mf = std::fabs(fx) > std::fabs(fy) ? std::fabs(fx) : std::fabs(fy);
if (std::fabs(fz) > mf) mf = std::fabs(fz);
int period_steps = 0;
if (mf > 1e-12)
period_steps = (int)(drv.ncycles[d] / mf / dt);
if (step > period_steps) {
x[idx] = drv.freeze[d*3+0];
y[idx] = drv.freeze[d*3+1];
z[idx] = drv.freeze[d*3+2];
vx[idx] = vy[idx] = vz[idx] = 0.0;
continue;
}
double px = drv.eq[d*3+0] + drv.amp[d*3+0]*std::cos(TWO_PI*fx*t + drv.phi[d*3+0]);
double py = drv.eq[d*3+1] + drv.amp[d*3+1]*std::cos(TWO_PI*fy*t + drv.phi[d*3+1]);
double pz = drv.eq[d*3+2] + drv.amp[d*3+2]*std::cos(TWO_PI*fz*t + drv.phi[d*3+2]);
if (step == period_steps) {
drv.freeze[d*3+0] = px;
drv.freeze[d*3+1] = py;
drv.freeze[d*3+2] = pz;
}
}
x[idx] = drv.eq[d*3+0] + drv.amp[d*3+0]*std::cos(TWO_PI*fx*t + drv.phi[d*3+0]);
y[idx] = drv.eq[d*3+1] + drv.amp[d*3+1]*std::cos(TWO_PI*fy*t + drv.phi[d*3+1]);
z[idx] = drv.eq[d*3+2] + drv.amp[d*3+2]*std::cos(TWO_PI*fz*t + drv.phi[d*3+2]);
vx[idx] = -drv.amp[d*3+0]*TWO_PI*fx*std::sin(TWO_PI*fx*t + drv.phi[d*3+0]);
vy[idx] = -drv.amp[d*3+1]*TWO_PI*fy*std::sin(TWO_PI*fy*t + drv.phi[d*3+1]);
vz[idx] = -drv.amp[d*3+2]*TWO_PI*fz*std::sin(TWO_PI*fz*t + drv.phi[d*3+2]);
}
}
/* ══════════════════════════════════════════════════════════
* 导出函数:run_dynamics(接口与 C 版完全相同)
* ══════════════════════════════════════════════════════════ */
EXPORT int run_dynamics(
int n_atoms,
const double *pos_init,
const double *vel_init,
const double *masses,
const int *fixed,
int n_bonds,
const int *bond_pairs,
const double *bond_k,
const double *bond_r0,
double box_a, double dt,
int NT, int NSTEP, int warmup_steps, int method_id,
double Gx, double Gy, double Gz,
double Bx, double By, double Bz,
int gravity_field, int elastic_force, int damping_force,
double gravity_strength,
int n_drivers,
const int *drv_idx,
const double *drv_amp,
const double *drv_freq,
const double *drv_phi,
const double *drv_eq,
const double *drv_ncycles,
const int *drv_has_period,
int n_frames,
double *out_x, double *out_y, double *out_z,
double *out_vx, double *out_vy, double *out_vz,
void (*progress_cb)(int step, int total))
{
(void)gravity_strength;
int n = n_atoms;
std::vector<double> xv(n), yv(n), zv(n);
std::vector<double> vxv(n), vyv(n), vzv(n);
for (int i = 0; i < n; i++) {
xv[i]=pos_init[i*3+0]; yv[i]=pos_init[i*3+1]; zv[i]=pos_init[i*3+2];
vxv[i]=vel_init[i*3+0]; vyv[i]=vel_init[i*3+1]; vzv[i]=vel_init[i*3+2];
}
double *x=xv.data(), *y=yv.data(), *z=zv.data();
double *vx=vxv.data(), *vy=vyv.data(), *vz=vzv.data();
Drivers drv;
drv.n_drivers = n_drivers;
drv.idx = drv_idx;
drv.amp = drv_amp;
drv.freq = drv_freq;
drv.phi = drv_phi;
drv.eq = drv_eq;
drv.ncycles = drv_ncycles;
drv.has_period = drv_has_period;
if (n_drivers > 0)
drv.freeze.assign(n_drivers * 3, 0.0);
#define DO_STEP() do { \
switch (method_id) { \
case 0: euler_step(n,x,y,z,vx,vy,vz,masses,fixed,Gx,Gy,Gz,Bx,By,Bz, \
gravity_field,elastic_force,damping_force, \
n_bonds,bond_pairs,bond_k,bond_r0,dt); break; \
case 1: implicit_euler_step(n,x,y,z,vx,vy,vz,masses,fixed,Gx,Gy,Gz,Bx,By,Bz, \
gravity_field,elastic_force,damping_force, \
n_bonds,bond_pairs,bond_k,bond_r0,dt); break; \
case 2: midpoint_step(n,x,y,z,vx,vy,vz,masses,fixed,Gx,Gy,Gz,Bx,By,Bz, \
gravity_field,elastic_force,damping_force, \
n_bonds,bond_pairs,bond_k,bond_r0,dt); break; \
default: leapfrog_step(n,x,y,z,vx,vy,vz,masses,fixed,Gx,Gy,Gz,Bx,By,Bz, \
gravity_field,elastic_force,damping_force, \
n_bonds,bond_pairs,bond_k,bond_r0,dt); break; \
} \
apply_boundary_and_constraints(n,x,y,z,vx,vy,vz,fixed,pos_init,box_a); \
} while(0)
/* 蛙跳法:初始化 v(-dt/2) */
if (method_id == 3) {
std::vector<double> ibuf(n * 3);
double *ax0=ibuf.data(), *ay0=ax0+n, *az0=ay0+n;
accel_conservative(n, x, y, z, masses, Gx, Gy, Gz,
gravity_field, elastic_force,
n_bonds, bond_pairs, bond_k, bond_r0,
ax0, ay0, az0);
for (int i = 0; i < n; i++) {
if (fixed[i*3] && fixed[i*3+1] && fixed[i*3+2]) continue;
vx[i] -= 0.5*ax0[i]*dt;
vy[i] -= 0.5*ay0[i]*dt;
vz[i] -= 0.5*az0[i]*dt;
}
}
/* 初始驱动 t=0 */
if (n_drivers > 0) apply_driving(n, x, y, z, vx, vy, vz, 0.0, 0, dt, drv);
/* 预热 */
for (int s = 0; s < warmup_steps; s++) {
double tw = (s + 1) * dt;
if (n_drivers > 0) apply_driving(n, x, y, z, vx, vy, vz, tw, s, dt, drv);
DO_STEP();
}
/* 记录循环 */
int record_steps = NT - warmup_steps;
int prog_interval = std::max(1, record_steps / 100);
int frame_idx = 0;
for (int s = 0; s < record_steps; s++) {
if (progress_cb && s % prog_interval == 0 && s > 0)
progress_cb(s, record_steps);
double t = (s + warmup_steps) * dt;
if (n_drivers > 0) apply_driving(n, x, y, z, vx, vy, vz, t, s, dt, drv);
if (s % NSTEP == 0 && frame_idx < n_frames) {
int base = frame_idx * n;
for (int i = 0; i < n; i++) {
out_x [base+i] = x[i]; out_y [base+i] = y[i]; out_z [base+i] = z[i];
out_vx[base+i] = vx[i]; out_vy[base+i] = vy[i]; out_vz[base+i] = vz[i];
}
frame_idx++;
}
DO_STEP();
}
#undef DO_STEP
return 0;
}
-1025
View File
File diff suppressed because it is too large Load Diff
+5 -3
View File
@@ -12,9 +12,11 @@ Python ctypes 包装器:加载 C/C++/Fortran 动态链接库并调用 run_dyna
# arrays: dict with keys x, y, z, vx, vy, vz shape=(n_frames, n_atoms)
DLL 编译(C 版本):
Windows: gcc -O3 -shared -o engines/c/build/dynamics_c.dll engines/c/dynamics_lib.c -lm
Linux: gcc -O3 -shared -fPIC -o engines/c/build/dynamics_c.so engines/c/dynamics_lib.c -lm
macOS: gcc -O3 -dynamiclib -o engines/c/build/dynamics_c.dylib engines/c/dynamics_lib.c -lm
Windows: gcc -O3 -shared -o engines/release/dynamics_c.dll engines/src/c/dynamics_lib.c -lm
Linux: gcc -O3 -shared -fPIC -o engines/release/dynamics_c.so engines/src/c/dynamics_lib.c -lm
macOS: gcc -O3 -dynamiclib -o engines/release/dynamics_c.dylib engines/src/c/dynamics_lib.c -lm
或用 make dll 一键编译:
cd engines/src/c && make dll
"""
import ctypes
-47
View File
@@ -1,47 +0,0 @@
# engines/fortran/Makefile
FC = gfortran
FFLAGS = -O3 -march=native -Wall -Wextra
SRCS = main.f90
LIB_SRC = dynamics_lib.f90
UNAME_S := $(shell uname -s 2>/dev/null || echo Windows)
ifeq ($(UNAME_S),Windows)
STATIC_FLAGS = -static-libgcc -static-libgfortran -static-libquadmath
else
STATIC_FLAGS =
endif
TARGET = build/dynamics_f90.exe
ifeq ($(UNAME_S),Linux)
DLL_TARGET = build/dynamics_f90.so
DLL_FLAGS = -shared -fPIC
else ifeq ($(UNAME_S),Darwin)
DLL_TARGET = build/dynamics_f90.dylib
DLL_FLAGS = -dynamiclib
else
DLL_TARGET = build/dynamics_f90.dll
DLL_FLAGS = -shared -fPIC
endif
.PHONY: all dll clean
all: $(TARGET)
dll: $(DLL_TARGET)
$(TARGET): $(SRCS) | build
$(FC) $(FFLAGS) $(STATIC_FLAGS) -o $@ $(SRCS)
@echo " === Fortran engine built: $@ ==="
$(DLL_TARGET): $(LIB_SRC) | build
$(FC) $(FFLAGS) $(STATIC_FLAGS) $(DLL_FLAGS) -o $@ $(LIB_SRC)
@echo " === Fortran DLL built: $@ ==="
build:
mkdir -p build
clean:
rm -rf build *.o *.mod
-1
View File
@@ -1 +0,0 @@
{"n_atoms": 40, "nt": 200000, "step_time": 0.005991018545627594}
-483
View File
@@ -1,483 +0,0 @@
! engines/fortran/dynamics_lib.f90
! ---------------------------------
! 纯计算 DLLFortran 版):无文件 I/O,由 Python ctypes 调用。
! 算法与 main.f90 / compute.py 完全一致。
! 使用 iso_c_binding 导出 C 兼容接口。
!
! 编译(Windows:
! gfortran -O3 -march=native -shared -fPIC -o build/dynamics_f90.dll dynamics_lib.f90
! 编译(Linux:
! gfortran -O3 -march=native -shared -fPIC -o build/dynamics_f90.so dynamics_lib.f90
! 编译(macOS:
! gfortran -O3 -march=native -dynamiclib -o build/dynamics_f90.dylib dynamics_lib.f90
module dynamics_dll
use iso_c_binding, only: c_int, c_double, c_funptr, c_f_procpointer, c_associated
implicit none
private
real(c_double), parameter :: TWO_PI = 2.0d0 * 3.14159265358979323846d0
public :: run_dynamics
contains
! ── 保守加速度 ───────────────────────────────────────────────
subroutine accel_conservative(n, x, y, z, m, Gx, Gy, Gz, &
gravity_field, elastic_force, &
n_bonds, bond_pairs, bond_k, bond_r0, &
ax, ay, az)
integer, intent(in) :: n, gravity_field, elastic_force, n_bonds
real(c_double), intent(in) :: x(n), y(n), z(n), m(n)
real(c_double), intent(in) :: Gx, Gy, Gz
integer, intent(in) :: bond_pairs(2, n_bonds)
real(c_double), intent(in) :: bond_k(n_bonds), bond_r0(n_bonds)
real(c_double), intent(out) :: ax(n), ay(n), az(n)
integer :: b, ii, jj
real(c_double) :: dx, dy, dz, dist, fac, fx, fy, fz_b
if (gravity_field /= 0) then
ax = Gx; ay = Gy; az = Gz
else
ax = 0.0d0; ay = 0.0d0; az = 0.0d0
end if
if (elastic_force == 0 .or. n_bonds == 0) return
do b = 1, n_bonds
ii = bond_pairs(1, b) + 1 ! 0-based → 1-based
jj = bond_pairs(2, b) + 1
dx = x(jj)-x(ii); dy = y(jj)-y(ii); dz = z(jj)-z(ii)
dist = sqrt(dx*dx + dy*dy + dz*dz)
if (dist < 1.0d-12) cycle
fac = bond_k(b) * (dist - bond_r0(b)) / dist
fx = fac*dx; fy = fac*dy; fz_b = fac*dz
ax(ii) = ax(ii) + fx/m(ii); ay(ii) = ay(ii) + fy/m(ii); az(ii) = az(ii) + fz_b/m(ii)
ax(jj) = ax(jj) - fx/m(jj); ay(jj) = ay(jj) - fy/m(jj); az(jj) = az(jj) - fz_b/m(jj)
end do
end subroutine
! ── 完整加速度(含阻尼)──────────────────────────────────────
subroutine accel_full(n, x, y, z, vx, vy, vz, m, Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, ax, ay, az)
integer, intent(in) :: n, gravity_field, elastic_force, damping_force, n_bonds
real(c_double), intent(in) :: x(n), y(n), z(n), vx(n), vy(n), vz(n), m(n)
real(c_double), intent(in) :: Gx, Gy, Gz, Bx, By, Bz
integer, intent(in) :: bond_pairs(2, n_bonds)
real(c_double), intent(in) :: bond_k(n_bonds), bond_r0(n_bonds)
real(c_double), intent(out) :: ax(n), ay(n), az(n)
integer :: i
call accel_conservative(n, x, y, z, m, Gx, Gy, Gz, &
gravity_field, elastic_force, &
n_bonds, bond_pairs, bond_k, bond_r0, ax, ay, az)
if (damping_force /= 0) then
do i = 1, n
ax(i) = ax(i) - Bx*vx(i)/m(i)
ay(i) = ay(i) - By*vy(i)/m(i)
az(i) = az(i) - Bz*vz(i)/m(i)
end do
end if
end subroutine
! ── 边界 + 固定约束 ──────────────────────────────────────────
subroutine apply_bc(n, x, y, z, vx, vy, vz, fixed, pos_init, box_a)
integer, intent(in) :: n
real(c_double), intent(inout) :: x(n), y(n), z(n), vx(n), vy(n), vz(n)
integer, intent(in) :: fixed(3, n)
real(c_double), intent(in) :: pos_init(3, n), box_a
integer :: i
real(c_double) :: lo, hi
lo = -box_a; hi = box_a
! 反弹
do i = 1, n
if (fixed(1,i)/=0 .and. fixed(2,i)/=0 .and. fixed(3,i)/=0) cycle
if (x(i)>hi) then; x(i)=hi; vx(i)=-abs(vx(i)); end if
if (x(i)<lo) then; x(i)=lo; vx(i)= abs(vx(i)); end if
if (y(i)>hi) then; y(i)=hi; vy(i)=-abs(vy(i)); end if
if (y(i)<lo) then; y(i)=lo; vy(i)= abs(vy(i)); end if
if (z(i)>hi) then; z(i)=hi; vz(i)=-abs(vz(i)); end if
if (z(i)<lo) then; z(i)=lo; vz(i)= abs(vz(i)); end if
end do
! 回绕
do i = 1, n
if (x(i)>hi) x(i)=lo; if (x(i)<lo) x(i)=hi
if (y(i)>hi) y(i)=lo; if (y(i)<lo) y(i)=hi
if (z(i)>hi) z(i)=lo; if (z(i)<lo) z(i)=hi
end do
! 逐自由度固定约束
do i = 1, n
if (fixed(1,i)/=0) then; x(i)=pos_init(1,i); vx(i)=0.0d0; end if
if (fixed(2,i)/=0) then; y(i)=pos_init(2,i); vy(i)=0.0d0; end if
if (fixed(3,i)/=0) then; z(i)=pos_init(3,i); vz(i)=0.0d0; end if
end do
end subroutine
! ── 蛙跳法 ───────────────────────────────────────────────────
subroutine leapfrog_step(n, x, y, z, vx, vy, vz, m, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt)
integer, intent(in) :: n, gravity_field, elastic_force, damping_force, n_bonds
real(c_double), intent(inout) :: x(n), y(n), z(n), vx(n), vy(n), vz(n)
real(c_double), intent(in) :: m(n), bond_k(n_bonds), bond_r0(n_bonds)
integer, intent(in) :: fixed(3,n), bond_pairs(2,n_bonds)
real(c_double), intent(in) :: Gx, Gy, Gz, Bx, By, Bz, dt
real(c_double) :: ax(n), ay(n), az(n), ax_, ay_, az_
logical :: has_damp
integer :: i
call accel_conservative(n, x, y, z, m, Gx, Gy, Gz, &
gravity_field, elastic_force, &
n_bonds, bond_pairs, bond_k, bond_r0, ax, ay, az)
has_damp = (damping_force/=0) .and. (abs(Bx)+abs(By)+abs(Bz) > 0.0d0)
do i = 1, n
if (fixed(1,i)/=0 .and. fixed(2,i)/=0 .and. fixed(3,i)/=0) cycle
if (has_damp) then
ax_ = Bx*dt/(2.0d0*m(i)); ay_ = By*dt/(2.0d0*m(i)); az_ = Bz*dt/(2.0d0*m(i))
vx(i) = (vx(i)*(1.0d0-ax_) + ax(i)*dt)/(1.0d0+ax_)
vy(i) = (vy(i)*(1.0d0-ay_) + ay(i)*dt)/(1.0d0+ay_)
vz(i) = (vz(i)*(1.0d0-az_) + az(i)*dt)/(1.0d0+az_)
else
vx(i) = vx(i)+ax(i)*dt; vy(i) = vy(i)+ay(i)*dt; vz(i) = vz(i)+az(i)*dt
end if
x(i) = x(i)+vx(i)*dt; y(i) = y(i)+vy(i)*dt; z(i) = z(i)+vz(i)*dt
end do
end subroutine
! ── 显式欧拉法 ───────────────────────────────────────────────
subroutine euler_step(n, x, y, z, vx, vy, vz, m, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt)
integer, intent(in) :: n, gravity_field, elastic_force, damping_force, n_bonds
real(c_double), intent(inout) :: x(n), y(n), z(n), vx(n), vy(n), vz(n)
real(c_double), intent(in) :: m(n), bond_k(n_bonds), bond_r0(n_bonds)
integer, intent(in) :: fixed(3,n), bond_pairs(2,n_bonds)
real(c_double), intent(in) :: Gx, Gy, Gz, Bx, By, Bz, dt
real(c_double) :: ax(n), ay(n), az(n)
integer :: i
call accel_full(n, x, y, z, vx, vy, vz, m, Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, ax, ay, az)
do i = 1, n
if (fixed(1,i)/=0 .and. fixed(2,i)/=0 .and. fixed(3,i)/=0) cycle
x(i) = x(i)+vx(i)*dt; y(i) = y(i)+vy(i)*dt; z(i) = z(i)+vz(i)*dt
vx(i)= vx(i)+ax(i)*dt; vy(i)= vy(i)+ay(i)*dt; vz(i)= vz(i)+az(i)*dt
end do
end subroutine
! ── 隐式欧拉法 ───────────────────────────────────────────────
subroutine implicit_euler_step(n, x, y, z, vx, vy, vz, m, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt)
integer, intent(in) :: n, gravity_field, elastic_force, damping_force, n_bonds
real(c_double), intent(inout) :: x(n), y(n), z(n), vx(n), vy(n), vz(n)
real(c_double), intent(in) :: m(n), bond_k(n_bonds), bond_r0(n_bonds)
integer, intent(in) :: fixed(3,n), bond_pairs(2,n_bonds)
real(c_double), intent(in) :: Gx, Gy, Gz, Bx, By, Bz, dt
real(c_double) :: vxn(n), vyn(n), vzn(n), ax(n), ay(n), az(n)
real(c_double) :: gamma_x, gamma_y, gamma_z
integer :: i
do i = 1, n
if (fixed(1,i)/=0 .and. fixed(2,i)/=0 .and. fixed(3,i)/=0) then
vxn(i)=0.0d0; vyn(i)=0.0d0; vzn(i)=0.0d0; cycle
end if
gamma_x = Bx/m(i); gamma_y = By/m(i); gamma_z = Bz/m(i)
vxn(i) = (vx(i)+Gx*dt)/(1.0d0+gamma_x*dt)
vyn(i) = (vy(i)+Gy*dt)/(1.0d0+gamma_y*dt)
vzn(i) = (vz(i)+Gz*dt)/(1.0d0+gamma_z*dt)
end do
call accel_full(n, x, y, z, vxn, vyn, vzn, m, Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, ax, ay, az)
do i = 1, n
if (fixed(1,i)/=0 .and. fixed(2,i)/=0 .and. fixed(3,i)/=0) cycle
vx(i)=vx(i)+ax(i)*dt; vy(i)=vy(i)+ay(i)*dt; vz(i)=vz(i)+az(i)*dt
x(i) =x(i) +vx(i)*dt; y(i) =y(i) +vy(i)*dt; z(i) =z(i) +vz(i)*dt
end do
end subroutine
! ── 中点法 ───────────────────────────────────────────────────
subroutine midpoint_step(n, x, y, z, vx, vy, vz, m, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt)
integer, intent(in) :: n, gravity_field, elastic_force, damping_force, n_bonds
real(c_double), intent(inout) :: x(n), y(n), z(n), vx(n), vy(n), vz(n)
real(c_double), intent(in) :: m(n), bond_k(n_bonds), bond_r0(n_bonds)
integer, intent(in) :: fixed(3,n), bond_pairs(2,n_bonds)
real(c_double), intent(in) :: Gx, Gy, Gz, Bx, By, Bz, dt
real(c_double) :: ax(n), ay(n), az(n)
real(c_double) :: xm(n), ym(n), zm(n), vxm(n), vym(n), vzm(n)
real(c_double) :: axm(n), aym(n), azm(n)
integer :: i
call accel_full(n, x, y, z, vx, vy, vz, m, Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, ax, ay, az)
do i = 1, n
if (fixed(1,i)/=0 .and. fixed(2,i)/=0 .and. fixed(3,i)/=0) then
xm(i)=x(i); ym(i)=y(i); zm(i)=z(i)
vxm(i)=0.0d0; vym(i)=0.0d0; vzm(i)=0.0d0; cycle
end if
xm(i) = x(i) +0.5d0*vx(i)*dt; ym(i) = y(i) +0.5d0*vy(i)*dt; zm(i) = z(i) +0.5d0*vz(i)*dt
vxm(i) = vx(i)+0.5d0*ax(i)*dt; vym(i) = vy(i)+0.5d0*ay(i)*dt; vzm(i) = vz(i)+0.5d0*az(i)*dt
x(i) = x(i) +vxm(i)*dt; y(i) = y(i) +vym(i)*dt; z(i) = z(i) +vzm(i)*dt
end do
call accel_full(n, xm, ym, zm, vxm, vym, vzm, m, Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, axm, aym, azm)
do i = 1, n
if (fixed(1,i)/=0 .and. fixed(2,i)/=0 .and. fixed(3,i)/=0) cycle
vx(i)=vx(i)+axm(i)*dt; vy(i)=vy(i)+aym(i)*dt; vz(i)=vz(i)+azm(i)*dt
end do
end subroutine
! ── 驱动力施加 ───────────────────────────────────────────────
subroutine apply_drive(n, x, y, z, vx, vy, vz, t, step, dt, &
nd, drv_idx, drv_amp, drv_freq, drv_phi, &
drv_eq, drv_ncycles, drv_has_period, freeze)
integer, intent(in) :: n, nd, step
real(c_double), intent(inout) :: x(n), y(n), z(n), vx(n), vy(n), vz(n)
real(c_double), intent(in) :: t, dt
integer, intent(in) :: drv_idx(nd), drv_has_period(nd)
real(c_double), intent(in) :: drv_amp(3,nd), drv_freq(3,nd)
real(c_double), intent(in) :: drv_phi(3,nd), drv_eq(3,nd)
real(c_double), intent(in) :: drv_ncycles(nd)
real(c_double), intent(inout) :: freeze(3,nd)
integer :: d, idx, ps
real(c_double) :: fx, fy, fz, mf, px, py, pz
do d = 1, nd
idx = drv_idx(d) + 1 ! 0-based → 1-based
fx = drv_freq(1,d); fy = drv_freq(2,d); fz = drv_freq(3,d)
if (drv_has_period(d) /= 0) then
mf = max(abs(fx), max(abs(fy), abs(fz)))
ps = 0
if (mf > 1.0d-12) ps = int(drv_ncycles(d)/mf/dt)
if (step > ps) then
x(idx)=freeze(1,d); y(idx)=freeze(2,d); z(idx)=freeze(3,d)
vx(idx)=0.0d0; vy(idx)=0.0d0; vz(idx)=0.0d0
cycle
end if
px = drv_eq(1,d)+drv_amp(1,d)*cos(TWO_PI*fx*t+drv_phi(1,d))
py = drv_eq(2,d)+drv_amp(2,d)*cos(TWO_PI*fy*t+drv_phi(2,d))
pz = drv_eq(3,d)+drv_amp(3,d)*cos(TWO_PI*fz*t+drv_phi(3,d))
if (step == ps) then
freeze(1,d)=px; freeze(2,d)=py; freeze(3,d)=pz
end if
end if
x(idx) = drv_eq(1,d)+drv_amp(1,d)*cos(TWO_PI*fx*t+drv_phi(1,d))
y(idx) = drv_eq(2,d)+drv_amp(2,d)*cos(TWO_PI*fy*t+drv_phi(2,d))
z(idx) = drv_eq(3,d)+drv_amp(3,d)*cos(TWO_PI*fz*t+drv_phi(3,d))
vx(idx) = -drv_amp(1,d)*TWO_PI*fx*sin(TWO_PI*fx*t+drv_phi(1,d))
vy(idx) = -drv_amp(2,d)*TWO_PI*fy*sin(TWO_PI*fy*t+drv_phi(2,d))
vz(idx) = -drv_amp(3,d)*TWO_PI*fz*sin(TWO_PI*fz*t+drv_phi(3,d))
end do
end subroutine
! ══════════════════════════════════════════════════════════════
! 导出函数:run_dynamicsC 兼容接口,bind(C)
! 接口与 C/C++ DLL 完全相同(扁平 C-contiguous 数组)。
! ══════════════════════════════════════════════════════════════
integer(c_int) function run_dynamics( &
n_atoms, pos_init, vel_init, masses, fixed, &
n_bonds, bond_pairs, bond_k, bond_r0, &
box_a, dt, NT, NSTEP, warmup_steps, method_id, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, gravity_strength, &
n_drivers, drv_idx, drv_amp, drv_freq, drv_phi, drv_eq, &
drv_ncycles, drv_has_period, &
n_frames, out_x, out_y, out_z, out_vx, out_vy, out_vz, &
progress_cb) &
bind(C, name="run_dynamics")
integer(c_int), value, intent(in) :: n_atoms, n_bonds, NT, NSTEP
integer(c_int), value, intent(in) :: warmup_steps, method_id
integer(c_int), value, intent(in) :: gravity_field, elastic_force, damping_force
integer(c_int), value, intent(in) :: n_drivers, n_frames
real(c_double), value, intent(in) :: box_a, dt
real(c_double), value, intent(in) :: Gx, Gy, Gz, Bx, By, Bz
real(c_double), value, intent(in) :: gravity_strength
! 扁平数组:Python 传入 C-contiguous int32/float64
! Fortran 以列优先解释,维度反转:(3,n) 对应 C 的 n×3
real(c_double), intent(in) :: pos_init(3, n_atoms)
real(c_double), intent(in) :: vel_init(3, n_atoms)
real(c_double), intent(in) :: masses(n_atoms)
integer(c_int), intent(in) :: fixed(3, n_atoms)
integer(c_int), intent(in) :: bond_pairs(2, n_bonds)
real(c_double), intent(in) :: bond_k(n_bonds), bond_r0(n_bonds)
integer(c_int), intent(in) :: drv_idx(n_drivers)
real(c_double), intent(in) :: drv_amp(3, n_drivers)
real(c_double), intent(in) :: drv_freq(3, n_drivers)
real(c_double), intent(in) :: drv_phi(3, n_drivers)
real(c_double), intent(in) :: drv_eq(3, n_drivers)
real(c_double), intent(in) :: drv_ncycles(n_drivers)
integer(c_int), intent(in) :: drv_has_period(n_drivers)
real(c_double), intent(out) :: out_x(n_atoms, n_frames)
real(c_double), intent(out) :: out_y(n_atoms, n_frames)
real(c_double), intent(out) :: out_z(n_atoms, n_frames)
real(c_double), intent(out) :: out_vx(n_atoms, n_frames)
real(c_double), intent(out) :: out_vy(n_atoms, n_frames)
real(c_double), intent(out) :: out_vz(n_atoms, n_frames)
type(c_funptr), value, intent(in) :: progress_cb
! 进度回调接口
abstract interface
subroutine cb_iface(step, total) bind(C)
use iso_c_binding
integer(c_int), value :: step, total
end subroutine
end interface
procedure(cb_iface), pointer :: cb_ptr
integer :: n, s, frame_idx, record_steps, prog_interval, nd
real(c_double) :: t, tw
real(c_double), allocatable :: x(:), y(:), z(:), vx(:), vy(:), vz(:)
real(c_double), allocatable :: ax0(:), ay0(:), az0(:)
real(c_double), allocatable :: freeze(:,:)
logical :: has_cb
n = n_atoms
nd = n_drivers
allocate(x(n), y(n), z(n), vx(n), vy(n), vz(n))
do s = 1, n
x(s) = pos_init(1,s); y(s) = pos_init(2,s); z(s) = pos_init(3,s)
vx(s) = vel_init(1,s); vy(s) = vel_init(2,s); vz(s) = vel_init(3,s)
end do
allocate(freeze(3, max(nd,1)))
freeze = 0.0d0
has_cb = c_associated(progress_cb)
if (has_cb) call c_f_procpointer(progress_cb, cb_ptr)
! ── 蛙跳法:初始化 v(-dt/2) ─────────────────────────────
if (method_id == 3) then
allocate(ax0(n), ay0(n), az0(n))
call accel_conservative(n, x, y, z, masses, Gx, Gy, Gz, &
gravity_field, elastic_force, &
n_bonds, bond_pairs, bond_k, bond_r0, ax0, ay0, az0)
do s = 1, n
if (fixed(1,s)/=0 .and. fixed(2,s)/=0 .and. fixed(3,s)/=0) cycle
vx(s)=vx(s)-0.5d0*ax0(s)*dt
vy(s)=vy(s)-0.5d0*ay0(s)*dt
vz(s)=vz(s)-0.5d0*az0(s)*dt
end do
deallocate(ax0, ay0, az0)
end if
! ── 初始驱动 t=0 ─────────────────────────────────────────
if (nd > 0) call apply_drive(n, x, y, z, vx, vy, vz, 0.0d0, 0, dt, &
nd, drv_idx, drv_amp, drv_freq, drv_phi, &
drv_eq, drv_ncycles, drv_has_period, freeze)
! ── 预热 ─────────────────────────────────────────────────
do s = 0, warmup_steps-1
tw = (s+1)*dt
if (nd>0) call apply_drive(n, x, y, z, vx, vy, vz, tw, s, dt, &
nd, drv_idx, drv_amp, drv_freq, drv_phi, &
drv_eq, drv_ncycles, drv_has_period, freeze)
call do_step(n, x, y, z, vx, vy, vz, masses, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt, method_id, &
pos_init, box_a)
end do
! ── 记录循环 ─────────────────────────────────────────────
record_steps = NT - warmup_steps
prog_interval = max(1, record_steps/100)
frame_idx = 0
do s = 0, record_steps-1
if (has_cb .and. mod(s, prog_interval)==0 .and. s>0) call cb_ptr(s, record_steps)
t = (s+warmup_steps)*dt
if (nd>0) call apply_drive(n, x, y, z, vx, vy, vz, t, s, dt, &
nd, drv_idx, drv_amp, drv_freq, drv_phi, &
drv_eq, drv_ncycles, drv_has_period, freeze)
if (mod(s, NSTEP)==0 .and. frame_idx<n_frames) then
frame_idx = frame_idx+1
out_x(:, frame_idx) = x
out_y(:, frame_idx) = y
out_z(:, frame_idx) = z
out_vx(:, frame_idx) = vx
out_vy(:, frame_idx) = vy
out_vz(:, frame_idx) = vz
end if
call do_step(n, x, y, z, vx, vy, vz, masses, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt, method_id, &
pos_init, box_a)
end do
deallocate(x, y, z, vx, vy, vz, freeze)
run_dynamics = 0
contains
subroutine do_step(n, x, y, z, vx, vy, vz, m, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt, method_id, &
pos_init, box_a)
integer, intent(in) :: n, gravity_field, elastic_force, damping_force
integer, intent(in) :: n_bonds, method_id
real(c_double), intent(inout) :: x(n), y(n), z(n), vx(n), vy(n), vz(n)
real(c_double), intent(in) :: m(n), bond_k(n_bonds), bond_r0(n_bonds)
integer, intent(in) :: fixed(3,n), bond_pairs(2,n_bonds)
real(c_double), intent(in) :: Gx, Gy, Gz, Bx, By, Bz, dt, box_a
real(c_double), intent(in) :: pos_init(3, n)
select case (method_id)
case (0)
call euler_step(n, x, y, z, vx, vy, vz, m, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt)
case (1)
call implicit_euler_step(n, x, y, z, vx, vy, vz, m, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt)
case (2)
call midpoint_step(n, x, y, z, vx, vy, vz, m, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt)
case default
call leapfrog_step(n, x, y, z, vx, vy, vz, m, fixed, &
Gx, Gy, Gz, Bx, By, Bz, &
gravity_field, elastic_force, damping_force, &
n_bonds, bond_pairs, bond_k, bond_r0, dt)
end select
call apply_bc(n, x, y, z, vx, vy, vz, fixed, pos_init, box_a)
end subroutine do_step
end function run_dynamics
end module dynamics_dll
File diff suppressed because it is too large Load Diff
+4
View File
@@ -212,6 +212,10 @@ def main():
"number_of_frames": str(n_frames),
"number_of_particles": str(len(atom_ids)),
"use_marker": str(use_marker),
"display_color": json.dumps(p.get("display_color",
{"x":[0,[1.0,0.0,0.0]],"y":[0,[0.0,1.0,0.0]],"z":[0,[0.0,0.0,1.0]],
"xy":[0,[1.0,1.0,0.0]],"yz":[0,[0.0,1.0,1.0]],"zx":[0,[1.0,0.0,1.0]],
"xyz":[1,[1.0,1.0,1.0]]})),
"ball_radius": str(ball_radius),
"ball_color_r": str(ball_color[0]),
"ball_color_g": str(ball_color[1]),
-1
View File
@@ -1 +0,0 @@
{"n_atoms": 40, "nt": 10000, "step_time": 0.00015027170181274412}
-51
View File
@@ -1,51 +0,0 @@
{
"box_a": 300.0,
"NT": 10000,
"DT": 0.001,
"NSTEP": 20,
"warmup_steps": 0,
"method": "leapfrog",
"G": [
0.0,
0.0,
0.0
],
"B": [
0.005,
0.0,
0.005
],
"gravity_field": 0,
"gravity_interaction": 0,
"elastic_force": 1,
"damping_force": 0,
"gravity_strength": 1.0,
"driving_force": 1,
"save_trajectory": 0,
"alpha": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"ball_radius": 0.5,
"ball_color": [
0.2,
0.6,
0.9
],
"box_color": [
0.8,
0.8,
0.85
],
"use_marker": 1,
"camera_distance": 120.0,
"camera_elevation": 0.0,
"camera_azimuth": 0.0,
"camera_center_x": 60.0,
"camera_center_y": 0.0,
"camera_center_z": 0.0
}
Binary file not shown.
Binary file not shown.
Binary file not shown.
+31 -62
View File
@@ -1,82 +1,51 @@
# engines/c/Makefile
# 跨平台编译:make → 本地系统编译
# make linux → Linux 交叉编译(需 x86_64-linux-gnu-gcc
# make windows → Windows 交叉编译(需 x86_64-w64-mingw32-gcc
# make macos → macOS 交叉编译(需 osxcross 工具链)
# engines/src/c/Makefile
# 编译 DLL 到 engines/release/(主程序通过 ctypes 直接调用)
CC = gcc
CFLAGS = -O3 -march=native -Wall -Wextra
LDFLAGS = -lm
SRCS = main.c
LIB_SRC = dynamics_lib.c
# 自动检测系统
UNAME_S := $(shell uname -s 2>/dev/null || echo Windows)
# 目标文件名:统一使用 .exe 后缀(方便 Python 跨平台调用)
TARGET = build/dynamics_c.exe
# DLL 目标(平台自动选择后缀)
ifeq ($(UNAME_S),Linux)
DLL_TARGET = build/dynamics_c.so
DLL_FLAGS = -shared -fPIC
else ifeq ($(UNAME_S),Darwin)
DLL_TARGET = build/dynamics_c.dylib
DLL_FLAGS = -dynamiclib
else
DLL_TARGET = build/dynamics_c.dll
DLL_FLAGS = -shared
# Windows 检测:Msys2/MINGW 也视为 Windows
IS_WINDOWS := $(findstring MINGW,$(UNAME_S))
ifneq ($(IS_WINDOWS),)
UNAME_S := Windows
endif
IS_WINDOWS := $(findstring MSYS,$(UNAME_S))
ifneq ($(IS_WINDOWS),)
UNAME_S := Windows
endif
# ── 本地编译 ─────────────────────────────────
.PHONY: all dll clean linux windows macos
# DLL 输出到 engines/release/
DLL_DIR = ../../release
all: $(TARGET)
ifeq ($(UNAME_S),Linux)
DLL_TARGET = $(DLL_DIR)/dynamics_c.so
DLL_FLAGS = -shared -fPIC
else ifeq ($(UNAME_S),Darwin)
DLL_TARGET = $(DLL_DIR)/dynamics_c.dylib
DLL_FLAGS = -dynamiclib
else
# Windows: 静态链接运行时,避免依赖 libgcc_s_seh-1.dll
DLL_TARGET = $(DLL_DIR)/dynamics_c.dll
DLL_FLAGS = -shared -static
endif
.PHONY: all dll clean
all: dll
dll: $(DLL_TARGET)
$(TARGET): $(SRCS) | build
$(CC) $(CFLAGS) -o $@ $(SRCS) $(LDFLAGS)
@echo " === C engine built: $@ ==="
$(DLL_TARGET): $(LIB_SRC) | build
$(DLL_TARGET): $(LIB_SRC) | $(DLL_DIR)
$(CC) $(CFLAGS) $(DLL_FLAGS) -o $@ $(LIB_SRC) $(LDFLAGS)
@echo " === C DLL built: $@ ==="
build:
mkdir -p build
# ── 交叉编译 ─────────────────────────────────
# Linux → Linux (x86_64)
linux: CROSS_PREFIX = x86_64-linux-gnu-
linux: CC = $(CROSS_PREFIX)gcc
linux: CFLAGS = -O3 -march=x86-64 -Wall -Wextra
linux: $(SRCS) | build
$(CC) $(CFLAGS) -o build/dynamics_c_linux.exe $(SRCS) $(LDFLAGS)
@echo " === Linux binary: build/dynamics_c_linux.exe ==="
# 任意平台 → Windows (x86_64)
# 需要安装 MinGW 交叉编译器:
# apt install mingw-w64 (Debian/Ubuntu)
# brew install mingw-w64 (macOS)
windows: CROSS_PREFIX = x86_64-w64-mingw32-
windows: CC = $(CROSS_PREFIX)gcc
windows: CFLAGS = -O3 -march=x86-64 -Wall -Wextra
windows: $(SRCS) | build
$(CC) $(CFLAGS) -o build/dynamics_c_win.exe $(SRCS) $(LDFLAGS)
@echo " === Windows binary: build/dynamics_c_win.exe ==="
# 任意平台 → macOS (x86_64)
# 需要安装 osxcross 工具链
macos: CROSS_PREFIX = x86_64-apple-darwin-
macos: CC = $(CROSS_PREFIX)gcc
macos: CFLAGS = -O3 -march=x86-64 -Wall -Wextra
macos: $(SRCS) | build
$(CC) $(CFLAGS) -o build/dynamics_c_mac.exe $(SRCS) $(LDFLAGS)
@echo " === macOS binary: build/dynamics_c_mac.exe ==="
# ── 编译所有平台 ──────────────────────────────
all-platforms: linux windows macos
$(DLL_DIR):
mkdir -p $(DLL_DIR)
clean:
rm -rf build *.o
rm -f $(DLL_TARGET)
-1114
View File
File diff suppressed because it is too large Load Diff
+25 -24
View File
@@ -1,49 +1,50 @@
# engines/cpp/Makefile
# engines/src/cpp/Makefile
# 编译 DLL 到 engines/release/(主程序通过 ctypes 直接调用)
CXX = g++
SRCS = main.cpp
LIB_SRC = dynamics_lib.cpp
UNAME_S := $(shell uname -s 2>/dev/null || echo Windows)
CXXFLAGS = -O3 -march=native -std=c++17 -Wall -Wextra -D_USE_MATH_DEFINES
# Windows 下静态链接运行时,避免 libstdc++-6.dll / libgcc_s_seh-1.dll 版本冲突
ifeq ($(UNAME_S),Windows)
STATIC_FLAGS = -static-libgcc -static-libstdc++
else
STATIC_FLAGS =
# Windows 检测:Msys2/MINGW 也视为 Windows
IS_WINDOWS := $(findstring MINGW,$(UNAME_S))
ifneq ($(IS_WINDOWS),)
UNAME_S := Windows
endif
IS_WINDOWS := $(findstring MSYS,$(UNAME_S))
ifneq ($(IS_WINDOWS),)
UNAME_S := Windows
endif
TARGET = build/dynamics_cpp.exe
CXXFLAGS = -O3 -march=native -std=c++17 -Wall -Wextra -D_USE_MATH_DEFINES
# DLL 输出到 engines/release/
DLL_DIR = ../../release
ifeq ($(UNAME_S),Linux)
DLL_TARGET = build/dynamics_cpp.so
DLL_TARGET = $(DLL_DIR)/dynamics_cpp.so
DLL_FLAGS = -shared -fPIC
else ifeq ($(UNAME_S),Darwin)
DLL_TARGET = build/dynamics_cpp.dylib
DLL_TARGET = $(DLL_DIR)/dynamics_cpp.dylib
DLL_FLAGS = -dynamiclib
else
DLL_TARGET = build/dynamics_cpp.dll
DLL_FLAGS = -shared
# Windows: 完全静态链接,避免依赖运行时 DLL
DLL_TARGET = $(DLL_DIR)/dynamics_cpp.dll
DLL_FLAGS = -shared -static
endif
.PHONY: all dll clean
all: $(TARGET)
all: dll
dll: $(DLL_TARGET)
$(TARGET): $(SRCS) | build
$(CXX) $(CXXFLAGS) $(STATIC_FLAGS) -o $@ $(SRCS)
@echo " === C++ engine built: $@ ==="
$(DLL_TARGET): $(LIB_SRC) | build
$(CXX) $(CXXFLAGS) $(STATIC_FLAGS) $(DLL_FLAGS) -o $@ $(LIB_SRC)
$(DLL_TARGET): $(LIB_SRC) | $(DLL_DIR)
$(CXX) $(CXXFLAGS) $(DLL_FLAGS) -o $@ $(LIB_SRC)
@echo " === C++ DLL built: $@ ==="
build:
mkdir -p build
$(DLL_DIR):
mkdir -p $(DLL_DIR)
clean:
rm -rf build *.o
rm -f $(DLL_TARGET)
File diff suppressed because it is too large Load Diff
+23 -21
View File
@@ -1,47 +1,49 @@
# engines/fortran/Makefile
# engines/src/fortran/Makefile
# 编译 DLL 到 engines/release/(主程序通过 ctypes 直接调用)
FC = gfortran
FFLAGS = -O3 -march=native -Wall -Wextra
SRCS = main.f90
LIB_SRC = dynamics_lib.f90
UNAME_S := $(shell uname -s 2>/dev/null || echo Windows)
ifeq ($(UNAME_S),Windows)
STATIC_FLAGS = -static-libgcc -static-libgfortran -static-libquadmath
else
STATIC_FLAGS =
# Windows 检测:Msys2/MINGW 也视为 Windows
IS_WINDOWS := $(findstring MINGW,$(UNAME_S))
ifneq ($(IS_WINDOWS),)
UNAME_S := Windows
endif
IS_WINDOWS := $(findstring MSYS,$(UNAME_S))
ifneq ($(IS_WINDOWS),)
UNAME_S := Windows
endif
TARGET = build/dynamics_f90.exe
# DLL 输出到 engines/release/
DLL_DIR = ../../release
ifeq ($(UNAME_S),Linux)
DLL_TARGET = build/dynamics_f90.so
DLL_TARGET = $(DLL_DIR)/dynamics_f90.so
DLL_FLAGS = -shared -fPIC
else ifeq ($(UNAME_S),Darwin)
DLL_TARGET = build/dynamics_f90.dylib
DLL_TARGET = $(DLL_DIR)/dynamics_f90.dylib
DLL_FLAGS = -dynamiclib
else
DLL_TARGET = build/dynamics_f90.dll
DLL_FLAGS = -shared -fPIC
# Windows: 完全静态链接,避免依赖 libgfortran-5.dll
DLL_TARGET = $(DLL_DIR)/dynamics_f90.dll
DLL_FLAGS = -shared -fPIC -static
endif
.PHONY: all dll clean
all: $(TARGET)
all: dll
dll: $(DLL_TARGET)
$(TARGET): $(SRCS) | build
$(FC) $(FFLAGS) $(STATIC_FLAGS) -o $@ $(SRCS)
@echo " === Fortran engine built: $@ ==="
$(DLL_TARGET): $(LIB_SRC) | build
$(FC) $(FFLAGS) $(STATIC_FLAGS) $(DLL_FLAGS) -o $@ $(LIB_SRC)
$(DLL_TARGET): $(LIB_SRC) | $(DLL_DIR)
$(FC) $(FFLAGS) $(DLL_FLAGS) -o $@ $(LIB_SRC)
@echo " === Fortran DLL built: $@ ==="
build:
mkdir -p build
$(DLL_DIR):
mkdir -p $(DLL_DIR)
clean:
rm -rf build *.o *.mod
rm -f $(DLL_TARGET)
File diff suppressed because it is too large Load Diff
+23 -10
View File
@@ -169,7 +169,7 @@
<body>
<div class="container">
<h1>Dynamics 示例案例 <small>v2.0</small></h1>
<h1>Dynamics 示例案例 <small>v2.1</small></h1>
<p class="subtitle">10 个从简单到复杂的物理模拟案例,展示分子动力学模拟框架的多种应用场景</p>
<h2>📋 案例总览</h2>
@@ -210,7 +210,7 @@
<div class="case-card">
<span class="num">04</span>
<h3>日地月系统(稳定)</h3>
<p>三体系统稳定椭圆轨道</p>
<p>三体系统稳定轨道,经希尔半径检验</p>
<div class="meta">
<span class="tag tag-python">Python</span>
<span class="tag tag-gravity">万有引力</span>
@@ -263,7 +263,7 @@
<div class="case-card">
<span class="num">09</span>
<h3>一维链纵波·Fortran 引擎</h3>
<p>Fortran 引擎驱动的纵波传播测试</p>
<p>Fortran 引擎驱动的纵波GPU 实例化渲染</p>
<div class="meta">
<span class="tag tag-fortran">Fortran</span>
<span class="tag tag-spring">弹簧</span>
@@ -323,8 +323,8 @@
<li class="tag tag-python">Python 引擎</li>
<li class="tag tag-gravity">万有引力</li>
</ul>
<p>与 case03 相同的三体系统,但采用恰当的初始条件地球和月球维持稳定的椭圆轨道运动</p>
<div class="highlight">✅ 与 case03 对比学习:初始条件对数值稳定性的影响</div>
<p>与 case03 相同的三体系统,但采用恰当的初始条件地球置于近日点($r=10$),$v_z=520$ 接近圆轨道;月球缩至希尔半径($r_H\approx1.0$)以内的 $r=0.5$,相对速度 $v_{\text{rel}}\approx127$,确保月球被地球稳定束缚</p>
<div class="highlight">✅ 与 case03 对比学习:初始条件对数值稳定性的影响。地月距 0.5 在地球希尔半径以内,满足稳定性条件</div>
</div>
<div class="detail-card">
@@ -377,8 +377,8 @@
<li class="tag tag-spring">弹簧键力</li>
<li class="tag tag-drive">驱动力</li>
</ul>
<p>40 原子沿 x 轴排列,Fortran 引擎驱动的纵波传播测试。验证 Fortran 引擎的输出兼容性。</p>
<div class="highlight">🔧 Fortran 引擎兼容性验证,T_total=200, NSTEP=100</div>
<p>40 原子沿 x 轴排列,Fortran 引擎驱动的纵波传播测试。使用 <code>use_marker: 1</code>GPU 实例化 Marker 模式)加速渲染,验证 Fortran 引擎与其他引擎的输出兼容性。</p>
<div class="highlight">🔧 Fortran 引擎兼容性验证,T_total=200, NSTEP=100。Marker 模式下 40 原子以 GPU 点精灵渲染,帧率高</div>
</div>
<div class="detail-card">
@@ -412,7 +412,12 @@
<h3>配置</h3>
<p style="color:var(--text-dim); font-size:0.875rem;">
每个案例的 <code>input/input.txt</code> 可配置物理参数、力开关、算法、引擎、渲染方式等。
从 case06 起支持 <code>save_trajectory</code> 开关、摄像机初始位置、<code>display_amp</code> 视觉放大等高级功能。
</p>
<h3>引擎架构</h3>
<p style="color:var(--text-dim); font-size:0.875rem;">
外部引擎(C / C++ / Fortran)以 <strong>DLL 方式</strong> 运行,主程序通过 ctypes 在进程内直接调用,不启动子进程。DLL 预编译在 <code>engines/release/</code> 中,源码位于 <code>engines/src/{c,cpp,fortran}/</code>。重新编译:
<code>cd engines/src/c && make dll</code>
</p>
</div>
@@ -421,10 +426,18 @@
<pre style="font-size:0.825rem; color:var(--text-dim); line-height:1.5;">
dynamics/
├── dynamics.py # 统一运行入口
├── compute.py # Python 物理引擎
├── compute.py # 物理引擎 + 显示数据生成
├── draw.py # VisPy 3D 动画
├── plot_wave.py # 波形能量图
├── engines/ # C / C++ / Fortran 引擎
├── .gitattributes # DLL/二进制文件保护
├── engines/
│ ├── engine_dll.py # DLL 加载器(ctypes
│ ├── python/ # Python 引擎(dynamics_lib.py
│ ├── release/ # 预编译 DLLC / C++ / Fortran
│ └── src/ # 引擎源码
│ ├── c/ # C 源码 + Makefile → dynamics_c.dll
│ ├── cpp/ # C++ 源码 + Makefile → dynamics_cpp.dll
│ └── fortran/ # Fortran 源码 + Makefile → dynamics_f90.dll
├── examples/ # 案例目录
│ ├── case01/ ~ case10/
└── output/ # 默认输出目录
+92
View File
@@ -0,0 +1,92 @@
"""
为指定案例添加次紧邻键 (k2, k=100, r0=1.41421356)
用法: python add_k2.py case16
python add_k2.py case16 case17 case18
python add_k2.py --all
"""
import os
import sys
def add_k2(case_dir):
coord_path = os.path.join(case_dir, "input", "coord.txt")
conn_path = os.path.join(case_dir, "input", "connection.txt")
bond_path = os.path.join(case_dir, "input", "bond.txt")
if not os.path.exists(coord_path):
print(f" [跳过] {case_dir}: 找不到 coord.txt")
return False
# 读取 coord.txt 获取网格尺寸
with open(coord_path, "r", encoding="utf-8") as f:
lines = f.readlines()
n_atoms = len(lines) - 1 # 去掉表头
N = int(n_atoms ** 0.5)
if N * N != n_atoms:
print(f" [跳过] {case_dir}: 非正方形网格 (n_atoms={n_atoms})")
return False
print(f" {case_dir}: {N}x{N} 网格")
# 读取现有 connection.txt,检查是否已有 k2
has_k2 = False
if os.path.exists(conn_path):
with open(conn_path, "r") as f:
for line in f:
if "k2" in line:
has_k2 = True
break
if has_k2:
print(f" k2 已存在,跳过 connection.txt")
else:
# 追加 k2 键到 connection.txt
with open(conn_path, "a", encoding="utf-8") as f:
cnt = 0
for row in range(N):
for col in range(N):
id1 = row * N + col + 1
if col + 1 < N and row + 1 < N:
f.write(f"{id1} {(row + 1) * N + (col + 1) + 1} k2\n")
cnt += 1
if col - 1 >= 0 and row + 1 < N:
f.write(f"{id1} {(row + 1) * N + (col - 1) + 1} k2\n")
cnt += 1
print(f" connection.txt: 追加 {cnt} 条 k2 键")
# 检查 bond.txt 是否有 k2
has_bond = False
if os.path.exists(bond_path):
with open(bond_path, "r") as f:
for line in f:
if line.startswith("k2"):
has_bond = True
break
if has_bond:
print(f" bond.txt: k2 已存在")
else:
with open(bond_path, "a", encoding="utf-8") as f:
f.write("k2 100.0 1.41421356\n")
print(f" bond.txt: 追加 k2 定义")
return True
if __name__ == "__main__":
targets = []
if "--all" in sys.argv:
base = os.path.dirname(os.path.abspath(__file__))
for d in sorted(os.listdir(base)):
if d.startswith("case") and os.path.isdir(os.path.join(base, d)):
targets.append(os.path.join(base, d))
else:
for arg in sys.argv[1:]:
if arg.startswith("--"):
continue
p = arg if os.path.isabs(arg) else os.path.join(os.path.dirname(os.path.abspath(__file__)), arg)
targets.append(p)
for t in targets:
add_k2(t)
+1 -1
View File
@@ -6,7 +6,7 @@
# 每步用 0/1 单独开关,1=执行,0=跳过
# 依赖关系:抽帧依赖模拟结果,绘图依赖模拟+抽帧
step_simulate: 1 # 运行物理模拟 → output/trajectory.txt
step_sample: 1 # 抽帧 → output/display.txt
step_sample: 0 # 抽帧 → output/display.txt
step_plot: 1 # 绘制轨迹/能量图 → output/trajectory_plots.png
step_plot_wave: 0 # 绘制波形能量动画
plot_wave_save_gif: 0 # 输出波形 GIF(需 step_plot_wave=1
+829
View File
@@ -0,0 +1,829 @@
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>case04 — 日地月三体系统</title>
<script>
MathJax = {
tex: { inlineMath: [['$','$'], ['\\(','\\)']], displayMath: [['$$','$$'], ['\\[','\\]']] },
svg: { fontCache: 'global' }
};
</script>
<script src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-svg.js" async></script>
<style>
:root {
--bg: #0d1117;
--surface: #161b22;
--border: #30363d;
--text: #c9d1d9;
--text-dim: #8b949e;
--accent: #58a6ff;
--green: #3fb950;
--orange: #d29922;
--red: #f85149;
--teal: #56d4dd;
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", "Noto Sans SC", Helvetica, Arial, sans-serif;
background: var(--bg);
color: var(--text);
line-height: 1.8;
padding: 40px 24px;
}
.container { max-width: 860px; margin: 0 auto; }
h1 { font-size: 1.8rem; margin-bottom: 6px; color: #f0f6fc; }
h1 small { font-size: 1rem; color: var(--text-dim); font-weight: 400; }
.subtitle { color: var(--text-dim); margin-bottom: 32px; }
h2 {
font-size: 1.35rem;
margin-top: 36px;
margin-bottom: 14px;
padding-bottom: 6px;
border-bottom: 1px solid var(--border);
color: #f0f6fc;
}
h3 { font-size: 1.1rem; margin-top: 24px; margin-bottom: 10px; color: #f0f6fc; }
p { margin-bottom: 12px; color: var(--text); }
.card {
background: var(--surface);
border: 1px solid var(--border);
border-radius: 8px;
padding: 20px 24px;
margin-bottom: 20px;
}
table {
width: 100%;
border-collapse: collapse;
margin: 12px 0;
font-size: 0.9rem;
}
th, td {
text-align: left;
padding: 8px 14px;
border-bottom: 1px solid var(--border);
}
th { color: var(--text-dim); font-weight: 600; background: #1c2128; }
.highlight {
background: #1f242e;
border-left: 3px solid var(--accent);
padding: 10px 16px;
margin: 12px 0;
border-radius: 0 6px 6px 0;
font-size: 0.9rem;
}
.highlight-eq {
background: #1f242e;
border-left: 3px solid var(--teal);
padding: 14px 18px;
margin: 14px 0;
border-radius: 0 6px 6px 0;
}
.highlight-eq p { margin: 6px 0; }
code {
background: #1c2128;
padding: 2px 6px;
border-radius: 4px;
font-family: "SF Mono", "Fira Code", monospace;
font-size: 0.85em;
color: var(--green);
}
.tag {
display: inline-block;
padding: 2px 10px;
border-radius: 4px;
font-size: 0.78rem;
font-weight: 500;
margin: 2px;
}
.tag-python { background: #3572A533; color: #3572A5; }
.tag-gravity { background: #d2992233; color: var(--orange); }
ul, ol { margin: 8px 0 12px 24px; }
li { margin-bottom: 4px; }
.formula-block {
overflow-x: auto;
padding: 8px 0;
}
.diagram-wrap {
background: #0d1117;
border: 1px solid var(--border);
border-radius: 8px;
padding: 16px;
margin: 16px 0;
text-align: center;
}
.diagram-wrap canvas {
display: block;
margin: 0 auto;
max-width: 100%;
height: auto;
}
.diagram-controls {
display: flex;
flex-wrap: wrap;
align-items: center;
justify-content: center;
gap: 16px;
margin-top: 12px;
}
.diagram-controls label {
font-size: 0.85rem;
color: var(--text-dim);
}
.diagram-controls input[type="range"] {
width: 200px;
accent-color: var(--accent);
}
.diagram-controls .val {
font-size: 0.9rem;
font-weight: 500;
color: var(--teal);
min-width: 48px;
display: inline-block;
text-align: center;
}
.diagram-info {
display: flex;
flex-wrap: wrap;
justify-content: center;
gap: 20px;
margin-top: 10px;
font-size: 0.82rem;
color: var(--text-dim);
}
.diagram-info span {
display: inline-flex;
align-items: center;
gap: 4px;
}
.diagram-info .dot {
display: inline-block;
width: 10px;
height: 10px;
border-radius: 50%;
margin-right: 2px;
}
@media (max-width: 640px) {
body { padding: 16px; }
.diagram-controls { flex-direction: column; gap: 8px; }
}
</style>
</head>
<body>
<div class="container">
<h1>case04 — 日地月三体系统 <small>稳定轨道版本</small></h1>
<p class="subtitle">太阳、地球、月球三体系统,采用真实比例的质量和万有引力模拟,地球和月球维持稳定椭圆轨道。</p>
<div class="card" style="display:flex; flex-wrap:wrap; gap:8px; align-items:center; margin-bottom:20px;">
<span class="tag tag-python">Python 引擎</span>
<span class="tag tag-gravity">万有引力</span>
<span style="color:var(--text-dim); font-size:0.85rem; margin-left:8px;">3 粒子 | leapfrog 算法 | 稳定轨道</span>
</div>
<h2>一、物理系统概览</h2>
<div class="card">
<p>本案例模拟真实的日-地-月三体系统:</p>
<ul>
<li><strong>太阳</strong>:位于原点,质量为 $M_\odot$,固定不动(fix_x=fix_y=fix_z=1</li>
<li><strong>地球</strong>:绕太阳公转,质量为 $M_\oplus$,近日点出发</li>
<li><strong>月球</strong>:绕地球公转(同时跟随地球绕太阳),质量为 $M_\text{moon}$</li>
</ul>
<p>原子间万有引力由 <code>gravity_strength</code> 缩放控制,$F = G \dfrac{m_1 m_2}{r^2}$。</p>
</div>
<h2>二、天体真实参数</h2>
<h3>2.1 质量</h3>
<table>
<tr><th>天体</th><th>质量 / kg</th><th>模拟质量(月球=1</th></tr>
<tr><td>太阳</td><td>$1.989 \times 10^{30}$</td><td>$27\,000$</td></tr>
<tr><td>地球</td><td>$5.972 \times 10^{24}$</td><td>$81$</td></tr>
<tr><td>月球</td><td>$7.35 \times 10^{22}$</td><td>$1$</td></tr>
</table>
<div class="highlight">
<strong>质量比</strong>$M_\odot : M_\oplus : M_\text{moon} \approx 27\,100\,000 : 81.3 : 1$<br>
模拟中采用缩放比例 $27\,000 : 81 : 1$(太阳质量缩至 $1/1000$ 以保持数值稳定)。
</div>
<h3>2.2 轨道参数</h3>
<table>
<tr><th>轨道</th><th>半长轴 $a$</th><th>偏心率 $e$</th><th>半短轴 $b$</th><th>周期</th></tr>
<tr><td>地球绕太阳</td><td>$1.4960 \times 10^8$ km (1 AU)</td><td>$0.0167$</td><td>$1.4958 \times 10^8$ km</td><td>365.25 天</td></tr>
<tr><td>月球绕地球</td><td>$3.844 \times 10^5$ km</td><td>$0.0549$</td><td>$3.838 \times 10^5$ km</td><td>27.32 天</td></tr>
</table>
<div class="highlight">
<strong>比例</strong>$R_{\text{日地}} : R_{\text{地月}} \approx 389 : 1$<br>
地月距离约为日地距离的 $1/389$。
</div>
<h2>三、轨道力学</h2>
<h3>3.1 交互式轨道示意图</h3>
<p>拖动下方滑块改变偏心率 $e$,观察轨道形状的变化和参数标注:</p>
<div class="diagram-wrap">
<canvas id="orbitCanvas" width="760" height="400"></canvas>
<div class="diagram-controls">
<label>偏心率 e = <span class="val" id="eDisplay">0.0170</span>(地球真实值 0.0167</label>
<input type="range" id="eSlider" min="0" max="850" value="17">
</div>
<div class="diagram-info" id="diagramInfo"></div>
</div>
<h3>3.2 月球绕地球轨道</h3>
<p>拖动下方滑块改变月球轨道偏心率 $e_\text{moon}$(真实值 0.0549):</p>
<div class="diagram-wrap">
<canvas id="moonCanvas" width="760" height="400"></canvas>
<div class="diagram-controls">
<label>偏心率 e = <span class="val" id="moonEDisplay">0.0549</span>(月球真实值 0.0549</label>
<input type="range" id="moonESlider" min="0" max="850" value="55">
</div>
<div class="diagram-info" id="moonInfo"></div>
</div>
<h3>3.3 偏心率定义</h3>
<div class="highlight-eq">
<p>偏心率 $e$ 描述椭圆轨道偏离正圆的程度:</p>
<div class="formula-block">
$$e = \frac{c}{a}$$
</div>
<p>其中 $c = ea$ 为偏心距(焦点到椭圆中心的距离),$a$ 为半长轴。</p>
</div>
<table>
<tr><th>$e$ 值</th><th>轨道形状</th><th>说明</th></tr>
<tr><td>$e = 0$</td><td>正圆形</td><td>速度恒定,距离恒定</td></tr>
<tr><td>$0 &lt; e &lt; 1$</td><td>椭圆</td><td>近日点/近地点速度最大,远日点/远地点速度最小</td></tr>
<tr><td>$e = 1$</td><td>抛物线</td><td>逃逸轨道,速度恰好达到逃逸速度</td></tr>
<tr><td>$e &gt; 1$</td><td>双曲线</td><td>飞越轨道,速度超过逃逸速度</td></tr>
</table>
<h3>3.4 半长轴与半短轴的关系</h3>
<div class="highlight-eq">
<div class="formula-block">
$$b = a\sqrt{1 - e^2}$$
</div>
<p>当 $e \ll 1$ 时,$b \approx a\left(1 - \dfrac{e^2}{2}\right)$,椭圆度非常微小。</p>
</div>
<h3>3.5 近日点与远日点</h3>
<div class="highlight-eq">
<p>近日点(距太阳最近)和远日点(距太阳最远)的距离分别为:</p>
<div class="formula-block">
$$r_{\text{peri}} = a(1 - e), \qquad r_{\text{ap}} = a(1 + e)$$
</div>
<p>在近日点轨道速度最大,在远日点轨道速度最小:</p>
<div class="formula-block">
$$v_{\text{peri}} = \sqrt{\frac{GM(1+e)}{a(1-e)}}, \qquad
v_{\text{ap}} = \sqrt{\frac{GM(1-e)}{a(1+e)}}$$
</div>
</div>
<h3>3.6 开普勒三大定律</h3>
<ol>
<li><strong>椭圆定律</strong>:行星轨道是椭圆,太阳位于椭圆的一个焦点上。</li>
<li><strong>面积定律</strong>:行星与太阳的连线在相等时间内扫过相等的面积。</li>
<li><strong>周期定律</strong>:公转周期的平方与半长轴的立方成正比:$T^2 \propto a^3$。</li>
</ol>
<div class="highlight">
<strong>验证</strong>:在本模拟中,你可以通过轨迹图观察面积定律是否成立——地球在近日点移动更快,远日点移动更慢。
</div>
<h2>四、模拟参数</h2>
<h3>4.1 缩放说明</h3>
<p>真实尺度无法直接用于模拟(日地距离 1.5 亿 km),因此采用缩放参数。当前 case04 的配置为教学演示而简化:</p>
<table>
<tr><th>参数</th><th></th><th>说明</th></tr>
<tr><td><code>gravity_strength</code></td><td>100.0</td><td>万有引力强度</td></tr>
<tr><td><code>box_a</code></td><td>30.0</td><td>盒子半边长</td></tr>
<tr><td><code>method</code></td><td>leapfrog</td><td>蛙跳法(辛积分器,能量守恒)</td></tr>
<tr><td><code>T_total</code></td><td>10.0 s</td><td>总模拟时间</td></tr>
<tr><td><code>NSTEP</code></td><td>2</td><td>抽帧间隔(密采帧)</td></tr>
</table>
<h3>4.2 初始构型</h3>
<table>
<tr><th>天体</th><th>质量</th><th>位置 $(x,y,z)$</th><th>速度 $(v_x,v_y,v_z)$</th><th>固定约束</th></tr>
<tr><td>太阳</td><td>$27\,000$</td><td>$(0,0,0)$</td><td>$(0,0,0)$</td><td>全部固定</td></tr>
<tr><td>地球</td><td>$81$</td><td>$(10,0,0)$</td><td>$(0,0,520)$</td><td></td></tr>
<tr><td>月球</td><td>$1$</td><td>$(10.5,0,0)$</td><td>$(0,0,647)$</td><td></td></tr>
</table>
<p>地球在 $z$ 方向获得初速度 $v=520$,产生绕太阳的轨道运动(接近圆轨道);月球在地球基础上附加 $v\approx127$ 的绕地速度,形成绕地球的轨道。速度由圆形轨道公式 $v = \sqrt{G_{\text{eff}} M / r}$ 计算,其中 $G_{\text{eff}} = \text{gravity\_strength} = 100$。</p>
<div class="highlight">
<strong>验证</strong>:地球速度 $v_\oplus = \sqrt{100 \times 27\,000 / 10} \approx 519.6$,设 520 正确。月球相对速度 $v_{\text{rel}} = \sqrt{100 \times 81 / 0.5} \approx 127.3$,设 127 正确。地月距离 0.5 在地球希尔半径 $r_H \approx 10 \times (81 / 81\,000)^{1/3} \approx 1.0$ 之内,可确保轨道稳定。
</div>
<h2>五、使用方法</h2>
<div class="card">
<code style="display:block; padding:14px 18px; margin-bottom:10px;">
# 进入 case04 目录并运行<br>
cd examples/case04<br>
python run_dynamics.py<br><br>
# 仅输出轨迹图,跳过动画<br>
python run_dynamics.py --no-plot
</code>
<p>配置文件:<code>input/input.txt</code>(物理参数)、<code>input/coord.txt</code>(初始位置/速度)。</p>
</div>
<h2>六、与 case03 的对比</h2>
<table>
<tr><th></th><th>case03(失败案例)</th><th>case04(成功案例)</th></tr>
<tr><td>轨道状态</td><td>轨道发散或碰撞</td><td>稳定椭圆轨道</td></tr>
<tr><td>关键差异</td><td>初始速度或质量比不恰当</td><td>合理的初值和参数</td></tr>
<tr><td>教学意义</td><td>展示参数选择的重要性</td><td>展示正确的三体运动</td></tr>
</table>
<div class="highlight">
<strong>教学建议</strong>:先运行 case03 观察失稳,再运行 case04 对比稳定轨道,理解初始条件对数值模拟的关键影响。
</div>
<h2>七、已知局限</h2>
<ul>
<li>太阳真实质量为月球 $27\,100\,000$ 倍,模拟中缩至 $27\,000$ 倍以保持数值稳定($1/1000$)</li>
<li>轨道半径未按真实比例缩放(真实日地距是地月距的 389 倍,模拟中约为 20 倍,受希尔半径约束,月球已尽可能放远)</li>
<li>未考虑月球轨道倾角(真实地月轨道有约 5° 的倾角)</li>
<li>leapfrog 算法为能量守恒的辛积分器,长期稳定,但步长过大时仍可能偏离真实轨道</li>
<li>太阳固定不动可作为教学简化,但严格三体模拟中应让所有天体在质心系中自由运动</li>
</ul>
</div>
<script>
(function() {
var canvas = document.getElementById('orbitCanvas');
var ctx = canvas.getContext('2d');
var slider = document.getElementById('eSlider');
var eDisplay = document.getElementById('eDisplay');
var infoDiv = document.getElementById('diagramInfo');
var W = 760, H = 400;
var cx = 380, cy = 200;
// 实际绘图区域半径
var A = 140;
function draw(e) {
ctx.clearRect(0, 0, W, H);
var b = A * Math.sqrt(1 - e * e);
var f = e * A;
// 长轴辅助线
ctx.beginPath();
ctx.moveTo(cx - A - 20, cy);
ctx.lineTo(cx + A + 20, cy);
ctx.strokeStyle = '#30363d';
ctx.lineWidth = 0.5;
ctx.setLineDash([4, 3]);
ctx.stroke();
ctx.setLineDash([]);
// 椭圆轨道(中心在 cx+f,左焦点=cx 为太阳)
ctx.beginPath();
ctx.ellipse(cx + f, cy, A, b, 0, 0, Math.PI * 2);
ctx.strokeStyle = '#58a6ff';
ctx.lineWidth = 1.5;
ctx.stroke();
// 半长轴 a(标注线,从椭圆中心到右顶点)
var aY = cy + 28;
ctx.beginPath();
ctx.moveTo(cx + f, aY);
ctx.lineTo(cx + f + A, aY);
ctx.strokeStyle = '#c9d1d9';
ctx.lineWidth = 1.5;
ctx.stroke();
// 箭头(左端:椭圆中心 cx+f)
ctx.beginPath();
ctx.moveTo(cx + f + 6, aY - 4);
ctx.lineTo(cx + f, aY);
ctx.lineTo(cx + f + 6, aY + 4);
ctx.strokeStyle = '#c9d1d9';
ctx.lineWidth = 1.5;
ctx.stroke();
// 箭头(右端:右顶点 cx+f+A)
ctx.beginPath();
ctx.moveTo(cx + f + A - 6, aY - 4);
ctx.lineTo(cx + f + A, aY);
ctx.lineTo(cx + f + A - 6, aY + 4);
ctx.strokeStyle = '#c9d1d9';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.fillStyle = '#c9d1d9';
ctx.font = '13px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'top';
ctx.fillText('a = ' + A.toFixed(0), cx + f + A/2, aY + 6);
// 偏心距 c = ea(从左焦点到椭圆中心)
var cY = cy - b - 20;
ctx.beginPath();
ctx.moveTo(cx, cY);
ctx.lineTo(cx + f, cY);
ctx.strokeStyle = '#d29922';
ctx.lineWidth = 1.5;
ctx.stroke();
// 箭头(左端:左焦点=cx 太阳)
ctx.beginPath();
ctx.moveTo(cx + 5, cY - 4);
ctx.lineTo(cx, cY);
ctx.lineTo(cx + 5, cY + 4);
ctx.strokeStyle = '#d29922';
ctx.lineWidth = 1.5;
ctx.stroke();
// 箭头(右端:椭圆中心 cx+f)
ctx.beginPath();
ctx.moveTo(cx + f - 5, cY - 4);
ctx.lineTo(cx + f, cY);
ctx.lineTo(cx + f - 5, cY + 4);
ctx.strokeStyle = '#d29922';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.fillStyle = '#d29922';
ctx.font = '13px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'bottom';
ctx.fillText('c = ea = ' + f.toFixed(1), cx + f/2, cY - 4);
// 半短轴 b 标注(从椭圆中心到上顶点)
var bX = cx + f + A + 18;
ctx.beginPath();
ctx.moveTo(bX, cy);
ctx.lineTo(bX, cy - b);
ctx.strokeStyle = '#3fb950';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.beginPath();
ctx.moveTo(bX - 4, cy - 6);
ctx.lineTo(bX, cy);
ctx.lineTo(bX + 4, cy - 6);
ctx.strokeStyle = '#3fb950';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.beginPath();
ctx.moveTo(bX - 4, cy - b + 6);
ctx.lineTo(bX, cy - b);
ctx.lineTo(bX + 4, cy - b + 6);
ctx.strokeStyle = '#3fb950';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.fillStyle = '#3fb950';
ctx.font = '13px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'left';
ctx.textBaseline = 'middle';
ctx.fillText('b = ' + b.toFixed(1), bX + 8, cy - b/2);
// 太阳(焦点)
ctx.beginPath();
ctx.arc(cx, cy, 7, 0, Math.PI * 2);
ctx.fillStyle = '#d29922';
ctx.fill();
ctx.strokeStyle = '#f0883e';
ctx.lineWidth = 1;
ctx.stroke();
ctx.fillStyle = '#f0f6fc';
ctx.font = '13px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'bottom';
ctx.fillText('太阳(左焦点)', cx, cy - 12);
// 近日点(左顶点:距左焦点最近)
var periX = cx + f - A;
ctx.fillStyle = '#58a6ff';
ctx.beginPath();
ctx.arc(periX, cy, 4, 0, Math.PI * 2);
ctx.fill();
ctx.fillStyle = '#8b949e';
ctx.font = '12px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'top';
ctx.fillText('近日点 r = a(1-e) = ' + (A * (1 - e)).toFixed(1), periX, cy + 12);
// 远日点标注(右顶点:距左焦点最远)
var apX = cx + f + A;
ctx.fillStyle = '#58a6ff';
ctx.beginPath();
ctx.arc(apX, cy, 4, 0, Math.PI * 2);
ctx.fill();
ctx.fillStyle = '#8b949e';
ctx.font = '12px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'top';
ctx.fillText('远日点 r = a(1+e) = ' + (A * (1 + e)).toFixed(1), apX, cy + 12);
// 地球(轨道上的点)
var angle = 0.4;
var ex = cx + f + A * Math.cos(angle);
var ey = cy - b * Math.sin(angle);
ctx.beginPath();
ctx.arc(ex, ey, 5, 0, Math.PI * 2);
ctx.fillStyle = '#58a6ff';
ctx.fill();
ctx.strokeStyle = '#1f6feb';
ctx.lineWidth = 1;
ctx.stroke();
ctx.fillStyle = '#58a6ff';
ctx.font = '12px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'left';
ctx.textBaseline = 'bottom';
ctx.fillText('地球', ex + 8, ey);
// 速度矢量
var vlen = 32;
var vx = -A * Math.sin(angle) * 0.7;
var vy = b * Math.cos(angle) * 0.7;
var vl = Math.sqrt(vx*vx + vy*vy);
if (vl > 0) {
vx = vx / vl * vlen;
vy = vy / vl * vlen;
}
ctx.beginPath();
ctx.moveTo(ex, ey);
ctx.lineTo(ex + vx, ey + vy);
ctx.strokeStyle = '#3fb950';
ctx.lineWidth = 2;
ctx.stroke();
ctx.beginPath();
var tipX = ex + vx, tipY = ey + vy;
var a1 = 0.3;
ctx.moveTo(tipX, tipY);
ctx.lineTo(tipX - vx * a1 + vy * a1 * 0.5, tipY - vy * a1 - vx * a1 * 0.5);
ctx.lineTo(tipX - vx * a1 - vy * a1 * 0.5, tipY - vy * a1 + vx * a1 * 0.5);
ctx.closePath();
ctx.fillStyle = '#3fb950';
ctx.fill();
// 信息面板
var roundE = Math.round(e * 10000) / 10000;
infoDiv.innerHTML =
'<span><span class="dot" style="background:#58a6ff"></span>椭圆轨道 e = ' + roundE.toFixed(4) + '</span>' +
'<span><span class="dot" style="background:#d29922"></span>偏心距 c = ' + f.toFixed(1) + '</span>' +
'<span><span class="dot" style="background:#3fb950"></span>b/a = ' + (b / A).toFixed(4) + '</span>' +
'<span>近日点 ' + (A * (1 - e)).toFixed(1) + ' / 远日点 ' + (A * (1 + e)).toFixed(1) + '</span>';
}
function update() {
var val = parseInt(slider.value);
var e = val / 1000;
eDisplay.textContent = e.toFixed(4);
draw(e);
}
slider.addEventListener('input', update);
update();
})();
// ── 月球绕地球轨道图 ──
(function() {
var canvas = document.getElementById('moonCanvas');
var ctx = canvas.getContext('2d');
var slider = document.getElementById('moonESlider');
var eDisplay = document.getElementById('moonEDisplay');
var infoDiv = document.getElementById('moonInfo');
var W = 760, H = 400;
var cx = 380, cy = 200;
var A = 140;
function draw(e) {
ctx.clearRect(0, 0, W, H);
var b = A * Math.sqrt(1 - e * e);
var f = e * A;
// 长轴辅助线
ctx.beginPath();
ctx.moveTo(cx - A - 20, cy);
ctx.lineTo(cx + A + 20, cy);
ctx.strokeStyle = '#30363d';
ctx.lineWidth = 0.5;
ctx.setLineDash([4, 3]);
ctx.stroke();
ctx.setLineDash([]);
// 椭圆轨道(中心在 cx+f,左焦点=cx 为地球)
ctx.beginPath();
ctx.ellipse(cx + f, cy, A, b, 0, 0, Math.PI * 2);
ctx.strokeStyle = '#7ee787';
ctx.lineWidth = 1.5;
ctx.stroke();
// 半长轴 a(标注线,从椭圆中心到右顶点)
var aY = cy + 28;
ctx.beginPath();
ctx.moveTo(cx + f, aY);
ctx.lineTo(cx + f + A, aY);
ctx.strokeStyle = '#c9d1d9';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.beginPath();
ctx.moveTo(cx + f + 6, aY - 4);
ctx.lineTo(cx + f, aY);
ctx.lineTo(cx + f + 6, aY + 4);
ctx.strokeStyle = '#c9d1d9';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.beginPath();
ctx.moveTo(cx + f + A - 6, aY - 4);
ctx.lineTo(cx + f + A, aY);
ctx.lineTo(cx + f + A - 6, aY + 4);
ctx.strokeStyle = '#c9d1d9';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.fillStyle = '#c9d1d9';
ctx.font = '13px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'top';
ctx.fillText('a = ' + A.toFixed(0), cx + f + A/2, aY + 6);
// 偏心距 c = ea(从左焦点=地球 到椭圆中心)
var cY = cy - b - 20;
ctx.beginPath();
ctx.moveTo(cx, cY);
ctx.lineTo(cx + f, cY);
ctx.strokeStyle = '#d29922';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.beginPath();
ctx.moveTo(cx + 5, cY - 4);
ctx.lineTo(cx, cY);
ctx.lineTo(cx + 5, cY + 4);
ctx.strokeStyle = '#d29922';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.beginPath();
ctx.moveTo(cx + f - 5, cY - 4);
ctx.lineTo(cx + f, cY);
ctx.lineTo(cx + f - 5, cY + 4);
ctx.strokeStyle = '#d29922';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.fillStyle = '#d29922';
ctx.font = '13px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'bottom';
ctx.fillText('c = ea = ' + f.toFixed(1), cx + f/2, cY - 4);
// 半短轴 b 标注(从椭圆中心到上顶点)
var bX = cx + f + A + 18;
ctx.beginPath();
ctx.moveTo(bX, cy);
ctx.lineTo(bX, cy - b);
ctx.strokeStyle = '#3fb950';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.beginPath();
ctx.moveTo(bX - 4, cy - 6);
ctx.lineTo(bX, cy);
ctx.lineTo(bX + 4, cy - 6);
ctx.strokeStyle = '#3fb950';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.beginPath();
ctx.moveTo(bX - 4, cy - b + 6);
ctx.lineTo(bX, cy - b);
ctx.lineTo(bX + 4, cy - b + 6);
ctx.strokeStyle = '#3fb950';
ctx.lineWidth = 1.5;
ctx.stroke();
ctx.fillStyle = '#3fb950';
ctx.font = '13px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'left';
ctx.textBaseline = 'middle';
ctx.fillText('b = ' + b.toFixed(1), bX + 8, cy - b/2);
// 地球(左焦点)
ctx.beginPath();
ctx.arc(cx, cy, 8, 0, Math.PI * 2);
ctx.fillStyle = '#58a6ff';
ctx.fill();
ctx.strokeStyle = '#1f6feb';
ctx.lineWidth = 1;
ctx.stroke();
ctx.fillStyle = '#f0f6fc';
ctx.font = '13px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'bottom';
ctx.fillText('地球(左焦点)', cx, cy - 14);
// 近地点(左顶点:距地球最近)
var periX = cx + f - A;
ctx.fillStyle = '#7ee787';
ctx.beginPath();
ctx.arc(periX, cy, 4, 0, Math.PI * 2);
ctx.fill();
ctx.fillStyle = '#8b949e';
ctx.font = '12px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'top';
ctx.fillText('近地点 r = a(1-e) = ' + (A * (1 - e)).toFixed(1), periX, cy + 12);
// 远地点(右顶点:距地球最远)
var apX = cx + f + A;
ctx.fillStyle = '#7ee787';
ctx.beginPath();
ctx.arc(apX, cy, 4, 0, Math.PI * 2);
ctx.fill();
ctx.fillStyle = '#8b949e';
ctx.font = '12px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'center';
ctx.textBaseline = 'top';
ctx.fillText('远地点 r = a(1+e) = ' + (A * (1 + e)).toFixed(1), apX, cy + 12);
// 月球(轨道上的点)
var angle = 0.6;
var mx = cx + f + A * Math.cos(angle);
var my = cy - b * Math.sin(angle);
ctx.beginPath();
ctx.arc(mx, my, 4, 0, Math.PI * 2);
ctx.fillStyle = '#e2e8f0';
ctx.fill();
ctx.strokeStyle = '#8b949e';
ctx.lineWidth = 0.5;
ctx.stroke();
ctx.fillStyle = '#e2e8f0';
ctx.font = '12px "Segoe UI", Arial, sans-serif';
ctx.textAlign = 'left';
ctx.textBaseline = 'bottom';
ctx.fillText('月球', mx + 8, my);
// 速度矢量
var vlen = 32;
var vx = -A * Math.sin(angle) * 0.7;
var vy = b * Math.cos(angle) * 0.7;
var vl = Math.sqrt(vx*vx + vy*vy);
if (vl > 0) { vx = vx/vl*vlen; vy = vy/vl*vlen; }
ctx.beginPath();
ctx.moveTo(mx, my);
ctx.lineTo(mx + vx, my + vy);
ctx.strokeStyle = '#3fb950';
ctx.lineWidth = 2;
ctx.stroke();
var tipX = mx + vx, tipY = my + vy;
ctx.beginPath();
ctx.moveTo(tipX, tipY);
ctx.lineTo(tipX - vx*0.3 + vy*0.15, tipY - vy*0.3 - vx*0.15);
ctx.lineTo(tipX - vx*0.3 - vy*0.15, tipY - vy*0.3 + vx*0.15);
ctx.closePath();
ctx.fillStyle = '#3fb950';
ctx.fill();
// 信息面板
var roundE = Math.round(e * 10000) / 10000;
infoDiv.innerHTML =
'<span><span class="dot" style="background:#7ee787"></span>椭圆轨道 e = ' + roundE.toFixed(4) + '</span>' +
'<span><span class="dot" style="background:#d29922"></span>偏心距 c = ' + f.toFixed(1) + '</span>' +
'<span><span class="dot" style="background:#3fb950"></span>b/a = ' + (b / A).toFixed(4) + '</span>' +
'<span>近地点 ' + (A * (1 - e)).toFixed(1) + ' / 远地点 ' + (A * (1 + e)).toFixed(1) + '</span>';
}
function update() {
var val = parseInt(slider.value);
var e = val / 1000;
eDisplay.textContent = e.toFixed(4);
draw(e);
}
slider.addEventListener('input', update);
update();
})();
</script>
</body>
</html>
+3 -3
View File
@@ -1,4 +1,4 @@
n mass radius x y z vx vy vz fix_x fix_y fix_z
1 1 0.28 0 0 0 0 0 0 1 1 1
2 1 0.28 4 0 0 0 0 4 0 0 0
3 0.1 0.18 5 0 0 0 0 6 0 0 0
1 27000 1.0 0 0 0 0 0 0 1 1 1
2 81 0.2 10 0 0 0 0 520 0 0 0
3 1 0.1 10.5 0 0 0 0 647 0 0 0
+2 -2
View File
@@ -23,7 +23,7 @@ force_calc: 0 # 强制重新计算:1=跳过缓存强算,0=自动使用
engine: python # 默认使用 Python 引擎
# ── 盒子 ──────────────────────────────────────
box_a: 20.0 # 立方体半边长,粒子被限制在 [-box_a, box_a]³ 内
box_a: 30.0 # 立方体半边长,粒子被限制在 [-box_a, box_a]³ 内
# ── 初始构型 ──────────────────────────────────
# 坐标文件格式:
@@ -74,7 +74,7 @@ T_total: 10.0
NSTEP: 2
# ── 时间步长 ──────────────────────────────────
DT: 0.001 # 时间步长 (s)
DT: 0.0001 # 时间步长 (s)
# 抽帧范围:只保存 [sample_start, sample_end) 区间内的帧
sample_start: null # null 表示从头开始(帧索引从 0 起)
+1 -1
View File
@@ -1,5 +1,5 @@
n mass radius x y z vx vy vz fix_x fix_y fix_z
1 1 0.1 0 0 1 0 0 0 0 1 1
1 1 0.1 0 0 0 0 0 0 0 1 1
2 1 0.1 1 0 0 0 0 0 0 1 1
3 1 0.1 2 0 0 0 0 0 0 1 1
4 1 0.1 3 0 0 0 0 0 0 1 1
+40
View File
@@ -0,0 +1,40 @@
# case06: 一维原子链横波模拟
60 个原子沿 x 轴排列,相邻原子用弹簧连接。原子 1 受 z 方向驱动力作用,产生沿链传播的横波。
## 物理设定
| 参数 | 值 |
|---|---|
| 原子数 | 120 |
| 排列 | 沿 x 轴等间距排列,间距为 1 |
| 约束 | 原子**沿 z 方向自由振动**fix_x=1, fix_y=1, fix_z=0),x, y 锁定 |
| 弹簧 | 劲度系数 k=1.0,原长 L₀=1.0 |
| 重力 | 无 |
| 万有引力 | 无 |
| 阻尼 | 无 |
| 驱动力 | 原子 1(z 方向驱动) |
| 算法 | leapfrog(蛙跳法,能量守恒) |
## 驱动力
原子 1 的位置由 `input/driver.txt` 中的驱动力公式决定:
```math
z(t) = A_z \cdot \cos(2\pi f_z t + \phi_z)
```
当前参数:A_z = 0.5, f_z = 0.1 Hz, φ_z = 90°, period = all(全程驱动)。
## 动力学行为
原子 1 沿 z 方向的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的**横波**。由于 z 方向的振动是横向的,弹簧大部分张力在 x 方向,z 方向的有效刚度是非线性的——等效于一个三次方恢复力(FPU 型非线性),因此波速较慢。
## 使用方法
```bash
cd examples/case06
python run_dynamics.py
```
配置参数详见 `input/input.txt`,驱动力定义见 `input/driver.txt`,完整文档见 `doc/index.html`
+477
View File
@@ -0,0 +1,477 @@
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>case06 — 一维原子链驱动力学模拟 | 物理原理 &amp; 使用文档</title>
<style>
:root {
--bg: #f8f9fa;
--card: #fff;
--text: #1a1a2e;
--accent: #2563eb;
--accent-light: #dbeafe;
--code-bg: #1e293b;
--code-text: #e2e8f0;
--border: #e2e8f0;
--muted: #64748b;
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Noto Sans SC", sans-serif;
background: var(--bg);
color: var(--text);
line-height: 1.7;
}
/* ── Header ── */
.hero {
background: linear-gradient(135deg, #1e293b 0%, #334155 100%);
color: #fff;
padding: 56px 24px 48px;
text-align: center;
}
.hero h1 { font-size: 2rem; font-weight: 700; letter-spacing: -0.02em; }
.hero .subtitle {
margin-top: 10px;
font-size: 1.05rem;
opacity: 0.8;
}
.hero .badge {
display: inline-block;
margin-top: 14px;
padding: 4px 14px;
border-radius: 999px;
background: rgba(255,255,255,0.12);
font-size: 0.82rem;
}
/* ── Layout ── */
.container { max-width: 820px; margin: 0 auto; padding: 32px 20px; }
section { margin-bottom: 44px; }
h2 {
font-size: 1.35rem;
font-weight: 600;
margin-bottom: 16px;
padding-bottom: 8px;
border-bottom: 2px solid var(--accent);
display: inline-block;
}
h3 {
font-size: 1.05rem;
font-weight: 600;
margin: 20px 0 10px;
}
p, li { margin-bottom: 10px; }
ul, ol { padding-left: 22px; }
strong { color: var(--accent); }
/* ── Cards ── */
.card {
background: var(--card);
border-radius: 12px;
padding: 20px 24px;
margin-bottom: 16px;
border: 1px solid var(--border);
box-shadow: 0 1px 3px rgba(0,0,0,0.04);
}
/* ── Formula / Code blocks ── */
.formula {
background: var(--card);
border-left: 4px solid var(--accent);
padding: 14px 20px;
margin: 14px 0;
font-family: "Times New Roman", "STIX", serif;
font-size: 1.05rem;
overflow-x: auto;
border-radius: 0 8px 8px 0;
}
code {
background: var(--accent-light);
padding: 2px 7px;
border-radius: 4px;
font-family: "JetBrains Mono", "Fira Code", monospace;
font-size: 0.88em;
}
pre {
background: var(--code-bg);
color: var(--code-text);
padding: 16px 20px;
border-radius: 10px;
overflow-x: auto;
font-size: 0.85rem;
line-height: 1.5;
margin: 14px 0;
}
pre .cm { color: #94a3b8; font-style: italic; } /* comment */
/* ── Table ── */
table {
width: 100%;
border-collapse: collapse;
margin: 14px 0;
font-size: 0.92rem;
}
th, td {
padding: 8px 12px;
text-align: left;
border-bottom: 1px solid var(--border);
}
th { background: var(--accent-light); font-weight: 600; }
/* ── TOC ── */
.toc { counter-reset: toc; }
.toc li { counter-increment: toc; list-style: none; margin-bottom: 6px; }
.toc li::before { content: counter(toc) ". "; font-weight: 600; color: var(--accent); }
.toc a { color: var(--accent); text-decoration: none; }
.toc a:hover { text-decoration: underline; }
/* ── Flow diagram ── */
.flow { display: flex; flex-wrap: wrap; gap: 8px; align-items: center; justify-content: center; margin: 16px 0; }
.flow-step {
background: var(--accent-light);
border: 1px solid var(--accent);
border-radius: 8px;
padding: 8px 16px;
font-size: 0.88rem;
font-weight: 500;
}
.flow-arrow { color: var(--muted); font-size: 1.2rem; }
@media (max-width: 600px) {
.hero h1 { font-size: 1.5rem; }
.flow { flex-direction: column; }
.flow-arrow { transform: rotate(90deg); }
}
</style>
</head>
<body>
<!-- ============================================================ -->
<!-- Header -->
<!-- ============================================================ -->
<header class="hero">
<h1>一维原子链驱动力学模拟</h1>
<p class="subtitle">120 个原子沿 x 轴排列 · 弹簧连接 · z 方向受迫振动</p>
<span class="badge">case06 · examples/case06</span>
</header>
<div class="container">
<!-- ============================================================ -->
<!-- TOC -->
<!-- ============================================================ -->
<section>
<h2>目录</h2>
<ol class="toc">
<li><a href="#physics">物理原理</a></li>
<li><a href="#algorithm">数值算法</a></li>
<li><a href="#driver">驱动力模型</a></li>
<li><a href="#usage">使用方法</a></li>
<li><a href="#params">参数参考</a></li>
<li><a href="#files">文件结构</a></li>
<li><a href="#troubleshoot">常见问题</a></li>
</ol>
</section>
<!-- ============================================================ -->
<!-- 1. Physics -->
<!-- ============================================================ -->
<section id="physics">
<h2>一、物理原理</h2>
<div class="card">
<h3>1.1 一维原子链</h3>
<p>120 个原子沿 <strong>x 轴</strong> 等间距排列,原子间距为 1。相邻原子之间用 <strong>理想弹簧</strong> 连接,弹簧的劲度系数 <em>k</em> = 1.0,原长 <em>L</em>₀ = 1.0(与原子间距一致,初始状态弹簧无拉伸)。</p>
<p>每个原子被限制在 <strong>z 方向</strong> 自由振动,x 和 y 方向锁定(<code>fix_x=1, fix_y=1, fix_z=0</code>)。</p>
</div>
<div class="card">
<h3>1.2 弹簧力(胡克定律)</h3>
<p>当原子 <em>i</em><em>j</em> 之间有弹簧连接时,原子 <em>i</em> 受到的弹簧力为:</p>
<div class="formula">
<strong>F</strong> = <em>k</em> · (<em>d</em> <em>L</em>₀) · <strong>u</strong><sub><em>ij</em></sub>
</div>
<p>其中 <em>d</em> = |<strong>r</strong><sub><em>j</em></sub> <strong>r</strong><sub><em>i</em></sub>| 为两原子间距离,<strong>u</strong><sub><em>ij</em></sub> 为从 <em>i</em> 指向 <em>j</em> 的单位向量。由于原子只在 z 方向振动,弹簧在 z 方向的分量是 <strong>几何非线性</strong> 的——对于小振幅近似,z 方向等效于一个三次方恢复力(FPU 型非线性)。</p>
</div>
<div class="card">
<h3>1.3 运动方程</h3>
<p>对于第 <em>i</em> 个自由原子(非受驱),牛顿第二定律给出:</p>
<div class="formula">
<em>m</em> · <strong>a</strong><sub><em>i</em></sub> = <strong>F</strong><sub><em>i</em></sub><sup>spring</sup> + <strong>F</strong><sub><em>i</em></sub><sup>driving</sup>
</div>
<p>本案例中 <strong>唯一的外力</strong> 来自驱动力(仅施加于原子 1)。无重力、无万有引力、无阻尼,系统总能量守恒。</p>
</div>
<div class="card">
<h3>1.4 波传播</h3>
<p>原子 1 的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的 <strong>横波</strong>。由于横向振动的几何非线性(弹簧大部分张力在 x 方向,z 方向的有效刚度远小于 1),波的传播速度较慢,且高阶频率成分会在链中产生复杂的非线性动力学行为(类似 FPU 回波现象)。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 2. Algorithm -->
<!-- ============================================================ -->
<section id="algorithm">
<h2>二、数值算法</h2>
<div class="card">
<h3>2.1 蛙跳法(Leapfrog / Velocity-Verlet</h3>
<p>采用能量守恒特性优异的 <strong>蛙跳法</strong>(二阶辛积分器),更新公式为:</p>
<div class="formula">
<strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) = <strong>v</strong>(<em>t</em>) + ½ <strong>a</strong>(<em>t</em>) · Δ<em>t</em><br>
<strong>r</strong>(<em>t</em> + Δ<em>t</em>) = <strong>r</strong>(<em>t</em>) + <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) · Δ<em>t</em><br>
<strong>a</strong>(<em>t</em> + Δ<em>t</em>) = <strong>F</strong>(<strong>r</strong>(<em>t</em> + Δ<em>t</em>), <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>)) / <em>m</em><br>
<strong>v</strong>(<em>t</em> + Δ<em>t</em>) = <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) + ½ <strong>a</strong>(<em>t</em> + Δ<em>t</em>) · Δ<em>t</em>
</div>
<p>蛙跳法在长时间模拟中能量漂移极小(本案例验证 <strong>&lt; 0.004%</strong>),适合无阻尼的保守系统。</p>
</div>
<div class="card">
<h3>2.2 时间步长与采样</h3>
<table>
<tr><th>参数</th><th></th><th>说明</th></tr>
<tr><td>DT</td><td>0.01 s</td><td>积分步长(远小于 1/ω ≈ 0.16 s,满足稳定性条件)</td></tr>
<tr><td>T_total</td><td>100 s</td><td>总模拟时间 → NT = 10000 步</td></tr>
<tr><td>NSTEP</td><td>50</td><td>每 NSTEP 步取一帧用于动画 → 200 帧</td></tr>
<tr><td>method</td><td>leapfrog</td><td>蛙跳法(Velocity-Verlet</td></tr>
</table>
</div>
<div class="card">
<h3>2.3 计算流程</h3>
<div class="flow">
<span class="flow-step">读入 coord.txt<br>connection.txt<br>bond.txt</span>
<span class="flow-arrow"></span>
<span class="flow-step">施加驱动力<br>(驱动原子 1</span>
<span class="flow-arrow"></span>
<span class="flow-step">记录轨迹</span>
<span class="flow-arrow"></span>
<span class="flow-step">蛙跳法<br>更新位置/速度</span>
<span class="flow-arrow"></span>
<span class="flow-step">固定约束<br>x, y 锁定)</span>
<span class="flow-arrow"></span>
<span class="flow-step" style="background:#fef3c7;border-color:#f59e0b;">循环<br>NT 次</span>
</div>
<p style="margin-top:12px;">注意:驱动力在 <strong>每次积分前</strong> 施加,确保受驱原子的位置正确传递给弹簧力计算。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 3. Driving Force -->
<!-- ============================================================ -->
<section id="driver">
<h2>三、驱动力模型</h2>
<div class="card">
<h3>3.1 定义文件</h3>
<p>驱动力由 <code>input/driver.txt</code> 定义,格式如下:</p>
<pre>n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0 0 5 0 0 1 0 0 90 all</pre>
</div>
<div class="card">
<h3>3.2 数学公式</h3>
<p>受驱原子的位置由下式决定(<strong>完全替换</strong> coord.txt 中的初始坐标和固定约束):</p>
<div class="formula">
<strong>r</strong>(<em>t</em>) = <strong>A</strong> · cos(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>速度由解析导数给出:</p>
<div class="formula">
<strong>v</strong>(<em>t</em>) = <strong>A</strong> · 2π<em>f</em> · sin(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>其中 <strong>A</strong> = (amp_x, amp_y, amp_z)<strong>f</strong> = (freq_x, freq_y, freq_z) 为不同方向的驱动频率,<strong>φ</strong> = (phi_x, phi_y, phi_z) 为相位(<strong>角度制</strong>,代码自动转换为弧度)。</p>
</div>
<div class="card">
<h3>3.3 本案例驱动参数</h3>
<table>
<tr><th>参数</th><th></th><th>含义</th></tr>
<tr><td>amp_z</td><td>5.0</td><td>z 方向驱动振幅</td></tr>
<tr><td>freq_z</td><td>1.0 Hz</td><td>驱动频率(周期 1 s</td></tr>
<tr><td>phi_z</td><td>90°</td><td>驱动相位 → z(0) = 5·cos(90°) = 0</td></tr>
<tr><td>period</td><td>all</td><td>全程驱动,永不停止</td></tr>
</table>
<div class="formula">
<em>z</em>(<em>t</em>) = 5.0 · cos(2π · 1.0 · <em>t</em> + 90°)
</div>
</div>
<div class="card">
<h3>3.4 有限周期驱动</h3>
<p><code>period</code> 参数支持三种模式:</p>
<ul>
<li><strong>all</strong> — 全程驱动</li>
<li><strong>数值</strong> — 驱动指定周期数后 <strong>静止</strong>(冻结在最终位置,速度归零)。例如 <code>period: 1</code> 表示驱动 1 个完整周期后停止。</li>
</ul>
</div>
<div class="card">
<h3>3.5 驱动与固定约束的关系</h3>
<p>对于受驱原子(<code>driver.txt</code><code>n</code> 指定的原子),其在 <code>coord.txt</code> 中的初始坐标和 <code>fix_x/fix_y/fix_z</code> 约束被 <strong>完全忽略</strong>。原子的位置和速度完全由驱动力公式决定。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 4. Usage -->
<!-- ============================================================ -->
<section id="usage">
<h2>四、使用方法</h2>
<div class="card">
<h3>4.1 完整运行(模拟 + 动画)</h3>
<pre>cd examples/case06
python run_dynamics.py</pre>
<p>这步会依次执行:物理模拟 → 抽帧 → 打开 VisPy 3D 动画窗口。</p>
</div>
<div class="card">
<h3>4.2 仅查看已有结果</h3>
<p>如果已经跑完模拟且生成了 <code>output/display.txt</code>,可以通过修改 <code>input.txt</code> 跳过计算,只开动画:</p>
<pre>step_simulate: 0 # 跳过模拟
step_sample: 0 # 跳过抽帧
step_animation: 1 # 播放动画</pre>
<p>然后运行:<code>python run_dynamics.py</code></p>
</div>
<div class="card">
<h3>4.3 手动 3D 动画</h3>
<p>也可以单独启动 VisPy 窗口:</p>
<pre>python ../../draw.py output/</pre>
</div>
<div class="card">
<h3>4.4 强制重新计算</h3>
<p>修改参数后需要重新运行模拟时,设置:</p>
<pre>force_calc: 1 # 忽略缓存,强制重新计算</pre>
</div>
<div class="card">
<h3>4.5 动画交互</h3>
<table>
<tr><th>操作</th><th>效果</th></tr>
<tr><td>鼠标拖动</td><td>旋转视角</td></tr>
<tr><td>滚轮</td><td>缩放</td></tr>
<tr><td>W / S 键</td><td>相机沿 Z 轴向前 / 向后移动(靠近/远离场景)</td></tr>
<tr><td>A / D 键</td><td>视角向右 / 向左平移</td></tr>
<tr><td>E / Q 键</td><td>视角上升 / 下降(屏幕方向)</td></tr>
<tr><td>C / X 键</td><td>增大 / 减小步长</td></tr>
<tr><td>V 键</td><td>切换透视 / 正交投影</td></tr>
<tr><td>左上角 <strong>reset</strong> 按钮</td><td>复位视角到初始位置</td></tr>
<tr><td>左上角 <strong>info</strong> 按钮</td><td>切换信息面板显示/隐藏</td></tr>
<tr><td>左上角 <strong>axes</strong> 按钮</td><td>切换坐标轴显示/隐藏</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 5. Parameters -->
<!-- ============================================================ -->
<section id="params">
<h2>五、参数参考</h2>
<div class="card">
<h3>5.1 input.txt 关键参数</h3>
<table>
<tr><th>参数</th><th>默认值</th><th>说明</th></tr>
<tr><td>gravity_field</td><td>0</td><td>均匀重力场(已关闭)</td></tr>
<tr><td>gravity_interaction</td><td>0</td><td>原子间万有引力(已关闭)</td></tr>
<tr><td>elastic_force</td><td>1</td><td>弹簧键力(已开启)</td></tr>
<tr><td>damping_force</td><td>0</td><td>阻尼(已关闭)</td></tr>
<tr><td><strong>driving_force</strong></td><td><strong>1</strong></td><td>驱动力开关(1=开启,需 driver.txt</td></tr>
<tr><td>method</td><td>leapfrog</td><td>数值积分方法</td></tr>
<tr><td>DT</td><td>0.01</td><td>积分步长 (s)</td></tr>
<tr><td>T_total</td><td>100.0</td><td>总模拟时间 (s)</td></tr>
<tr><td>NSTEP</td><td>50</td><td>抽帧步数间隔</td></tr>
<tr><td>engine</td><td>python</td><td>计算引擎(python / c / cpp / fortran</td></tr>
<tr><td>use_marker</td><td>1</td><td>渲染模式(0=Sphere 网格, 1=Marker GPU 实例化)</td></tr>
</table>
</div>
<div class="card">
<h3>5.2 流程控制参数</h3>
<table>
<tr><th>参数</th><th>0</th><th>1</th></tr>
<tr><td>step_simulate</td><td>跳过模拟(加载已有轨迹)</td><td>运行物理模拟</td></tr>
<tr><td>step_sample</td><td>跳过抽帧</td><td>从轨迹抽取显示帧</td></tr>
<tr><td>step_plot</td><td>不生成图表</td><td>生成轨迹/能量图</td></tr>
<tr><td><strong>step_plot_wave</strong></td><td>不生成波形图</td><td>生成波形能量动画 GIF</td></tr>
<tr><td>step_animation</td><td>不启动动画</td><td>自动打开 VisPy 3D 窗口</td></tr>
<tr><td>force_calc</td><td>自动检测缓存</td><td>强制重新计算</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 6. File Structure -->
<!-- ============================================================ -->
<section id="files">
<h2>六、文件结构</h2>
<pre>case06/
├── input/
│ ├── input.txt # 主配置文件(YAML 格式)
│ ├── coord.txt # 原子坐标(120 个原子)
│ ├── connection.txt # 弹簧连接关系(59 条键)
│ ├── bond.txt # 弹簧参数(k=1.0, L₀=1.0
│ └── <strong>driver.txt</strong> # <span class="cm">驱动力定义(本案例新增)</span>
├── output/
│ ├── trajectory.txt # 全量轨迹数据(50000 步 × 120 原子)
│ ├── display.txt # 抽帧后的动画数据(500 帧 × 120 原子)
│ ├── dynamics.log # 计算日志
│ ├── animation.log # 动画启动日志(闪退时排查用)
│ └── wave_animation.gif # 波形能量动画(step_plot_wave=1 时生成)
├── doc/
│ └── index.html # <span class="cm">本文档</span>
├── Readme.md # 案例简介
└── run_dynamics.py # 案例运行入口</pre>
</section>
<!-- ============================================================ -->
<!-- 7. Troubleshooting -->
<!-- ============================================================ -->
<section id="troubleshoot">
<h2>七、常见问题</h2>
<div class="card">
<h3>7.1 动画窗口闪退</h3>
<p>如果 VisPy 窗口一闪就消失,请检查:</p>
<ul>
<li><code>output/animation.log</code> 中是否有错误信息</li>
<li><code>output/display.txt</code> 是否存在(需先跑 <code>step_sample: 1</code></li>
</ul>
</div>
<div class="card">
<h3>7.2 原子不振动</h3>
<p>可能原因:</p>
<ul>
<li><strong>NSTEP 过大</strong>:抽帧间隔大于驱动周期的一半时,动画会丢失振动细节。建议 NSTEP ≤ 1/(freq · DT · 10)</li>
<li><strong>相位 φ 使采样点落在零值</strong>:试试 <code>phi_z: 0</code> 让原子在 t=0 处于振幅峰值</li>
<li>确认 <code>driving_force: 1</code><code>driver.txt</code> 中 amp_z 不为 0</li>
</ul>
</div>
<div class="card">
<h3>7.3 渲染性能慢</h3>
<p>原子数多时动画卡顿:</p>
<ul>
<li>设置 <code>use_marker: 1</code>(使用 GPU 实例化渲染替代独立网格球体)</li>
<li>增大 <code>NSTEP</code> 减少动画帧数</li>
</ul>
</div>
</section>
<hr style="border:none;border-top:1px solid var(--border);margin:40px 0;">
<footer style="text-align:center;color:var(--muted);font-size:0.85rem;margin-bottom:40px;">
Dynamics Simulation Framework &nbsp;·&nbsp; 生成于 2026-06-10
</footer>
</div>
</body>
</html>
+2
View File
@@ -0,0 +1,2 @@
bond_name k rest_length
h 100.0 1.0
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
+2
View File
@@ -0,0 +1,2 @@
n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
5101 0 0 2.0 0 0 0.05 90 90 90 all
+89
View File
@@ -0,0 +1,89 @@
# 物理模拟参数配置
# 格式:YAML
# 用法:python run_dynamics.py
# ── 流程控制 ──────────────────────────────────
step_simulate: 1 # 运行物理模拟
step_sample: 0 # 重新抽帧,默认0=不执行
step_plot: 0 # 绘制轨迹/能量图
step_animation: 1 # 自动播放 VisPy 3D 动画窗口
step_plot_wave: 0 # 绘制波形能量动画
force_calc: 1 # 强制重新计算
# ── 文件保存 ──────────────────────────────────
save_trajectory: 0 # 0=不保留完整轨迹文件
# ── 计算引擎 ──────────────────────────────────
engine: c
# ── 盒子 ──────────────────────────────────────
box_a: 120.0
# ── 初始构型 ──────────────────────────────────
coord_file: input/coord.txt
connection_file: input/connection.txt
bond_file: input/bond.txt
driver_file: input/driver.txt
# 绘图/动画展示的原子序号
plot_atom: 5101 # 中心原子 (0,0)
# ── 物理参数 ──────────────────────────────────
G: [0.000, 0.000, 0.000]
B: [0.000, 0.000, 0.000]
# ── 力开关 ────────────────────────────────────
gravity_field: 0
gravity_interaction: 0
elastic_force: 1
damping_force: 0
driving_force: 1
gravity_strength: 1.0
# ── 数值算法 ──────────────────────────────────
method: leapfrog
# ── 步骤控制 ──────────────────────────────────
warmup_steps: 0 # 受迫波动,无需预热
T_total: 100.0
NSTEP: 10
DT: 0.01
sample_start: null
sample_end: null
# ── 渲染方式 ──────────────────────────────────
use_marker: 1
# ── 位移着色 ──────────────────────────────────
display_color: {
x : [0, [255, 0, 0]],
y : [0, [ 0, 255, 0]],
z : [1, [ 0, 0, 255]],
xy : [0, [255, 255, 0]],
yz : [0, [ 0, 255, 255]],
zx : [0, [255, 0, 255]],
xyz : [0, [ 0, 0, 0]],
}
# ── 显示参数 ──────────────────────────────────
alpha: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
ball_color_r: 0.20
ball_color_g: 0.60
ball_color_b: 0.90
box_color_r: 0.80
box_color_g: 0.80
box_color_b: 0.85
# ── 摄像机 ────────────────────────────────────
camera_distance: 120.0
camera_elevation: 60.0
camera_azimuth: -45.0
camera_center_x: 0.0
camera_center_y: 0.0
camera_center_z: 0.0
move_camera: 0
display_amp: [1.0, 1.0, 1.0]
+2
View File
@@ -0,0 +1,2 @@
0 0 50
0 0 80
+54
View File
@@ -0,0 +1,54 @@
"""
Case runner for Dynamics case11 2D grid (61x61 atomic mesh).
This script keeps program and data separated:
- program: ../../dynamics.py
- input: ./input
- output: ./output
"""
from __future__ import annotations
import argparse
import importlib.util
from pathlib import Path
CASE_DIR = Path(__file__).resolve().parent
DYNAMICS_PATH = Path("..") / ".." / "dynamics.py"
INPUT_DIR = Path("input")
OUTPUT_DIR = Path("output")
CONFIG_FILE = INPUT_DIR / "input.txt"
def load_dynamics_module(module_path: Path):
spec = importlib.util.spec_from_file_location("dynamics_module", module_path)
if spec is None or spec.loader is None:
raise ImportError(f"无法加载 dynamics.py: {module_path}")
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
return module
def main():
parser = argparse.ArgumentParser(description="运行 Dynamics 示例案例 case11")
parser.add_argument("--no-plot", action="store_true", help="跳过 matplotlib 绘图")
args = parser.parse_args()
dynamics_path = (CASE_DIR / DYNAMICS_PATH).resolve()
input_dir = (CASE_DIR / INPUT_DIR).resolve()
output_dir = (CASE_DIR / OUTPUT_DIR).resolve()
config_path = (CASE_DIR / CONFIG_FILE).resolve()
module = load_dynamics_module(dynamics_path)
module.run_case(
config_path=config_path,
runtime_base=CASE_DIR,
input_dir=input_dir,
output_dir=output_dir,
no_plot=args.no_plot,
)
if __name__ == "__main__":
main()
+40
View File
@@ -0,0 +1,40 @@
# case06: 一维原子链横波模拟
60 个原子沿 x 轴排列,相邻原子用弹簧连接。原子 1 受 z 方向驱动力作用,产生沿链传播的横波。
## 物理设定
| 参数 | 值 |
|---|---|
| 原子数 | 120 |
| 排列 | 沿 x 轴等间距排列,间距为 1 |
| 约束 | 原子**沿 z 方向自由振动**fix_x=1, fix_y=1, fix_z=0),x, y 锁定 |
| 弹簧 | 劲度系数 k=1.0,原长 L₀=1.0 |
| 重力 | 无 |
| 万有引力 | 无 |
| 阻尼 | 无 |
| 驱动力 | 原子 1(z 方向驱动) |
| 算法 | leapfrog(蛙跳法,能量守恒) |
## 驱动力
原子 1 的位置由 `input/driver.txt` 中的驱动力公式决定:
```math
z(t) = A_z \cdot \cos(2\pi f_z t + \phi_z)
```
当前参数:A_z = 0.5, f_z = 0.1 Hz, φ_z = 90°, period = all(全程驱动)。
## 动力学行为
原子 1 沿 z 方向的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的**横波**。由于 z 方向的振动是横向的,弹簧大部分张力在 x 方向,z 方向的有效刚度是非线性的——等效于一个三次方恢复力(FPU 型非线性),因此波速较慢。
## 使用方法
```bash
cd examples/case06
python run_dynamics.py
```
配置参数详见 `input/input.txt`,驱动力定义见 `input/driver.txt`,完整文档见 `doc/index.html`
+477
View File
@@ -0,0 +1,477 @@
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>case06 — 一维原子链驱动力学模拟 | 物理原理 &amp; 使用文档</title>
<style>
:root {
--bg: #f8f9fa;
--card: #fff;
--text: #1a1a2e;
--accent: #2563eb;
--accent-light: #dbeafe;
--code-bg: #1e293b;
--code-text: #e2e8f0;
--border: #e2e8f0;
--muted: #64748b;
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Noto Sans SC", sans-serif;
background: var(--bg);
color: var(--text);
line-height: 1.7;
}
/* ── Header ── */
.hero {
background: linear-gradient(135deg, #1e293b 0%, #334155 100%);
color: #fff;
padding: 56px 24px 48px;
text-align: center;
}
.hero h1 { font-size: 2rem; font-weight: 700; letter-spacing: -0.02em; }
.hero .subtitle {
margin-top: 10px;
font-size: 1.05rem;
opacity: 0.8;
}
.hero .badge {
display: inline-block;
margin-top: 14px;
padding: 4px 14px;
border-radius: 999px;
background: rgba(255,255,255,0.12);
font-size: 0.82rem;
}
/* ── Layout ── */
.container { max-width: 820px; margin: 0 auto; padding: 32px 20px; }
section { margin-bottom: 44px; }
h2 {
font-size: 1.35rem;
font-weight: 600;
margin-bottom: 16px;
padding-bottom: 8px;
border-bottom: 2px solid var(--accent);
display: inline-block;
}
h3 {
font-size: 1.05rem;
font-weight: 600;
margin: 20px 0 10px;
}
p, li { margin-bottom: 10px; }
ul, ol { padding-left: 22px; }
strong { color: var(--accent); }
/* ── Cards ── */
.card {
background: var(--card);
border-radius: 12px;
padding: 20px 24px;
margin-bottom: 16px;
border: 1px solid var(--border);
box-shadow: 0 1px 3px rgba(0,0,0,0.04);
}
/* ── Formula / Code blocks ── */
.formula {
background: var(--card);
border-left: 4px solid var(--accent);
padding: 14px 20px;
margin: 14px 0;
font-family: "Times New Roman", "STIX", serif;
font-size: 1.05rem;
overflow-x: auto;
border-radius: 0 8px 8px 0;
}
code {
background: var(--accent-light);
padding: 2px 7px;
border-radius: 4px;
font-family: "JetBrains Mono", "Fira Code", monospace;
font-size: 0.88em;
}
pre {
background: var(--code-bg);
color: var(--code-text);
padding: 16px 20px;
border-radius: 10px;
overflow-x: auto;
font-size: 0.85rem;
line-height: 1.5;
margin: 14px 0;
}
pre .cm { color: #94a3b8; font-style: italic; } /* comment */
/* ── Table ── */
table {
width: 100%;
border-collapse: collapse;
margin: 14px 0;
font-size: 0.92rem;
}
th, td {
padding: 8px 12px;
text-align: left;
border-bottom: 1px solid var(--border);
}
th { background: var(--accent-light); font-weight: 600; }
/* ── TOC ── */
.toc { counter-reset: toc; }
.toc li { counter-increment: toc; list-style: none; margin-bottom: 6px; }
.toc li::before { content: counter(toc) ". "; font-weight: 600; color: var(--accent); }
.toc a { color: var(--accent); text-decoration: none; }
.toc a:hover { text-decoration: underline; }
/* ── Flow diagram ── */
.flow { display: flex; flex-wrap: wrap; gap: 8px; align-items: center; justify-content: center; margin: 16px 0; }
.flow-step {
background: var(--accent-light);
border: 1px solid var(--accent);
border-radius: 8px;
padding: 8px 16px;
font-size: 0.88rem;
font-weight: 500;
}
.flow-arrow { color: var(--muted); font-size: 1.2rem; }
@media (max-width: 600px) {
.hero h1 { font-size: 1.5rem; }
.flow { flex-direction: column; }
.flow-arrow { transform: rotate(90deg); }
}
</style>
</head>
<body>
<!-- ============================================================ -->
<!-- Header -->
<!-- ============================================================ -->
<header class="hero">
<h1>一维原子链驱动力学模拟</h1>
<p class="subtitle">120 个原子沿 x 轴排列 · 弹簧连接 · z 方向受迫振动</p>
<span class="badge">case06 · examples/case06</span>
</header>
<div class="container">
<!-- ============================================================ -->
<!-- TOC -->
<!-- ============================================================ -->
<section>
<h2>目录</h2>
<ol class="toc">
<li><a href="#physics">物理原理</a></li>
<li><a href="#algorithm">数值算法</a></li>
<li><a href="#driver">驱动力模型</a></li>
<li><a href="#usage">使用方法</a></li>
<li><a href="#params">参数参考</a></li>
<li><a href="#files">文件结构</a></li>
<li><a href="#troubleshoot">常见问题</a></li>
</ol>
</section>
<!-- ============================================================ -->
<!-- 1. Physics -->
<!-- ============================================================ -->
<section id="physics">
<h2>一、物理原理</h2>
<div class="card">
<h3>1.1 一维原子链</h3>
<p>120 个原子沿 <strong>x 轴</strong> 等间距排列,原子间距为 1。相邻原子之间用 <strong>理想弹簧</strong> 连接,弹簧的劲度系数 <em>k</em> = 1.0,原长 <em>L</em>₀ = 1.0(与原子间距一致,初始状态弹簧无拉伸)。</p>
<p>每个原子被限制在 <strong>z 方向</strong> 自由振动,x 和 y 方向锁定(<code>fix_x=1, fix_y=1, fix_z=0</code>)。</p>
</div>
<div class="card">
<h3>1.2 弹簧力(胡克定律)</h3>
<p>当原子 <em>i</em><em>j</em> 之间有弹簧连接时,原子 <em>i</em> 受到的弹簧力为:</p>
<div class="formula">
<strong>F</strong> = <em>k</em> · (<em>d</em> <em>L</em>₀) · <strong>u</strong><sub><em>ij</em></sub>
</div>
<p>其中 <em>d</em> = |<strong>r</strong><sub><em>j</em></sub> <strong>r</strong><sub><em>i</em></sub>| 为两原子间距离,<strong>u</strong><sub><em>ij</em></sub> 为从 <em>i</em> 指向 <em>j</em> 的单位向量。由于原子只在 z 方向振动,弹簧在 z 方向的分量是 <strong>几何非线性</strong> 的——对于小振幅近似,z 方向等效于一个三次方恢复力(FPU 型非线性)。</p>
</div>
<div class="card">
<h3>1.3 运动方程</h3>
<p>对于第 <em>i</em> 个自由原子(非受驱),牛顿第二定律给出:</p>
<div class="formula">
<em>m</em> · <strong>a</strong><sub><em>i</em></sub> = <strong>F</strong><sub><em>i</em></sub><sup>spring</sup> + <strong>F</strong><sub><em>i</em></sub><sup>driving</sup>
</div>
<p>本案例中 <strong>唯一的外力</strong> 来自驱动力(仅施加于原子 1)。无重力、无万有引力、无阻尼,系统总能量守恒。</p>
</div>
<div class="card">
<h3>1.4 波传播</h3>
<p>原子 1 的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的 <strong>横波</strong>。由于横向振动的几何非线性(弹簧大部分张力在 x 方向,z 方向的有效刚度远小于 1),波的传播速度较慢,且高阶频率成分会在链中产生复杂的非线性动力学行为(类似 FPU 回波现象)。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 2. Algorithm -->
<!-- ============================================================ -->
<section id="algorithm">
<h2>二、数值算法</h2>
<div class="card">
<h3>2.1 蛙跳法(Leapfrog / Velocity-Verlet</h3>
<p>采用能量守恒特性优异的 <strong>蛙跳法</strong>(二阶辛积分器),更新公式为:</p>
<div class="formula">
<strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) = <strong>v</strong>(<em>t</em>) + ½ <strong>a</strong>(<em>t</em>) · Δ<em>t</em><br>
<strong>r</strong>(<em>t</em> + Δ<em>t</em>) = <strong>r</strong>(<em>t</em>) + <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) · Δ<em>t</em><br>
<strong>a</strong>(<em>t</em> + Δ<em>t</em>) = <strong>F</strong>(<strong>r</strong>(<em>t</em> + Δ<em>t</em>), <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>)) / <em>m</em><br>
<strong>v</strong>(<em>t</em> + Δ<em>t</em>) = <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) + ½ <strong>a</strong>(<em>t</em> + Δ<em>t</em>) · Δ<em>t</em>
</div>
<p>蛙跳法在长时间模拟中能量漂移极小(本案例验证 <strong>&lt; 0.004%</strong>),适合无阻尼的保守系统。</p>
</div>
<div class="card">
<h3>2.2 时间步长与采样</h3>
<table>
<tr><th>参数</th><th></th><th>说明</th></tr>
<tr><td>DT</td><td>0.01 s</td><td>积分步长(远小于 1/ω ≈ 0.16 s,满足稳定性条件)</td></tr>
<tr><td>T_total</td><td>100 s</td><td>总模拟时间 → NT = 10000 步</td></tr>
<tr><td>NSTEP</td><td>50</td><td>每 NSTEP 步取一帧用于动画 → 200 帧</td></tr>
<tr><td>method</td><td>leapfrog</td><td>蛙跳法(Velocity-Verlet</td></tr>
</table>
</div>
<div class="card">
<h3>2.3 计算流程</h3>
<div class="flow">
<span class="flow-step">读入 coord.txt<br>connection.txt<br>bond.txt</span>
<span class="flow-arrow"></span>
<span class="flow-step">施加驱动力<br>(驱动原子 1</span>
<span class="flow-arrow"></span>
<span class="flow-step">记录轨迹</span>
<span class="flow-arrow"></span>
<span class="flow-step">蛙跳法<br>更新位置/速度</span>
<span class="flow-arrow"></span>
<span class="flow-step">固定约束<br>x, y 锁定)</span>
<span class="flow-arrow"></span>
<span class="flow-step" style="background:#fef3c7;border-color:#f59e0b;">循环<br>NT 次</span>
</div>
<p style="margin-top:12px;">注意:驱动力在 <strong>每次积分前</strong> 施加,确保受驱原子的位置正确传递给弹簧力计算。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 3. Driving Force -->
<!-- ============================================================ -->
<section id="driver">
<h2>三、驱动力模型</h2>
<div class="card">
<h3>3.1 定义文件</h3>
<p>驱动力由 <code>input/driver.txt</code> 定义,格式如下:</p>
<pre>n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0 0 5 0 0 1 0 0 90 all</pre>
</div>
<div class="card">
<h3>3.2 数学公式</h3>
<p>受驱原子的位置由下式决定(<strong>完全替换</strong> coord.txt 中的初始坐标和固定约束):</p>
<div class="formula">
<strong>r</strong>(<em>t</em>) = <strong>A</strong> · cos(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>速度由解析导数给出:</p>
<div class="formula">
<strong>v</strong>(<em>t</em>) = <strong>A</strong> · 2π<em>f</em> · sin(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>其中 <strong>A</strong> = (amp_x, amp_y, amp_z)<strong>f</strong> = (freq_x, freq_y, freq_z) 为不同方向的驱动频率,<strong>φ</strong> = (phi_x, phi_y, phi_z) 为相位(<strong>角度制</strong>,代码自动转换为弧度)。</p>
</div>
<div class="card">
<h3>3.3 本案例驱动参数</h3>
<table>
<tr><th>参数</th><th></th><th>含义</th></tr>
<tr><td>amp_z</td><td>5.0</td><td>z 方向驱动振幅</td></tr>
<tr><td>freq_z</td><td>1.0 Hz</td><td>驱动频率(周期 1 s</td></tr>
<tr><td>phi_z</td><td>90°</td><td>驱动相位 → z(0) = 5·cos(90°) = 0</td></tr>
<tr><td>period</td><td>all</td><td>全程驱动,永不停止</td></tr>
</table>
<div class="formula">
<em>z</em>(<em>t</em>) = 5.0 · cos(2π · 1.0 · <em>t</em> + 90°)
</div>
</div>
<div class="card">
<h3>3.4 有限周期驱动</h3>
<p><code>period</code> 参数支持三种模式:</p>
<ul>
<li><strong>all</strong> — 全程驱动</li>
<li><strong>数值</strong> — 驱动指定周期数后 <strong>静止</strong>(冻结在最终位置,速度归零)。例如 <code>period: 1</code> 表示驱动 1 个完整周期后停止。</li>
</ul>
</div>
<div class="card">
<h3>3.5 驱动与固定约束的关系</h3>
<p>对于受驱原子(<code>driver.txt</code><code>n</code> 指定的原子),其在 <code>coord.txt</code> 中的初始坐标和 <code>fix_x/fix_y/fix_z</code> 约束被 <strong>完全忽略</strong>。原子的位置和速度完全由驱动力公式决定。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 4. Usage -->
<!-- ============================================================ -->
<section id="usage">
<h2>四、使用方法</h2>
<div class="card">
<h3>4.1 完整运行(模拟 + 动画)</h3>
<pre>cd examples/case06
python run_dynamics.py</pre>
<p>这步会依次执行:物理模拟 → 抽帧 → 打开 VisPy 3D 动画窗口。</p>
</div>
<div class="card">
<h3>4.2 仅查看已有结果</h3>
<p>如果已经跑完模拟且生成了 <code>output/display.txt</code>,可以通过修改 <code>input.txt</code> 跳过计算,只开动画:</p>
<pre>step_simulate: 0 # 跳过模拟
step_sample: 0 # 跳过抽帧
step_animation: 1 # 播放动画</pre>
<p>然后运行:<code>python run_dynamics.py</code></p>
</div>
<div class="card">
<h3>4.3 手动 3D 动画</h3>
<p>也可以单独启动 VisPy 窗口:</p>
<pre>python ../../draw.py output/</pre>
</div>
<div class="card">
<h3>4.4 强制重新计算</h3>
<p>修改参数后需要重新运行模拟时,设置:</p>
<pre>force_calc: 1 # 忽略缓存,强制重新计算</pre>
</div>
<div class="card">
<h3>4.5 动画交互</h3>
<table>
<tr><th>操作</th><th>效果</th></tr>
<tr><td>鼠标拖动</td><td>旋转视角</td></tr>
<tr><td>滚轮</td><td>缩放</td></tr>
<tr><td>W / S 键</td><td>相机沿 Z 轴向前 / 向后移动(靠近/远离场景)</td></tr>
<tr><td>A / D 键</td><td>视角向右 / 向左平移</td></tr>
<tr><td>E / Q 键</td><td>视角上升 / 下降(屏幕方向)</td></tr>
<tr><td>C / X 键</td><td>增大 / 减小步长</td></tr>
<tr><td>V 键</td><td>切换透视 / 正交投影</td></tr>
<tr><td>左上角 <strong>reset</strong> 按钮</td><td>复位视角到初始位置</td></tr>
<tr><td>左上角 <strong>info</strong> 按钮</td><td>切换信息面板显示/隐藏</td></tr>
<tr><td>左上角 <strong>axes</strong> 按钮</td><td>切换坐标轴显示/隐藏</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 5. Parameters -->
<!-- ============================================================ -->
<section id="params">
<h2>五、参数参考</h2>
<div class="card">
<h3>5.1 input.txt 关键参数</h3>
<table>
<tr><th>参数</th><th>默认值</th><th>说明</th></tr>
<tr><td>gravity_field</td><td>0</td><td>均匀重力场(已关闭)</td></tr>
<tr><td>gravity_interaction</td><td>0</td><td>原子间万有引力(已关闭)</td></tr>
<tr><td>elastic_force</td><td>1</td><td>弹簧键力(已开启)</td></tr>
<tr><td>damping_force</td><td>0</td><td>阻尼(已关闭)</td></tr>
<tr><td><strong>driving_force</strong></td><td><strong>1</strong></td><td>驱动力开关(1=开启,需 driver.txt</td></tr>
<tr><td>method</td><td>leapfrog</td><td>数值积分方法</td></tr>
<tr><td>DT</td><td>0.01</td><td>积分步长 (s)</td></tr>
<tr><td>T_total</td><td>100.0</td><td>总模拟时间 (s)</td></tr>
<tr><td>NSTEP</td><td>50</td><td>抽帧步数间隔</td></tr>
<tr><td>engine</td><td>python</td><td>计算引擎(python / c / cpp / fortran</td></tr>
<tr><td>use_marker</td><td>1</td><td>渲染模式(0=Sphere 网格, 1=Marker GPU 实例化)</td></tr>
</table>
</div>
<div class="card">
<h3>5.2 流程控制参数</h3>
<table>
<tr><th>参数</th><th>0</th><th>1</th></tr>
<tr><td>step_simulate</td><td>跳过模拟(加载已有轨迹)</td><td>运行物理模拟</td></tr>
<tr><td>step_sample</td><td>跳过抽帧</td><td>从轨迹抽取显示帧</td></tr>
<tr><td>step_plot</td><td>不生成图表</td><td>生成轨迹/能量图</td></tr>
<tr><td><strong>step_plot_wave</strong></td><td>不生成波形图</td><td>生成波形能量动画 GIF</td></tr>
<tr><td>step_animation</td><td>不启动动画</td><td>自动打开 VisPy 3D 窗口</td></tr>
<tr><td>force_calc</td><td>自动检测缓存</td><td>强制重新计算</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 6. File Structure -->
<!-- ============================================================ -->
<section id="files">
<h2>六、文件结构</h2>
<pre>case06/
├── input/
│ ├── input.txt # 主配置文件(YAML 格式)
│ ├── coord.txt # 原子坐标(120 个原子)
│ ├── connection.txt # 弹簧连接关系(59 条键)
│ ├── bond.txt # 弹簧参数(k=1.0, L₀=1.0
│ └── <strong>driver.txt</strong> # <span class="cm">驱动力定义(本案例新增)</span>
├── output/
│ ├── trajectory.txt # 全量轨迹数据(50000 步 × 120 原子)
│ ├── display.txt # 抽帧后的动画数据(500 帧 × 120 原子)
│ ├── dynamics.log # 计算日志
│ ├── animation.log # 动画启动日志(闪退时排查用)
│ └── wave_animation.gif # 波形能量动画(step_plot_wave=1 时生成)
├── doc/
│ └── index.html # <span class="cm">本文档</span>
├── Readme.md # 案例简介
└── run_dynamics.py # 案例运行入口</pre>
</section>
<!-- ============================================================ -->
<!-- 7. Troubleshooting -->
<!-- ============================================================ -->
<section id="troubleshoot">
<h2>七、常见问题</h2>
<div class="card">
<h3>7.1 动画窗口闪退</h3>
<p>如果 VisPy 窗口一闪就消失,请检查:</p>
<ul>
<li><code>output/animation.log</code> 中是否有错误信息</li>
<li><code>output/display.txt</code> 是否存在(需先跑 <code>step_sample: 1</code></li>
</ul>
</div>
<div class="card">
<h3>7.2 原子不振动</h3>
<p>可能原因:</p>
<ul>
<li><strong>NSTEP 过大</strong>:抽帧间隔大于驱动周期的一半时,动画会丢失振动细节。建议 NSTEP ≤ 1/(freq · DT · 10)</li>
<li><strong>相位 φ 使采样点落在零值</strong>:试试 <code>phi_z: 0</code> 让原子在 t=0 处于振幅峰值</li>
<li>确认 <code>driving_force: 1</code><code>driver.txt</code> 中 amp_z 不为 0</li>
</ul>
</div>
<div class="card">
<h3>7.3 渲染性能慢</h3>
<p>原子数多时动画卡顿:</p>
<ul>
<li>设置 <code>use_marker: 1</code>(使用 GPU 实例化渲染替代独立网格球体)</li>
<li>增大 <code>NSTEP</code> 减少动画帧数</li>
</ul>
</div>
</section>
<hr style="border:none;border-top:1px solid var(--border);margin:40px 0;">
<footer style="text-align:center;color:var(--muted);font-size:0.85rem;margin-bottom:40px;">
Dynamics Simulation Framework &nbsp;·&nbsp; 生成于 2026-06-10
</footer>
</div>
</body>
</html>
+2
View File
@@ -0,0 +1,2 @@
bond_name k rest_length
h 100.0 1.0
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
+3
View File
@@ -0,0 +1,3 @@
n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
3081 0 0 2.0 0 0 0.05 90 90 90 all
7121 0 0 2.0 0 0 0.05 90 90 90 all
+83
View File
@@ -0,0 +1,83 @@
# 物理模拟参数配置
# case12 — 二维网格两点干涉(双点源 z 方向驱动)
# 驱动点: (0,-10) 和 (0,10),波从两点向外传播,在中心区域干涉
# ── 流程控制 ──────────────────────────────────
step_simulate: 1 # 运行物理模拟
step_sample: 0 # 重新抽帧,默认0=不执行
step_plot: 0 # 绘制轨迹/能量图
step_animation: 1 # 自动播放 VisPy 3D 动画窗口
step_plot_wave: 0 # 绘制波形能量动画
force_calc: 1 # 强制重新计算
# ── 文件保存 ──────────────────────────────────
save_trajectory: 0
# ── 计算引擎 ──────────────────────────────────
engine: c
# ── 盒子 ──────────────────────────────────────
box_a: 120.0
# ── 初始构型 ──────────────────────────────────
coord_file: input/coord.txt
connection_file: input/connection.txt
bond_file: input/bond.txt
driver_file: input/driver.txt
plot_atom: 5101 # 中心区域用于信息显示
# ── 物理参数 ──────────────────────────────────
G: [0.000, 0.000, 0.000]
B: [0.000, 0.000, 0.000]
gravity_field: 0
gravity_interaction: 0
elastic_force: 1
damping_force: 0
driving_force: 1
gravity_strength: 1.0
method: leapfrog
# ── 步骤控制 ──────────────────────────────────
warmup_steps: 0
T_total: 100.0
NSTEP: 500
DT: 0.001
sample_start: null
sample_end: null
# ── 渲染/着色 ─────────────────────────────────
use_marker: 1
display_color: {
x : [0, [255, 0, 0]],
y : [0, [ 0, 255, 0]],
z : [0, [ 0, 0, 255]],
xy : [0, [255, 255, 0]],
yz : [0, [ 0, 255, 255]],
zx : [0, [255, 0, 255]],
xyz : [1, [255, 255, 255]],
}
# ── 显示参数 ──────────────────────────────────
alpha: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
ball_color_r: 0.20
ball_color_g: 0.60
ball_color_b: 0.90
box_color_r: 0.80
box_color_g: 0.80
box_color_b: 0.85
# ── 摄像机 ────────────────────────────────────
camera_distance: 120.0
camera_elevation: 60.0
camera_azimuth: -45.0
camera_center_x: 0.0
camera_center_y: 0.0
camera_center_z: 0.0
move_camera: 0
display_amp: [1.0, 1.0, 1.0]
+2
View File
@@ -0,0 +1,2 @@
0 0 50
0 0 80
+54
View File
@@ -0,0 +1,54 @@
"""
Case runner for Dynamics case12 2D grid dual source interference.
This script keeps program and data separated:
- program: ../../dynamics.py
- input: ./input
- output: ./output
"""
from __future__ import annotations
import argparse
import importlib.util
from pathlib import Path
CASE_DIR = Path(__file__).resolve().parent
DYNAMICS_PATH = Path("..") / ".." / "dynamics.py"
INPUT_DIR = Path("input")
OUTPUT_DIR = Path("output")
CONFIG_FILE = INPUT_DIR / "input.txt"
def load_dynamics_module(module_path: Path):
spec = importlib.util.spec_from_file_location("dynamics_module", module_path)
if spec is None or spec.loader is None:
raise ImportError(f"无法加载 dynamics.py: {module_path}")
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
return module
def main():
parser = argparse.ArgumentParser(description="运行 Dynamics 示例案例 case11")
parser.add_argument("--no-plot", action="store_true", help="跳过 matplotlib 绘图")
args = parser.parse_args()
dynamics_path = (CASE_DIR / DYNAMICS_PATH).resolve()
input_dir = (CASE_DIR / INPUT_DIR).resolve()
output_dir = (CASE_DIR / OUTPUT_DIR).resolve()
config_path = (CASE_DIR / CONFIG_FILE).resolve()
module = load_dynamics_module(dynamics_path)
module.run_case(
config_path=config_path,
runtime_base=CASE_DIR,
input_dir=input_dir,
output_dir=output_dir,
no_plot=args.no_plot,
)
if __name__ == "__main__":
main()
+40
View File
@@ -0,0 +1,40 @@
# case06: 一维原子链横波模拟
60 个原子沿 x 轴排列,相邻原子用弹簧连接。原子 1 受 z 方向驱动力作用,产生沿链传播的横波。
## 物理设定
| 参数 | 值 |
|---|---|
| 原子数 | 120 |
| 排列 | 沿 x 轴等间距排列,间距为 1 |
| 约束 | 原子**沿 z 方向自由振动**fix_x=1, fix_y=1, fix_z=0),x, y 锁定 |
| 弹簧 | 劲度系数 k=1.0,原长 L₀=1.0 |
| 重力 | 无 |
| 万有引力 | 无 |
| 阻尼 | 无 |
| 驱动力 | 原子 1(z 方向驱动) |
| 算法 | leapfrog(蛙跳法,能量守恒) |
## 驱动力
原子 1 的位置由 `input/driver.txt` 中的驱动力公式决定:
```math
z(t) = A_z \cdot \cos(2\pi f_z t + \phi_z)
```
当前参数:A_z = 0.5, f_z = 0.1 Hz, φ_z = 90°, period = all(全程驱动)。
## 动力学行为
原子 1 沿 z 方向的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的**横波**。由于 z 方向的振动是横向的,弹簧大部分张力在 x 方向,z 方向的有效刚度是非线性的——等效于一个三次方恢复力(FPU 型非线性),因此波速较慢。
## 使用方法
```bash
cd examples/case06
python run_dynamics.py
```
配置参数详见 `input/input.txt`,驱动力定义见 `input/driver.txt`,完整文档见 `doc/index.html`
+477
View File
@@ -0,0 +1,477 @@
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>case06 — 一维原子链驱动力学模拟 | 物理原理 &amp; 使用文档</title>
<style>
:root {
--bg: #f8f9fa;
--card: #fff;
--text: #1a1a2e;
--accent: #2563eb;
--accent-light: #dbeafe;
--code-bg: #1e293b;
--code-text: #e2e8f0;
--border: #e2e8f0;
--muted: #64748b;
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Noto Sans SC", sans-serif;
background: var(--bg);
color: var(--text);
line-height: 1.7;
}
/* ── Header ── */
.hero {
background: linear-gradient(135deg, #1e293b 0%, #334155 100%);
color: #fff;
padding: 56px 24px 48px;
text-align: center;
}
.hero h1 { font-size: 2rem; font-weight: 700; letter-spacing: -0.02em; }
.hero .subtitle {
margin-top: 10px;
font-size: 1.05rem;
opacity: 0.8;
}
.hero .badge {
display: inline-block;
margin-top: 14px;
padding: 4px 14px;
border-radius: 999px;
background: rgba(255,255,255,0.12);
font-size: 0.82rem;
}
/* ── Layout ── */
.container { max-width: 820px; margin: 0 auto; padding: 32px 20px; }
section { margin-bottom: 44px; }
h2 {
font-size: 1.35rem;
font-weight: 600;
margin-bottom: 16px;
padding-bottom: 8px;
border-bottom: 2px solid var(--accent);
display: inline-block;
}
h3 {
font-size: 1.05rem;
font-weight: 600;
margin: 20px 0 10px;
}
p, li { margin-bottom: 10px; }
ul, ol { padding-left: 22px; }
strong { color: var(--accent); }
/* ── Cards ── */
.card {
background: var(--card);
border-radius: 12px;
padding: 20px 24px;
margin-bottom: 16px;
border: 1px solid var(--border);
box-shadow: 0 1px 3px rgba(0,0,0,0.04);
}
/* ── Formula / Code blocks ── */
.formula {
background: var(--card);
border-left: 4px solid var(--accent);
padding: 14px 20px;
margin: 14px 0;
font-family: "Times New Roman", "STIX", serif;
font-size: 1.05rem;
overflow-x: auto;
border-radius: 0 8px 8px 0;
}
code {
background: var(--accent-light);
padding: 2px 7px;
border-radius: 4px;
font-family: "JetBrains Mono", "Fira Code", monospace;
font-size: 0.88em;
}
pre {
background: var(--code-bg);
color: var(--code-text);
padding: 16px 20px;
border-radius: 10px;
overflow-x: auto;
font-size: 0.85rem;
line-height: 1.5;
margin: 14px 0;
}
pre .cm { color: #94a3b8; font-style: italic; } /* comment */
/* ── Table ── */
table {
width: 100%;
border-collapse: collapse;
margin: 14px 0;
font-size: 0.92rem;
}
th, td {
padding: 8px 12px;
text-align: left;
border-bottom: 1px solid var(--border);
}
th { background: var(--accent-light); font-weight: 600; }
/* ── TOC ── */
.toc { counter-reset: toc; }
.toc li { counter-increment: toc; list-style: none; margin-bottom: 6px; }
.toc li::before { content: counter(toc) ". "; font-weight: 600; color: var(--accent); }
.toc a { color: var(--accent); text-decoration: none; }
.toc a:hover { text-decoration: underline; }
/* ── Flow diagram ── */
.flow { display: flex; flex-wrap: wrap; gap: 8px; align-items: center; justify-content: center; margin: 16px 0; }
.flow-step {
background: var(--accent-light);
border: 1px solid var(--accent);
border-radius: 8px;
padding: 8px 16px;
font-size: 0.88rem;
font-weight: 500;
}
.flow-arrow { color: var(--muted); font-size: 1.2rem; }
@media (max-width: 600px) {
.hero h1 { font-size: 1.5rem; }
.flow { flex-direction: column; }
.flow-arrow { transform: rotate(90deg); }
}
</style>
</head>
<body>
<!-- ============================================================ -->
<!-- Header -->
<!-- ============================================================ -->
<header class="hero">
<h1>一维原子链驱动力学模拟</h1>
<p class="subtitle">120 个原子沿 x 轴排列 · 弹簧连接 · z 方向受迫振动</p>
<span class="badge">case06 · examples/case06</span>
</header>
<div class="container">
<!-- ============================================================ -->
<!-- TOC -->
<!-- ============================================================ -->
<section>
<h2>目录</h2>
<ol class="toc">
<li><a href="#physics">物理原理</a></li>
<li><a href="#algorithm">数值算法</a></li>
<li><a href="#driver">驱动力模型</a></li>
<li><a href="#usage">使用方法</a></li>
<li><a href="#params">参数参考</a></li>
<li><a href="#files">文件结构</a></li>
<li><a href="#troubleshoot">常见问题</a></li>
</ol>
</section>
<!-- ============================================================ -->
<!-- 1. Physics -->
<!-- ============================================================ -->
<section id="physics">
<h2>一、物理原理</h2>
<div class="card">
<h3>1.1 一维原子链</h3>
<p>120 个原子沿 <strong>x 轴</strong> 等间距排列,原子间距为 1。相邻原子之间用 <strong>理想弹簧</strong> 连接,弹簧的劲度系数 <em>k</em> = 1.0,原长 <em>L</em>₀ = 1.0(与原子间距一致,初始状态弹簧无拉伸)。</p>
<p>每个原子被限制在 <strong>z 方向</strong> 自由振动,x 和 y 方向锁定(<code>fix_x=1, fix_y=1, fix_z=0</code>)。</p>
</div>
<div class="card">
<h3>1.2 弹簧力(胡克定律)</h3>
<p>当原子 <em>i</em><em>j</em> 之间有弹簧连接时,原子 <em>i</em> 受到的弹簧力为:</p>
<div class="formula">
<strong>F</strong> = <em>k</em> · (<em>d</em> <em>L</em>₀) · <strong>u</strong><sub><em>ij</em></sub>
</div>
<p>其中 <em>d</em> = |<strong>r</strong><sub><em>j</em></sub> <strong>r</strong><sub><em>i</em></sub>| 为两原子间距离,<strong>u</strong><sub><em>ij</em></sub> 为从 <em>i</em> 指向 <em>j</em> 的单位向量。由于原子只在 z 方向振动,弹簧在 z 方向的分量是 <strong>几何非线性</strong> 的——对于小振幅近似,z 方向等效于一个三次方恢复力(FPU 型非线性)。</p>
</div>
<div class="card">
<h3>1.3 运动方程</h3>
<p>对于第 <em>i</em> 个自由原子(非受驱),牛顿第二定律给出:</p>
<div class="formula">
<em>m</em> · <strong>a</strong><sub><em>i</em></sub> = <strong>F</strong><sub><em>i</em></sub><sup>spring</sup> + <strong>F</strong><sub><em>i</em></sub><sup>driving</sup>
</div>
<p>本案例中 <strong>唯一的外力</strong> 来自驱动力(仅施加于原子 1)。无重力、无万有引力、无阻尼,系统总能量守恒。</p>
</div>
<div class="card">
<h3>1.4 波传播</h3>
<p>原子 1 的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的 <strong>横波</strong>。由于横向振动的几何非线性(弹簧大部分张力在 x 方向,z 方向的有效刚度远小于 1),波的传播速度较慢,且高阶频率成分会在链中产生复杂的非线性动力学行为(类似 FPU 回波现象)。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 2. Algorithm -->
<!-- ============================================================ -->
<section id="algorithm">
<h2>二、数值算法</h2>
<div class="card">
<h3>2.1 蛙跳法(Leapfrog / Velocity-Verlet</h3>
<p>采用能量守恒特性优异的 <strong>蛙跳法</strong>(二阶辛积分器),更新公式为:</p>
<div class="formula">
<strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) = <strong>v</strong>(<em>t</em>) + ½ <strong>a</strong>(<em>t</em>) · Δ<em>t</em><br>
<strong>r</strong>(<em>t</em> + Δ<em>t</em>) = <strong>r</strong>(<em>t</em>) + <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) · Δ<em>t</em><br>
<strong>a</strong>(<em>t</em> + Δ<em>t</em>) = <strong>F</strong>(<strong>r</strong>(<em>t</em> + Δ<em>t</em>), <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>)) / <em>m</em><br>
<strong>v</strong>(<em>t</em> + Δ<em>t</em>) = <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) + ½ <strong>a</strong>(<em>t</em> + Δ<em>t</em>) · Δ<em>t</em>
</div>
<p>蛙跳法在长时间模拟中能量漂移极小(本案例验证 <strong>&lt; 0.004%</strong>),适合无阻尼的保守系统。</p>
</div>
<div class="card">
<h3>2.2 时间步长与采样</h3>
<table>
<tr><th>参数</th><th></th><th>说明</th></tr>
<tr><td>DT</td><td>0.01 s</td><td>积分步长(远小于 1/ω ≈ 0.16 s,满足稳定性条件)</td></tr>
<tr><td>T_total</td><td>100 s</td><td>总模拟时间 → NT = 10000 步</td></tr>
<tr><td>NSTEP</td><td>50</td><td>每 NSTEP 步取一帧用于动画 → 200 帧</td></tr>
<tr><td>method</td><td>leapfrog</td><td>蛙跳法(Velocity-Verlet</td></tr>
</table>
</div>
<div class="card">
<h3>2.3 计算流程</h3>
<div class="flow">
<span class="flow-step">读入 coord.txt<br>connection.txt<br>bond.txt</span>
<span class="flow-arrow"></span>
<span class="flow-step">施加驱动力<br>(驱动原子 1</span>
<span class="flow-arrow"></span>
<span class="flow-step">记录轨迹</span>
<span class="flow-arrow"></span>
<span class="flow-step">蛙跳法<br>更新位置/速度</span>
<span class="flow-arrow"></span>
<span class="flow-step">固定约束<br>x, y 锁定)</span>
<span class="flow-arrow"></span>
<span class="flow-step" style="background:#fef3c7;border-color:#f59e0b;">循环<br>NT 次</span>
</div>
<p style="margin-top:12px;">注意:驱动力在 <strong>每次积分前</strong> 施加,确保受驱原子的位置正确传递给弹簧力计算。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 3. Driving Force -->
<!-- ============================================================ -->
<section id="driver">
<h2>三、驱动力模型</h2>
<div class="card">
<h3>3.1 定义文件</h3>
<p>驱动力由 <code>input/driver.txt</code> 定义,格式如下:</p>
<pre>n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0 0 5 0 0 1 0 0 90 all</pre>
</div>
<div class="card">
<h3>3.2 数学公式</h3>
<p>受驱原子的位置由下式决定(<strong>完全替换</strong> coord.txt 中的初始坐标和固定约束):</p>
<div class="formula">
<strong>r</strong>(<em>t</em>) = <strong>A</strong> · cos(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>速度由解析导数给出:</p>
<div class="formula">
<strong>v</strong>(<em>t</em>) = <strong>A</strong> · 2π<em>f</em> · sin(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>其中 <strong>A</strong> = (amp_x, amp_y, amp_z)<strong>f</strong> = (freq_x, freq_y, freq_z) 为不同方向的驱动频率,<strong>φ</strong> = (phi_x, phi_y, phi_z) 为相位(<strong>角度制</strong>,代码自动转换为弧度)。</p>
</div>
<div class="card">
<h3>3.3 本案例驱动参数</h3>
<table>
<tr><th>参数</th><th></th><th>含义</th></tr>
<tr><td>amp_z</td><td>5.0</td><td>z 方向驱动振幅</td></tr>
<tr><td>freq_z</td><td>1.0 Hz</td><td>驱动频率(周期 1 s</td></tr>
<tr><td>phi_z</td><td>90°</td><td>驱动相位 → z(0) = 5·cos(90°) = 0</td></tr>
<tr><td>period</td><td>all</td><td>全程驱动,永不停止</td></tr>
</table>
<div class="formula">
<em>z</em>(<em>t</em>) = 5.0 · cos(2π · 1.0 · <em>t</em> + 90°)
</div>
</div>
<div class="card">
<h3>3.4 有限周期驱动</h3>
<p><code>period</code> 参数支持三种模式:</p>
<ul>
<li><strong>all</strong> — 全程驱动</li>
<li><strong>数值</strong> — 驱动指定周期数后 <strong>静止</strong>(冻结在最终位置,速度归零)。例如 <code>period: 1</code> 表示驱动 1 个完整周期后停止。</li>
</ul>
</div>
<div class="card">
<h3>3.5 驱动与固定约束的关系</h3>
<p>对于受驱原子(<code>driver.txt</code><code>n</code> 指定的原子),其在 <code>coord.txt</code> 中的初始坐标和 <code>fix_x/fix_y/fix_z</code> 约束被 <strong>完全忽略</strong>。原子的位置和速度完全由驱动力公式决定。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 4. Usage -->
<!-- ============================================================ -->
<section id="usage">
<h2>四、使用方法</h2>
<div class="card">
<h3>4.1 完整运行(模拟 + 动画)</h3>
<pre>cd examples/case06
python run_dynamics.py</pre>
<p>这步会依次执行:物理模拟 → 抽帧 → 打开 VisPy 3D 动画窗口。</p>
</div>
<div class="card">
<h3>4.2 仅查看已有结果</h3>
<p>如果已经跑完模拟且生成了 <code>output/display.txt</code>,可以通过修改 <code>input.txt</code> 跳过计算,只开动画:</p>
<pre>step_simulate: 0 # 跳过模拟
step_sample: 0 # 跳过抽帧
step_animation: 1 # 播放动画</pre>
<p>然后运行:<code>python run_dynamics.py</code></p>
</div>
<div class="card">
<h3>4.3 手动 3D 动画</h3>
<p>也可以单独启动 VisPy 窗口:</p>
<pre>python ../../draw.py output/</pre>
</div>
<div class="card">
<h3>4.4 强制重新计算</h3>
<p>修改参数后需要重新运行模拟时,设置:</p>
<pre>force_calc: 1 # 忽略缓存,强制重新计算</pre>
</div>
<div class="card">
<h3>4.5 动画交互</h3>
<table>
<tr><th>操作</th><th>效果</th></tr>
<tr><td>鼠标拖动</td><td>旋转视角</td></tr>
<tr><td>滚轮</td><td>缩放</td></tr>
<tr><td>W / S 键</td><td>相机沿 Z 轴向前 / 向后移动(靠近/远离场景)</td></tr>
<tr><td>A / D 键</td><td>视角向右 / 向左平移</td></tr>
<tr><td>E / Q 键</td><td>视角上升 / 下降(屏幕方向)</td></tr>
<tr><td>C / X 键</td><td>增大 / 减小步长</td></tr>
<tr><td>V 键</td><td>切换透视 / 正交投影</td></tr>
<tr><td>左上角 <strong>reset</strong> 按钮</td><td>复位视角到初始位置</td></tr>
<tr><td>左上角 <strong>info</strong> 按钮</td><td>切换信息面板显示/隐藏</td></tr>
<tr><td>左上角 <strong>axes</strong> 按钮</td><td>切换坐标轴显示/隐藏</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 5. Parameters -->
<!-- ============================================================ -->
<section id="params">
<h2>五、参数参考</h2>
<div class="card">
<h3>5.1 input.txt 关键参数</h3>
<table>
<tr><th>参数</th><th>默认值</th><th>说明</th></tr>
<tr><td>gravity_field</td><td>0</td><td>均匀重力场(已关闭)</td></tr>
<tr><td>gravity_interaction</td><td>0</td><td>原子间万有引力(已关闭)</td></tr>
<tr><td>elastic_force</td><td>1</td><td>弹簧键力(已开启)</td></tr>
<tr><td>damping_force</td><td>0</td><td>阻尼(已关闭)</td></tr>
<tr><td><strong>driving_force</strong></td><td><strong>1</strong></td><td>驱动力开关(1=开启,需 driver.txt</td></tr>
<tr><td>method</td><td>leapfrog</td><td>数值积分方法</td></tr>
<tr><td>DT</td><td>0.01</td><td>积分步长 (s)</td></tr>
<tr><td>T_total</td><td>100.0</td><td>总模拟时间 (s)</td></tr>
<tr><td>NSTEP</td><td>50</td><td>抽帧步数间隔</td></tr>
<tr><td>engine</td><td>python</td><td>计算引擎(python / c / cpp / fortran</td></tr>
<tr><td>use_marker</td><td>1</td><td>渲染模式(0=Sphere 网格, 1=Marker GPU 实例化)</td></tr>
</table>
</div>
<div class="card">
<h3>5.2 流程控制参数</h3>
<table>
<tr><th>参数</th><th>0</th><th>1</th></tr>
<tr><td>step_simulate</td><td>跳过模拟(加载已有轨迹)</td><td>运行物理模拟</td></tr>
<tr><td>step_sample</td><td>跳过抽帧</td><td>从轨迹抽取显示帧</td></tr>
<tr><td>step_plot</td><td>不生成图表</td><td>生成轨迹/能量图</td></tr>
<tr><td><strong>step_plot_wave</strong></td><td>不生成波形图</td><td>生成波形能量动画 GIF</td></tr>
<tr><td>step_animation</td><td>不启动动画</td><td>自动打开 VisPy 3D 窗口</td></tr>
<tr><td>force_calc</td><td>自动检测缓存</td><td>强制重新计算</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 6. File Structure -->
<!-- ============================================================ -->
<section id="files">
<h2>六、文件结构</h2>
<pre>case06/
├── input/
│ ├── input.txt # 主配置文件(YAML 格式)
│ ├── coord.txt # 原子坐标(120 个原子)
│ ├── connection.txt # 弹簧连接关系(59 条键)
│ ├── bond.txt # 弹簧参数(k=1.0, L₀=1.0
│ └── <strong>driver.txt</strong> # <span class="cm">驱动力定义(本案例新增)</span>
├── output/
│ ├── trajectory.txt # 全量轨迹数据(50000 步 × 120 原子)
│ ├── display.txt # 抽帧后的动画数据(500 帧 × 120 原子)
│ ├── dynamics.log # 计算日志
│ ├── animation.log # 动画启动日志(闪退时排查用)
│ └── wave_animation.gif # 波形能量动画(step_plot_wave=1 时生成)
├── doc/
│ └── index.html # <span class="cm">本文档</span>
├── Readme.md # 案例简介
└── run_dynamics.py # 案例运行入口</pre>
</section>
<!-- ============================================================ -->
<!-- 7. Troubleshooting -->
<!-- ============================================================ -->
<section id="troubleshoot">
<h2>七、常见问题</h2>
<div class="card">
<h3>7.1 动画窗口闪退</h3>
<p>如果 VisPy 窗口一闪就消失,请检查:</p>
<ul>
<li><code>output/animation.log</code> 中是否有错误信息</li>
<li><code>output/display.txt</code> 是否存在(需先跑 <code>step_sample: 1</code></li>
</ul>
</div>
<div class="card">
<h3>7.2 原子不振动</h3>
<p>可能原因:</p>
<ul>
<li><strong>NSTEP 过大</strong>:抽帧间隔大于驱动周期的一半时,动画会丢失振动细节。建议 NSTEP ≤ 1/(freq · DT · 10)</li>
<li><strong>相位 φ 使采样点落在零值</strong>:试试 <code>phi_z: 0</code> 让原子在 t=0 处于振幅峰值</li>
<li>确认 <code>driving_force: 1</code><code>driver.txt</code> 中 amp_z 不为 0</li>
</ul>
</div>
<div class="card">
<h3>7.3 渲染性能慢</h3>
<p>原子数多时动画卡顿:</p>
<ul>
<li>设置 <code>use_marker: 1</code>(使用 GPU 实例化渲染替代独立网格球体)</li>
<li>增大 <code>NSTEP</code> 减少动画帧数</li>
</ul>
</div>
</section>
<hr style="border:none;border-top:1px solid var(--border);margin:40px 0;">
<footer style="text-align:center;color:var(--muted);font-size:0.85rem;margin-bottom:40px;">
Dynamics Simulation Framework &nbsp;·&nbsp; 生成于 2026-06-10
</footer>
</div>
</body>
</html>
+2
View File
@@ -0,0 +1,2 @@
bond_name k rest_length
h 100.0 1.0
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
+102
View File
@@ -0,0 +1,102 @@
n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0 0 2.0 0 0 0.05 90 90 90 all
102 0 0 2.0 0 0 0.05 90 90 90 all
203 0 0 2.0 0 0 0.05 90 90 90 all
304 0 0 2.0 0 0 0.05 90 90 90 all
405 0 0 2.0 0 0 0.05 90 90 90 all
506 0 0 2.0 0 0 0.05 90 90 90 all
607 0 0 2.0 0 0 0.05 90 90 90 all
708 0 0 2.0 0 0 0.05 90 90 90 all
809 0 0 2.0 0 0 0.05 90 90 90 all
910 0 0 2.0 0 0 0.05 90 90 90 all
1011 0 0 2.0 0 0 0.05 90 90 90 all
1112 0 0 2.0 0 0 0.05 90 90 90 all
1213 0 0 2.0 0 0 0.05 90 90 90 all
1314 0 0 2.0 0 0 0.05 90 90 90 all
1415 0 0 2.0 0 0 0.05 90 90 90 all
1516 0 0 2.0 0 0 0.05 90 90 90 all
1617 0 0 2.0 0 0 0.05 90 90 90 all
1718 0 0 2.0 0 0 0.05 90 90 90 all
1819 0 0 2.0 0 0 0.05 90 90 90 all
1920 0 0 2.0 0 0 0.05 90 90 90 all
2021 0 0 2.0 0 0 0.05 90 90 90 all
2122 0 0 2.0 0 0 0.05 90 90 90 all
2223 0 0 2.0 0 0 0.05 90 90 90 all
2324 0 0 2.0 0 0 0.05 90 90 90 all
2425 0 0 2.0 0 0 0.05 90 90 90 all
2526 0 0 2.0 0 0 0.05 90 90 90 all
2627 0 0 2.0 0 0 0.05 90 90 90 all
2728 0 0 2.0 0 0 0.05 90 90 90 all
2829 0 0 2.0 0 0 0.05 90 90 90 all
2930 0 0 2.0 0 0 0.05 90 90 90 all
3031 0 0 2.0 0 0 0.05 90 90 90 all
3132 0 0 2.0 0 0 0.05 90 90 90 all
3233 0 0 2.0 0 0 0.05 90 90 90 all
3334 0 0 2.0 0 0 0.05 90 90 90 all
3435 0 0 2.0 0 0 0.05 90 90 90 all
3536 0 0 2.0 0 0 0.05 90 90 90 all
3637 0 0 2.0 0 0 0.05 90 90 90 all
3738 0 0 2.0 0 0 0.05 90 90 90 all
3839 0 0 2.0 0 0 0.05 90 90 90 all
3940 0 0 2.0 0 0 0.05 90 90 90 all
4041 0 0 2.0 0 0 0.05 90 90 90 all
4142 0 0 2.0 0 0 0.05 90 90 90 all
4243 0 0 2.0 0 0 0.05 90 90 90 all
4344 0 0 2.0 0 0 0.05 90 90 90 all
4445 0 0 2.0 0 0 0.05 90 90 90 all
4546 0 0 2.0 0 0 0.05 90 90 90 all
4647 0 0 2.0 0 0 0.05 90 90 90 all
4748 0 0 2.0 0 0 0.05 90 90 90 all
4849 0 0 2.0 0 0 0.05 90 90 90 all
4950 0 0 2.0 0 0 0.05 90 90 90 all
5051 0 0 2.0 0 0 0.05 90 90 90 all
5152 0 0 2.0 0 0 0.05 90 90 90 all
5253 0 0 2.0 0 0 0.05 90 90 90 all
5354 0 0 2.0 0 0 0.05 90 90 90 all
5455 0 0 2.0 0 0 0.05 90 90 90 all
5556 0 0 2.0 0 0 0.05 90 90 90 all
5657 0 0 2.0 0 0 0.05 90 90 90 all
5758 0 0 2.0 0 0 0.05 90 90 90 all
5859 0 0 2.0 0 0 0.05 90 90 90 all
5960 0 0 2.0 0 0 0.05 90 90 90 all
6061 0 0 2.0 0 0 0.05 90 90 90 all
6162 0 0 2.0 0 0 0.05 90 90 90 all
6263 0 0 2.0 0 0 0.05 90 90 90 all
6364 0 0 2.0 0 0 0.05 90 90 90 all
6465 0 0 2.0 0 0 0.05 90 90 90 all
6566 0 0 2.0 0 0 0.05 90 90 90 all
6667 0 0 2.0 0 0 0.05 90 90 90 all
6768 0 0 2.0 0 0 0.05 90 90 90 all
6869 0 0 2.0 0 0 0.05 90 90 90 all
6970 0 0 2.0 0 0 0.05 90 90 90 all
7071 0 0 2.0 0 0 0.05 90 90 90 all
7172 0 0 2.0 0 0 0.05 90 90 90 all
7273 0 0 2.0 0 0 0.05 90 90 90 all
7374 0 0 2.0 0 0 0.05 90 90 90 all
7475 0 0 2.0 0 0 0.05 90 90 90 all
7576 0 0 2.0 0 0 0.05 90 90 90 all
7677 0 0 2.0 0 0 0.05 90 90 90 all
7778 0 0 2.0 0 0 0.05 90 90 90 all
7879 0 0 2.0 0 0 0.05 90 90 90 all
7980 0 0 2.0 0 0 0.05 90 90 90 all
8081 0 0 2.0 0 0 0.05 90 90 90 all
8182 0 0 2.0 0 0 0.05 90 90 90 all
8283 0 0 2.0 0 0 0.05 90 90 90 all
8384 0 0 2.0 0 0 0.05 90 90 90 all
8485 0 0 2.0 0 0 0.05 90 90 90 all
8586 0 0 2.0 0 0 0.05 90 90 90 all
8687 0 0 2.0 0 0 0.05 90 90 90 all
8788 0 0 2.0 0 0 0.05 90 90 90 all
8889 0 0 2.0 0 0 0.05 90 90 90 all
8990 0 0 2.0 0 0 0.05 90 90 90 all
9091 0 0 2.0 0 0 0.05 90 90 90 all
9192 0 0 2.0 0 0 0.05 90 90 90 all
9293 0 0 2.0 0 0 0.05 90 90 90 all
9394 0 0 2.0 0 0 0.05 90 90 90 all
9495 0 0 2.0 0 0 0.05 90 90 90 all
9596 0 0 2.0 0 0 0.05 90 90 90 all
9697 0 0 2.0 0 0 0.05 90 90 90 all
9798 0 0 2.0 0 0 0.05 90 90 90 all
9899 0 0 2.0 0 0 0.05 90 90 90 all
10000 0 0 2.0 0 0 0.05 90 90 90 all
10101 0 0 2.0 0 0 0.05 90 90 90 all
+73
View File
@@ -0,0 +1,73 @@
# 物理模拟参数配置
# case13 — 二维网格平面波(左边界驱动,右边界吸收)
# 左边界全部 101 原子齐振驱动 → 波从左向右传播 → 右边界全固定
step_simulate: 1
step_sample: 0
step_plot: 0
step_animation: 1
step_plot_wave: 0
force_calc: 1
save_trajectory: 0
engine: c
box_a: 120.0
coord_file: input/coord.txt
connection_file: input/connection.txt
bond_file: input/bond.txt
driver_file: input/driver.txt
plot_atom: 51 # 左边界中间原子用于信息显示
G: [0.000, 0.000, 0.000]
B: [0.000, 0.000, 0.000]
gravity_field: 0
gravity_interaction: 0
elastic_force: 1
damping_force: 0
driving_force: 1
gravity_strength: 1.0
method: leapfrog
warmup_steps: 0
T_total: 200.0
NSTEP: 100
DT: 0.01
sample_start: null
sample_end: null
# ── 渲染/着色 ─────────────────────────────────
use_marker: 1
display_color: {
x : [1, [255, 0, 0]],
y : [1, [ 0, 255, 0]],
z : [1, [ 0, 0, 255]],
xy : [0, [255, 255, 0]],
yz : [0, [ 0, 255, 255]],
zx : [0, [255, 0, 255]],
xyz : [0, [255, 255, 255]],
}
alpha: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
ball_color_r: 0.20
ball_color_g: 0.60
ball_color_b: 0.90
box_color_r: 0.80
box_color_g: 0.80
box_color_b: 0.85
camera_distance: 120.0
camera_elevation: 60.0
camera_azimuth: -45.0
camera_center_x: 0.0
camera_center_y: 0.0
camera_center_z: 0.0
move_camera: 0
display_amp: [1.0, 1.0, 1.0]
+2
View File
@@ -0,0 +1,2 @@
0 0 50
0 0 80
+54
View File
@@ -0,0 +1,54 @@
"""
Case runner for Dynamics case13 2D grid (61x61 atomic mesh).
This script keeps program and data separated:
- program: ../../dynamics.py
- input: ./input
- output: ./output
"""
from __future__ import annotations
import argparse
import importlib.util
from pathlib import Path
CASE_DIR = Path(__file__).resolve().parent
DYNAMICS_PATH = Path("..") / ".." / "dynamics.py"
INPUT_DIR = Path("input")
OUTPUT_DIR = Path("output")
CONFIG_FILE = INPUT_DIR / "input.txt"
def load_dynamics_module(module_path: Path):
spec = importlib.util.spec_from_file_location("dynamics_module", module_path)
if spec is None or spec.loader is None:
raise ImportError(f"无法加载 dynamics.py: {module_path}")
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
return module
def main():
parser = argparse.ArgumentParser(description="运行 Dynamics 示例案例 case13")
parser.add_argument("--no-plot", action="store_true", help="跳过 matplotlib 绘图")
args = parser.parse_args()
dynamics_path = (CASE_DIR / DYNAMICS_PATH).resolve()
input_dir = (CASE_DIR / INPUT_DIR).resolve()
output_dir = (CASE_DIR / OUTPUT_DIR).resolve()
config_path = (CASE_DIR / CONFIG_FILE).resolve()
module = load_dynamics_module(dynamics_path)
module.run_case(
config_path=config_path,
runtime_base=CASE_DIR,
input_dir=input_dir,
output_dir=output_dir,
no_plot=args.no_plot,
)
if __name__ == "__main__":
main()
+40
View File
@@ -0,0 +1,40 @@
# case06: 一维原子链横波模拟
60 个原子沿 x 轴排列,相邻原子用弹簧连接。原子 1 受 z 方向驱动力作用,产生沿链传播的横波。
## 物理设定
| 参数 | 值 |
|---|---|
| 原子数 | 120 |
| 排列 | 沿 x 轴等间距排列,间距为 1 |
| 约束 | 原子**沿 z 方向自由振动**fix_x=1, fix_y=1, fix_z=0),x, y 锁定 |
| 弹簧 | 劲度系数 k=1.0,原长 L₀=1.0 |
| 重力 | 无 |
| 万有引力 | 无 |
| 阻尼 | 无 |
| 驱动力 | 原子 1(z 方向驱动) |
| 算法 | leapfrog(蛙跳法,能量守恒) |
## 驱动力
原子 1 的位置由 `input/driver.txt` 中的驱动力公式决定:
```math
z(t) = A_z \cdot \cos(2\pi f_z t + \phi_z)
```
当前参数:A_z = 0.5, f_z = 0.1 Hz, φ_z = 90°, period = all(全程驱动)。
## 动力学行为
原子 1 沿 z 方向的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的**横波**。由于 z 方向的振动是横向的,弹簧大部分张力在 x 方向,z 方向的有效刚度是非线性的——等效于一个三次方恢复力(FPU 型非线性),因此波速较慢。
## 使用方法
```bash
cd examples/case06
python run_dynamics.py
```
配置参数详见 `input/input.txt`,驱动力定义见 `input/driver.txt`,完整文档见 `doc/index.html`
+477
View File
@@ -0,0 +1,477 @@
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>case06 — 一维原子链驱动力学模拟 | 物理原理 &amp; 使用文档</title>
<style>
:root {
--bg: #f8f9fa;
--card: #fff;
--text: #1a1a2e;
--accent: #2563eb;
--accent-light: #dbeafe;
--code-bg: #1e293b;
--code-text: #e2e8f0;
--border: #e2e8f0;
--muted: #64748b;
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Noto Sans SC", sans-serif;
background: var(--bg);
color: var(--text);
line-height: 1.7;
}
/* ── Header ── */
.hero {
background: linear-gradient(135deg, #1e293b 0%, #334155 100%);
color: #fff;
padding: 56px 24px 48px;
text-align: center;
}
.hero h1 { font-size: 2rem; font-weight: 700; letter-spacing: -0.02em; }
.hero .subtitle {
margin-top: 10px;
font-size: 1.05rem;
opacity: 0.8;
}
.hero .badge {
display: inline-block;
margin-top: 14px;
padding: 4px 14px;
border-radius: 999px;
background: rgba(255,255,255,0.12);
font-size: 0.82rem;
}
/* ── Layout ── */
.container { max-width: 820px; margin: 0 auto; padding: 32px 20px; }
section { margin-bottom: 44px; }
h2 {
font-size: 1.35rem;
font-weight: 600;
margin-bottom: 16px;
padding-bottom: 8px;
border-bottom: 2px solid var(--accent);
display: inline-block;
}
h3 {
font-size: 1.05rem;
font-weight: 600;
margin: 20px 0 10px;
}
p, li { margin-bottom: 10px; }
ul, ol { padding-left: 22px; }
strong { color: var(--accent); }
/* ── Cards ── */
.card {
background: var(--card);
border-radius: 12px;
padding: 20px 24px;
margin-bottom: 16px;
border: 1px solid var(--border);
box-shadow: 0 1px 3px rgba(0,0,0,0.04);
}
/* ── Formula / Code blocks ── */
.formula {
background: var(--card);
border-left: 4px solid var(--accent);
padding: 14px 20px;
margin: 14px 0;
font-family: "Times New Roman", "STIX", serif;
font-size: 1.05rem;
overflow-x: auto;
border-radius: 0 8px 8px 0;
}
code {
background: var(--accent-light);
padding: 2px 7px;
border-radius: 4px;
font-family: "JetBrains Mono", "Fira Code", monospace;
font-size: 0.88em;
}
pre {
background: var(--code-bg);
color: var(--code-text);
padding: 16px 20px;
border-radius: 10px;
overflow-x: auto;
font-size: 0.85rem;
line-height: 1.5;
margin: 14px 0;
}
pre .cm { color: #94a3b8; font-style: italic; } /* comment */
/* ── Table ── */
table {
width: 100%;
border-collapse: collapse;
margin: 14px 0;
font-size: 0.92rem;
}
th, td {
padding: 8px 12px;
text-align: left;
border-bottom: 1px solid var(--border);
}
th { background: var(--accent-light); font-weight: 600; }
/* ── TOC ── */
.toc { counter-reset: toc; }
.toc li { counter-increment: toc; list-style: none; margin-bottom: 6px; }
.toc li::before { content: counter(toc) ". "; font-weight: 600; color: var(--accent); }
.toc a { color: var(--accent); text-decoration: none; }
.toc a:hover { text-decoration: underline; }
/* ── Flow diagram ── */
.flow { display: flex; flex-wrap: wrap; gap: 8px; align-items: center; justify-content: center; margin: 16px 0; }
.flow-step {
background: var(--accent-light);
border: 1px solid var(--accent);
border-radius: 8px;
padding: 8px 16px;
font-size: 0.88rem;
font-weight: 500;
}
.flow-arrow { color: var(--muted); font-size: 1.2rem; }
@media (max-width: 600px) {
.hero h1 { font-size: 1.5rem; }
.flow { flex-direction: column; }
.flow-arrow { transform: rotate(90deg); }
}
</style>
</head>
<body>
<!-- ============================================================ -->
<!-- Header -->
<!-- ============================================================ -->
<header class="hero">
<h1>一维原子链驱动力学模拟</h1>
<p class="subtitle">120 个原子沿 x 轴排列 · 弹簧连接 · z 方向受迫振动</p>
<span class="badge">case06 · examples/case06</span>
</header>
<div class="container">
<!-- ============================================================ -->
<!-- TOC -->
<!-- ============================================================ -->
<section>
<h2>目录</h2>
<ol class="toc">
<li><a href="#physics">物理原理</a></li>
<li><a href="#algorithm">数值算法</a></li>
<li><a href="#driver">驱动力模型</a></li>
<li><a href="#usage">使用方法</a></li>
<li><a href="#params">参数参考</a></li>
<li><a href="#files">文件结构</a></li>
<li><a href="#troubleshoot">常见问题</a></li>
</ol>
</section>
<!-- ============================================================ -->
<!-- 1. Physics -->
<!-- ============================================================ -->
<section id="physics">
<h2>一、物理原理</h2>
<div class="card">
<h3>1.1 一维原子链</h3>
<p>120 个原子沿 <strong>x 轴</strong> 等间距排列,原子间距为 1。相邻原子之间用 <strong>理想弹簧</strong> 连接,弹簧的劲度系数 <em>k</em> = 1.0,原长 <em>L</em>₀ = 1.0(与原子间距一致,初始状态弹簧无拉伸)。</p>
<p>每个原子被限制在 <strong>z 方向</strong> 自由振动,x 和 y 方向锁定(<code>fix_x=1, fix_y=1, fix_z=0</code>)。</p>
</div>
<div class="card">
<h3>1.2 弹簧力(胡克定律)</h3>
<p>当原子 <em>i</em><em>j</em> 之间有弹簧连接时,原子 <em>i</em> 受到的弹簧力为:</p>
<div class="formula">
<strong>F</strong> = <em>k</em> · (<em>d</em> <em>L</em>₀) · <strong>u</strong><sub><em>ij</em></sub>
</div>
<p>其中 <em>d</em> = |<strong>r</strong><sub><em>j</em></sub> <strong>r</strong><sub><em>i</em></sub>| 为两原子间距离,<strong>u</strong><sub><em>ij</em></sub> 为从 <em>i</em> 指向 <em>j</em> 的单位向量。由于原子只在 z 方向振动,弹簧在 z 方向的分量是 <strong>几何非线性</strong> 的——对于小振幅近似,z 方向等效于一个三次方恢复力(FPU 型非线性)。</p>
</div>
<div class="card">
<h3>1.3 运动方程</h3>
<p>对于第 <em>i</em> 个自由原子(非受驱),牛顿第二定律给出:</p>
<div class="formula">
<em>m</em> · <strong>a</strong><sub><em>i</em></sub> = <strong>F</strong><sub><em>i</em></sub><sup>spring</sup> + <strong>F</strong><sub><em>i</em></sub><sup>driving</sup>
</div>
<p>本案例中 <strong>唯一的外力</strong> 来自驱动力(仅施加于原子 1)。无重力、无万有引力、无阻尼,系统总能量守恒。</p>
</div>
<div class="card">
<h3>1.4 波传播</h3>
<p>原子 1 的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的 <strong>横波</strong>。由于横向振动的几何非线性(弹簧大部分张力在 x 方向,z 方向的有效刚度远小于 1),波的传播速度较慢,且高阶频率成分会在链中产生复杂的非线性动力学行为(类似 FPU 回波现象)。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 2. Algorithm -->
<!-- ============================================================ -->
<section id="algorithm">
<h2>二、数值算法</h2>
<div class="card">
<h3>2.1 蛙跳法(Leapfrog / Velocity-Verlet</h3>
<p>采用能量守恒特性优异的 <strong>蛙跳法</strong>(二阶辛积分器),更新公式为:</p>
<div class="formula">
<strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) = <strong>v</strong>(<em>t</em>) + ½ <strong>a</strong>(<em>t</em>) · Δ<em>t</em><br>
<strong>r</strong>(<em>t</em> + Δ<em>t</em>) = <strong>r</strong>(<em>t</em>) + <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) · Δ<em>t</em><br>
<strong>a</strong>(<em>t</em> + Δ<em>t</em>) = <strong>F</strong>(<strong>r</strong>(<em>t</em> + Δ<em>t</em>), <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>)) / <em>m</em><br>
<strong>v</strong>(<em>t</em> + Δ<em>t</em>) = <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) + ½ <strong>a</strong>(<em>t</em> + Δ<em>t</em>) · Δ<em>t</em>
</div>
<p>蛙跳法在长时间模拟中能量漂移极小(本案例验证 <strong>&lt; 0.004%</strong>),适合无阻尼的保守系统。</p>
</div>
<div class="card">
<h3>2.2 时间步长与采样</h3>
<table>
<tr><th>参数</th><th></th><th>说明</th></tr>
<tr><td>DT</td><td>0.01 s</td><td>积分步长(远小于 1/ω ≈ 0.16 s,满足稳定性条件)</td></tr>
<tr><td>T_total</td><td>100 s</td><td>总模拟时间 → NT = 10000 步</td></tr>
<tr><td>NSTEP</td><td>50</td><td>每 NSTEP 步取一帧用于动画 → 200 帧</td></tr>
<tr><td>method</td><td>leapfrog</td><td>蛙跳法(Velocity-Verlet</td></tr>
</table>
</div>
<div class="card">
<h3>2.3 计算流程</h3>
<div class="flow">
<span class="flow-step">读入 coord.txt<br>connection.txt<br>bond.txt</span>
<span class="flow-arrow"></span>
<span class="flow-step">施加驱动力<br>(驱动原子 1</span>
<span class="flow-arrow"></span>
<span class="flow-step">记录轨迹</span>
<span class="flow-arrow"></span>
<span class="flow-step">蛙跳法<br>更新位置/速度</span>
<span class="flow-arrow"></span>
<span class="flow-step">固定约束<br>x, y 锁定)</span>
<span class="flow-arrow"></span>
<span class="flow-step" style="background:#fef3c7;border-color:#f59e0b;">循环<br>NT 次</span>
</div>
<p style="margin-top:12px;">注意:驱动力在 <strong>每次积分前</strong> 施加,确保受驱原子的位置正确传递给弹簧力计算。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 3. Driving Force -->
<!-- ============================================================ -->
<section id="driver">
<h2>三、驱动力模型</h2>
<div class="card">
<h3>3.1 定义文件</h3>
<p>驱动力由 <code>input/driver.txt</code> 定义,格式如下:</p>
<pre>n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0 0 5 0 0 1 0 0 90 all</pre>
</div>
<div class="card">
<h3>3.2 数学公式</h3>
<p>受驱原子的位置由下式决定(<strong>完全替换</strong> coord.txt 中的初始坐标和固定约束):</p>
<div class="formula">
<strong>r</strong>(<em>t</em>) = <strong>A</strong> · cos(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>速度由解析导数给出:</p>
<div class="formula">
<strong>v</strong>(<em>t</em>) = <strong>A</strong> · 2π<em>f</em> · sin(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>其中 <strong>A</strong> = (amp_x, amp_y, amp_z)<strong>f</strong> = (freq_x, freq_y, freq_z) 为不同方向的驱动频率,<strong>φ</strong> = (phi_x, phi_y, phi_z) 为相位(<strong>角度制</strong>,代码自动转换为弧度)。</p>
</div>
<div class="card">
<h3>3.3 本案例驱动参数</h3>
<table>
<tr><th>参数</th><th></th><th>含义</th></tr>
<tr><td>amp_z</td><td>5.0</td><td>z 方向驱动振幅</td></tr>
<tr><td>freq_z</td><td>1.0 Hz</td><td>驱动频率(周期 1 s</td></tr>
<tr><td>phi_z</td><td>90°</td><td>驱动相位 → z(0) = 5·cos(90°) = 0</td></tr>
<tr><td>period</td><td>all</td><td>全程驱动,永不停止</td></tr>
</table>
<div class="formula">
<em>z</em>(<em>t</em>) = 5.0 · cos(2π · 1.0 · <em>t</em> + 90°)
</div>
</div>
<div class="card">
<h3>3.4 有限周期驱动</h3>
<p><code>period</code> 参数支持三种模式:</p>
<ul>
<li><strong>all</strong> — 全程驱动</li>
<li><strong>数值</strong> — 驱动指定周期数后 <strong>静止</strong>(冻结在最终位置,速度归零)。例如 <code>period: 1</code> 表示驱动 1 个完整周期后停止。</li>
</ul>
</div>
<div class="card">
<h3>3.5 驱动与固定约束的关系</h3>
<p>对于受驱原子(<code>driver.txt</code><code>n</code> 指定的原子),其在 <code>coord.txt</code> 中的初始坐标和 <code>fix_x/fix_y/fix_z</code> 约束被 <strong>完全忽略</strong>。原子的位置和速度完全由驱动力公式决定。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 4. Usage -->
<!-- ============================================================ -->
<section id="usage">
<h2>四、使用方法</h2>
<div class="card">
<h3>4.1 完整运行(模拟 + 动画)</h3>
<pre>cd examples/case06
python run_dynamics.py</pre>
<p>这步会依次执行:物理模拟 → 抽帧 → 打开 VisPy 3D 动画窗口。</p>
</div>
<div class="card">
<h3>4.2 仅查看已有结果</h3>
<p>如果已经跑完模拟且生成了 <code>output/display.txt</code>,可以通过修改 <code>input.txt</code> 跳过计算,只开动画:</p>
<pre>step_simulate: 0 # 跳过模拟
step_sample: 0 # 跳过抽帧
step_animation: 1 # 播放动画</pre>
<p>然后运行:<code>python run_dynamics.py</code></p>
</div>
<div class="card">
<h3>4.3 手动 3D 动画</h3>
<p>也可以单独启动 VisPy 窗口:</p>
<pre>python ../../draw.py output/</pre>
</div>
<div class="card">
<h3>4.4 强制重新计算</h3>
<p>修改参数后需要重新运行模拟时,设置:</p>
<pre>force_calc: 1 # 忽略缓存,强制重新计算</pre>
</div>
<div class="card">
<h3>4.5 动画交互</h3>
<table>
<tr><th>操作</th><th>效果</th></tr>
<tr><td>鼠标拖动</td><td>旋转视角</td></tr>
<tr><td>滚轮</td><td>缩放</td></tr>
<tr><td>W / S 键</td><td>相机沿 Z 轴向前 / 向后移动(靠近/远离场景)</td></tr>
<tr><td>A / D 键</td><td>视角向右 / 向左平移</td></tr>
<tr><td>E / Q 键</td><td>视角上升 / 下降(屏幕方向)</td></tr>
<tr><td>C / X 键</td><td>增大 / 减小步长</td></tr>
<tr><td>V 键</td><td>切换透视 / 正交投影</td></tr>
<tr><td>左上角 <strong>reset</strong> 按钮</td><td>复位视角到初始位置</td></tr>
<tr><td>左上角 <strong>info</strong> 按钮</td><td>切换信息面板显示/隐藏</td></tr>
<tr><td>左上角 <strong>axes</strong> 按钮</td><td>切换坐标轴显示/隐藏</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 5. Parameters -->
<!-- ============================================================ -->
<section id="params">
<h2>五、参数参考</h2>
<div class="card">
<h3>5.1 input.txt 关键参数</h3>
<table>
<tr><th>参数</th><th>默认值</th><th>说明</th></tr>
<tr><td>gravity_field</td><td>0</td><td>均匀重力场(已关闭)</td></tr>
<tr><td>gravity_interaction</td><td>0</td><td>原子间万有引力(已关闭)</td></tr>
<tr><td>elastic_force</td><td>1</td><td>弹簧键力(已开启)</td></tr>
<tr><td>damping_force</td><td>0</td><td>阻尼(已关闭)</td></tr>
<tr><td><strong>driving_force</strong></td><td><strong>1</strong></td><td>驱动力开关(1=开启,需 driver.txt</td></tr>
<tr><td>method</td><td>leapfrog</td><td>数值积分方法</td></tr>
<tr><td>DT</td><td>0.01</td><td>积分步长 (s)</td></tr>
<tr><td>T_total</td><td>100.0</td><td>总模拟时间 (s)</td></tr>
<tr><td>NSTEP</td><td>50</td><td>抽帧步数间隔</td></tr>
<tr><td>engine</td><td>python</td><td>计算引擎(python / c / cpp / fortran</td></tr>
<tr><td>use_marker</td><td>1</td><td>渲染模式(0=Sphere 网格, 1=Marker GPU 实例化)</td></tr>
</table>
</div>
<div class="card">
<h3>5.2 流程控制参数</h3>
<table>
<tr><th>参数</th><th>0</th><th>1</th></tr>
<tr><td>step_simulate</td><td>跳过模拟(加载已有轨迹)</td><td>运行物理模拟</td></tr>
<tr><td>step_sample</td><td>跳过抽帧</td><td>从轨迹抽取显示帧</td></tr>
<tr><td>step_plot</td><td>不生成图表</td><td>生成轨迹/能量图</td></tr>
<tr><td><strong>step_plot_wave</strong></td><td>不生成波形图</td><td>生成波形能量动画 GIF</td></tr>
<tr><td>step_animation</td><td>不启动动画</td><td>自动打开 VisPy 3D 窗口</td></tr>
<tr><td>force_calc</td><td>自动检测缓存</td><td>强制重新计算</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 6. File Structure -->
<!-- ============================================================ -->
<section id="files">
<h2>六、文件结构</h2>
<pre>case06/
├── input/
│ ├── input.txt # 主配置文件(YAML 格式)
│ ├── coord.txt # 原子坐标(120 个原子)
│ ├── connection.txt # 弹簧连接关系(59 条键)
│ ├── bond.txt # 弹簧参数(k=1.0, L₀=1.0
│ └── <strong>driver.txt</strong> # <span class="cm">驱动力定义(本案例新增)</span>
├── output/
│ ├── trajectory.txt # 全量轨迹数据(50000 步 × 120 原子)
│ ├── display.txt # 抽帧后的动画数据(500 帧 × 120 原子)
│ ├── dynamics.log # 计算日志
│ ├── animation.log # 动画启动日志(闪退时排查用)
│ └── wave_animation.gif # 波形能量动画(step_plot_wave=1 时生成)
├── doc/
│ └── index.html # <span class="cm">本文档</span>
├── Readme.md # 案例简介
└── run_dynamics.py # 案例运行入口</pre>
</section>
<!-- ============================================================ -->
<!-- 7. Troubleshooting -->
<!-- ============================================================ -->
<section id="troubleshoot">
<h2>七、常见问题</h2>
<div class="card">
<h3>7.1 动画窗口闪退</h3>
<p>如果 VisPy 窗口一闪就消失,请检查:</p>
<ul>
<li><code>output/animation.log</code> 中是否有错误信息</li>
<li><code>output/display.txt</code> 是否存在(需先跑 <code>step_sample: 1</code></li>
</ul>
</div>
<div class="card">
<h3>7.2 原子不振动</h3>
<p>可能原因:</p>
<ul>
<li><strong>NSTEP 过大</strong>:抽帧间隔大于驱动周期的一半时,动画会丢失振动细节。建议 NSTEP ≤ 1/(freq · DT · 10)</li>
<li><strong>相位 φ 使采样点落在零值</strong>:试试 <code>phi_z: 0</code> 让原子在 t=0 处于振幅峰值</li>
<li>确认 <code>driving_force: 1</code><code>driver.txt</code> 中 amp_z 不为 0</li>
</ul>
</div>
<div class="card">
<h3>7.3 渲染性能慢</h3>
<p>原子数多时动画卡顿:</p>
<ul>
<li>设置 <code>use_marker: 1</code>(使用 GPU 实例化渲染替代独立网格球体)</li>
<li>增大 <code>NSTEP</code> 减少动画帧数</li>
</ul>
</div>
</section>
<hr style="border:none;border-top:1px solid var(--border);margin:40px 0;">
<footer style="text-align:center;color:var(--muted);font-size:0.85rem;margin-bottom:40px;">
Dynamics Simulation Framework &nbsp;·&nbsp; 生成于 2026-06-10
</footer>
</div>
</body>
</html>
+2
View File
@@ -0,0 +1,2 @@
bond_name k rest_length
h 100.0 1.0
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
+102
View File
@@ -0,0 +1,102 @@
n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0 0 2.0 0 0 0.05 90 90 90 all
102 0 0 2.0 0 0 0.05 90 90 90 all
203 0 0 2.0 0 0 0.05 90 90 90 all
304 0 0 2.0 0 0 0.05 90 90 90 all
405 0 0 2.0 0 0 0.05 90 90 90 all
506 0 0 2.0 0 0 0.05 90 90 90 all
607 0 0 2.0 0 0 0.05 90 90 90 all
708 0 0 2.0 0 0 0.05 90 90 90 all
809 0 0 2.0 0 0 0.05 90 90 90 all
910 0 0 2.0 0 0 0.05 90 90 90 all
1011 0 0 2.0 0 0 0.05 90 90 90 all
1112 0 0 2.0 0 0 0.05 90 90 90 all
1213 0 0 2.0 0 0 0.05 90 90 90 all
1314 0 0 2.0 0 0 0.05 90 90 90 all
1415 0 0 2.0 0 0 0.05 90 90 90 all
1516 0 0 2.0 0 0 0.05 90 90 90 all
1617 0 0 2.0 0 0 0.05 90 90 90 all
1718 0 0 2.0 0 0 0.05 90 90 90 all
1819 0 0 2.0 0 0 0.05 90 90 90 all
1920 0 0 2.0 0 0 0.05 90 90 90 all
2021 0 0 2.0 0 0 0.05 90 90 90 all
2122 0 0 2.0 0 0 0.05 90 90 90 all
2223 0 0 2.0 0 0 0.05 90 90 90 all
2324 0 0 2.0 0 0 0.05 90 90 90 all
2425 0 0 2.0 0 0 0.05 90 90 90 all
2526 0 0 2.0 0 0 0.05 90 90 90 all
2627 0 0 2.0 0 0 0.05 90 90 90 all
2728 0 0 2.0 0 0 0.05 90 90 90 all
2829 0 0 2.0 0 0 0.05 90 90 90 all
2930 0 0 2.0 0 0 0.05 90 90 90 all
3031 0 0 2.0 0 0 0.05 90 90 90 all
3132 0 0 2.0 0 0 0.05 90 90 90 all
3233 0 0 2.0 0 0 0.05 90 90 90 all
3334 0 0 2.0 0 0 0.05 90 90 90 all
3435 0 0 2.0 0 0 0.05 90 90 90 all
3536 0 0 2.0 0 0 0.05 90 90 90 all
3637 0 0 2.0 0 0 0.05 90 90 90 all
3738 0 0 2.0 0 0 0.05 90 90 90 all
3839 0 0 2.0 0 0 0.05 90 90 90 all
3940 0 0 2.0 0 0 0.05 90 90 90 all
4041 0 0 2.0 0 0 0.05 90 90 90 all
4142 0 0 2.0 0 0 0.05 90 90 90 all
4243 0 0 2.0 0 0 0.05 90 90 90 all
4344 0 0 2.0 0 0 0.05 90 90 90 all
4445 0 0 2.0 0 0 0.05 90 90 90 all
4546 0 0 2.0 0 0 0.05 90 90 90 all
4647 0 0 2.0 0 0 0.05 90 90 90 all
4748 0 0 2.0 0 0 0.05 90 90 90 all
4849 0 0 2.0 0 0 0.05 90 90 90 all
4950 0 0 2.0 0 0 0.05 90 90 90 all
5051 0 0 2.0 0 0 0.05 90 90 90 all
5152 0 0 2.0 0 0 0.05 90 90 90 all
5253 0 0 2.0 0 0 0.05 90 90 90 all
5354 0 0 2.0 0 0 0.05 90 90 90 all
5455 0 0 2.0 0 0 0.05 90 90 90 all
5556 0 0 2.0 0 0 0.05 90 90 90 all
5657 0 0 2.0 0 0 0.05 90 90 90 all
5758 0 0 2.0 0 0 0.05 90 90 90 all
5859 0 0 2.0 0 0 0.05 90 90 90 all
5960 0 0 2.0 0 0 0.05 90 90 90 all
6061 0 0 2.0 0 0 0.05 90 90 90 all
6162 0 0 2.0 0 0 0.05 90 90 90 all
6263 0 0 2.0 0 0 0.05 90 90 90 all
6364 0 0 2.0 0 0 0.05 90 90 90 all
6465 0 0 2.0 0 0 0.05 90 90 90 all
6566 0 0 2.0 0 0 0.05 90 90 90 all
6667 0 0 2.0 0 0 0.05 90 90 90 all
6768 0 0 2.0 0 0 0.05 90 90 90 all
6869 0 0 2.0 0 0 0.05 90 90 90 all
6970 0 0 2.0 0 0 0.05 90 90 90 all
7071 0 0 2.0 0 0 0.05 90 90 90 all
7172 0 0 2.0 0 0 0.05 90 90 90 all
7273 0 0 2.0 0 0 0.05 90 90 90 all
7374 0 0 2.0 0 0 0.05 90 90 90 all
7475 0 0 2.0 0 0 0.05 90 90 90 all
7576 0 0 2.0 0 0 0.05 90 90 90 all
7677 0 0 2.0 0 0 0.05 90 90 90 all
7778 0 0 2.0 0 0 0.05 90 90 90 all
7879 0 0 2.0 0 0 0.05 90 90 90 all
7980 0 0 2.0 0 0 0.05 90 90 90 all
8081 0 0 2.0 0 0 0.05 90 90 90 all
8182 0 0 2.0 0 0 0.05 90 90 90 all
8283 0 0 2.0 0 0 0.05 90 90 90 all
8384 0 0 2.0 0 0 0.05 90 90 90 all
8485 0 0 2.0 0 0 0.05 90 90 90 all
8586 0 0 2.0 0 0 0.05 90 90 90 all
8687 0 0 2.0 0 0 0.05 90 90 90 all
8788 0 0 2.0 0 0 0.05 90 90 90 all
8889 0 0 2.0 0 0 0.05 90 90 90 all
8990 0 0 2.0 0 0 0.05 90 90 90 all
9091 0 0 2.0 0 0 0.05 90 90 90 all
9192 0 0 2.0 0 0 0.05 90 90 90 all
9293 0 0 2.0 0 0 0.05 90 90 90 all
9394 0 0 2.0 0 0 0.05 90 90 90 all
9495 0 0 2.0 0 0 0.05 90 90 90 all
9596 0 0 2.0 0 0 0.05 90 90 90 all
9697 0 0 2.0 0 0 0.05 90 90 90 all
9798 0 0 2.0 0 0 0.05 90 90 90 all
9899 0 0 2.0 0 0 0.05 90 90 90 all
10000 0 0 2.0 0 0 0.05 90 90 90 all
10101 0 0 2.0 0 0 0.05 90 90 90 all
+72
View File
@@ -0,0 +1,72 @@
# 物理模拟参数配置
# case14 — 双缝干涉实验
# 左边界平面波 → 双缝势垒 (x=0) → 干涉图案 → 右边界反射
step_simulate: 1
step_sample: 0
step_plot: 0
step_animation: 1
step_plot_wave: 0
force_calc: 1
save_trajectory: 0
engine: c
box_a: 120.0
coord_file: input/coord.txt
connection_file: input/connection.txt
bond_file: input/bond.txt
driver_file: input/driver.txt
plot_atom: 51
G: [0.000, 0.000, 0.000]
B: [0.000, 0.000, 0.000]
gravity_field: 0
gravity_interaction: 0
elastic_force: 1
damping_force: 0
driving_force: 1
gravity_strength: 1.0
method: leapfrog
warmup_steps: 0
T_total: 100.0
NSTEP: 100
DT: 0.001
sample_start: null
sample_end: null
use_marker: 1
display_color: {
x : [0, [255, 0, 0]],
y : [0, [ 0, 255, 0]],
z : [1, [ 0, 0, 255]],
xy : [0, [255, 255, 0]],
yz : [0, [ 0, 255, 255]],
zx : [0, [255, 0, 255]],
xyz : [0, [255, 255, 255]],
}
alpha: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
ball_color_r: 0.20
ball_color_g: 0.60
ball_color_b: 0.90
box_color_r: 0.80
box_color_g: 0.80
box_color_b: 0.85
camera_distance: 120.0
camera_elevation: 60.0
camera_azimuth: -45.0
camera_center_x: 0.0
camera_center_y: 0.0
camera_center_z: 0.0
move_camera: 0
display_amp: [1.0, 1.0, 1.0]
+2
View File
@@ -0,0 +1,2 @@
0 0 50
0 0 80
+54
View File
@@ -0,0 +1,54 @@
"""
Case runner for Dynamics case14 2D grid (61x61 atomic mesh).
This script keeps program and data separated:
- program: ../../dynamics.py
- input: ./input
- output: ./output
"""
from __future__ import annotations
import argparse
import importlib.util
from pathlib import Path
CASE_DIR = Path(__file__).resolve().parent
DYNAMICS_PATH = Path("..") / ".." / "dynamics.py"
INPUT_DIR = Path("input")
OUTPUT_DIR = Path("output")
CONFIG_FILE = INPUT_DIR / "input.txt"
def load_dynamics_module(module_path: Path):
spec = importlib.util.spec_from_file_location("dynamics_module", module_path)
if spec is None or spec.loader is None:
raise ImportError(f"无法加载 dynamics.py: {module_path}")
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
return module
def main():
parser = argparse.ArgumentParser(description="运行 Dynamics 示例案例 case14")
parser.add_argument("--no-plot", action="store_true", help="跳过 matplotlib 绘图")
args = parser.parse_args()
dynamics_path = (CASE_DIR / DYNAMICS_PATH).resolve()
input_dir = (CASE_DIR / INPUT_DIR).resolve()
output_dir = (CASE_DIR / OUTPUT_DIR).resolve()
config_path = (CASE_DIR / CONFIG_FILE).resolve()
module = load_dynamics_module(dynamics_path)
module.run_case(
config_path=config_path,
runtime_base=CASE_DIR,
input_dir=input_dir,
output_dir=output_dir,
no_plot=args.no_plot,
)
if __name__ == "__main__":
main()
+40
View File
@@ -0,0 +1,40 @@
# case06: 一维原子链横波模拟
60 个原子沿 x 轴排列,相邻原子用弹簧连接。原子 1 受 z 方向驱动力作用,产生沿链传播的横波。
## 物理设定
| 参数 | 值 |
|---|---|
| 原子数 | 120 |
| 排列 | 沿 x 轴等间距排列,间距为 1 |
| 约束 | 原子**沿 z 方向自由振动**fix_x=1, fix_y=1, fix_z=0),x, y 锁定 |
| 弹簧 | 劲度系数 k=1.0,原长 L₀=1.0 |
| 重力 | 无 |
| 万有引力 | 无 |
| 阻尼 | 无 |
| 驱动力 | 原子 1(z 方向驱动) |
| 算法 | leapfrog(蛙跳法,能量守恒) |
## 驱动力
原子 1 的位置由 `input/driver.txt` 中的驱动力公式决定:
```math
z(t) = A_z \cdot \cos(2\pi f_z t + \phi_z)
```
当前参数:A_z = 0.5, f_z = 0.1 Hz, φ_z = 90°, period = all(全程驱动)。
## 动力学行为
原子 1 沿 z 方向的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的**横波**。由于 z 方向的振动是横向的,弹簧大部分张力在 x 方向,z 方向的有效刚度是非线性的——等效于一个三次方恢复力(FPU 型非线性),因此波速较慢。
## 使用方法
```bash
cd examples/case06
python run_dynamics.py
```
配置参数详见 `input/input.txt`,驱动力定义见 `input/driver.txt`,完整文档见 `doc/index.html`
+477
View File
@@ -0,0 +1,477 @@
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>case06 — 一维原子链驱动力学模拟 | 物理原理 &amp; 使用文档</title>
<style>
:root {
--bg: #f8f9fa;
--card: #fff;
--text: #1a1a2e;
--accent: #2563eb;
--accent-light: #dbeafe;
--code-bg: #1e293b;
--code-text: #e2e8f0;
--border: #e2e8f0;
--muted: #64748b;
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Noto Sans SC", sans-serif;
background: var(--bg);
color: var(--text);
line-height: 1.7;
}
/* ── Header ── */
.hero {
background: linear-gradient(135deg, #1e293b 0%, #334155 100%);
color: #fff;
padding: 56px 24px 48px;
text-align: center;
}
.hero h1 { font-size: 2rem; font-weight: 700; letter-spacing: -0.02em; }
.hero .subtitle {
margin-top: 10px;
font-size: 1.05rem;
opacity: 0.8;
}
.hero .badge {
display: inline-block;
margin-top: 14px;
padding: 4px 14px;
border-radius: 999px;
background: rgba(255,255,255,0.12);
font-size: 0.82rem;
}
/* ── Layout ── */
.container { max-width: 820px; margin: 0 auto; padding: 32px 20px; }
section { margin-bottom: 44px; }
h2 {
font-size: 1.35rem;
font-weight: 600;
margin-bottom: 16px;
padding-bottom: 8px;
border-bottom: 2px solid var(--accent);
display: inline-block;
}
h3 {
font-size: 1.05rem;
font-weight: 600;
margin: 20px 0 10px;
}
p, li { margin-bottom: 10px; }
ul, ol { padding-left: 22px; }
strong { color: var(--accent); }
/* ── Cards ── */
.card {
background: var(--card);
border-radius: 12px;
padding: 20px 24px;
margin-bottom: 16px;
border: 1px solid var(--border);
box-shadow: 0 1px 3px rgba(0,0,0,0.04);
}
/* ── Formula / Code blocks ── */
.formula {
background: var(--card);
border-left: 4px solid var(--accent);
padding: 14px 20px;
margin: 14px 0;
font-family: "Times New Roman", "STIX", serif;
font-size: 1.05rem;
overflow-x: auto;
border-radius: 0 8px 8px 0;
}
code {
background: var(--accent-light);
padding: 2px 7px;
border-radius: 4px;
font-family: "JetBrains Mono", "Fira Code", monospace;
font-size: 0.88em;
}
pre {
background: var(--code-bg);
color: var(--code-text);
padding: 16px 20px;
border-radius: 10px;
overflow-x: auto;
font-size: 0.85rem;
line-height: 1.5;
margin: 14px 0;
}
pre .cm { color: #94a3b8; font-style: italic; } /* comment */
/* ── Table ── */
table {
width: 100%;
border-collapse: collapse;
margin: 14px 0;
font-size: 0.92rem;
}
th, td {
padding: 8px 12px;
text-align: left;
border-bottom: 1px solid var(--border);
}
th { background: var(--accent-light); font-weight: 600; }
/* ── TOC ── */
.toc { counter-reset: toc; }
.toc li { counter-increment: toc; list-style: none; margin-bottom: 6px; }
.toc li::before { content: counter(toc) ". "; font-weight: 600; color: var(--accent); }
.toc a { color: var(--accent); text-decoration: none; }
.toc a:hover { text-decoration: underline; }
/* ── Flow diagram ── */
.flow { display: flex; flex-wrap: wrap; gap: 8px; align-items: center; justify-content: center; margin: 16px 0; }
.flow-step {
background: var(--accent-light);
border: 1px solid var(--accent);
border-radius: 8px;
padding: 8px 16px;
font-size: 0.88rem;
font-weight: 500;
}
.flow-arrow { color: var(--muted); font-size: 1.2rem; }
@media (max-width: 600px) {
.hero h1 { font-size: 1.5rem; }
.flow { flex-direction: column; }
.flow-arrow { transform: rotate(90deg); }
}
</style>
</head>
<body>
<!-- ============================================================ -->
<!-- Header -->
<!-- ============================================================ -->
<header class="hero">
<h1>一维原子链驱动力学模拟</h1>
<p class="subtitle">120 个原子沿 x 轴排列 · 弹簧连接 · z 方向受迫振动</p>
<span class="badge">case06 · examples/case06</span>
</header>
<div class="container">
<!-- ============================================================ -->
<!-- TOC -->
<!-- ============================================================ -->
<section>
<h2>目录</h2>
<ol class="toc">
<li><a href="#physics">物理原理</a></li>
<li><a href="#algorithm">数值算法</a></li>
<li><a href="#driver">驱动力模型</a></li>
<li><a href="#usage">使用方法</a></li>
<li><a href="#params">参数参考</a></li>
<li><a href="#files">文件结构</a></li>
<li><a href="#troubleshoot">常见问题</a></li>
</ol>
</section>
<!-- ============================================================ -->
<!-- 1. Physics -->
<!-- ============================================================ -->
<section id="physics">
<h2>一、物理原理</h2>
<div class="card">
<h3>1.1 一维原子链</h3>
<p>120 个原子沿 <strong>x 轴</strong> 等间距排列,原子间距为 1。相邻原子之间用 <strong>理想弹簧</strong> 连接,弹簧的劲度系数 <em>k</em> = 1.0,原长 <em>L</em>₀ = 1.0(与原子间距一致,初始状态弹簧无拉伸)。</p>
<p>每个原子被限制在 <strong>z 方向</strong> 自由振动,x 和 y 方向锁定(<code>fix_x=1, fix_y=1, fix_z=0</code>)。</p>
</div>
<div class="card">
<h3>1.2 弹簧力(胡克定律)</h3>
<p>当原子 <em>i</em><em>j</em> 之间有弹簧连接时,原子 <em>i</em> 受到的弹簧力为:</p>
<div class="formula">
<strong>F</strong> = <em>k</em> · (<em>d</em> <em>L</em>₀) · <strong>u</strong><sub><em>ij</em></sub>
</div>
<p>其中 <em>d</em> = |<strong>r</strong><sub><em>j</em></sub> <strong>r</strong><sub><em>i</em></sub>| 为两原子间距离,<strong>u</strong><sub><em>ij</em></sub> 为从 <em>i</em> 指向 <em>j</em> 的单位向量。由于原子只在 z 方向振动,弹簧在 z 方向的分量是 <strong>几何非线性</strong> 的——对于小振幅近似,z 方向等效于一个三次方恢复力(FPU 型非线性)。</p>
</div>
<div class="card">
<h3>1.3 运动方程</h3>
<p>对于第 <em>i</em> 个自由原子(非受驱),牛顿第二定律给出:</p>
<div class="formula">
<em>m</em> · <strong>a</strong><sub><em>i</em></sub> = <strong>F</strong><sub><em>i</em></sub><sup>spring</sup> + <strong>F</strong><sub><em>i</em></sub><sup>driving</sup>
</div>
<p>本案例中 <strong>唯一的外力</strong> 来自驱动力(仅施加于原子 1)。无重力、无万有引力、无阻尼,系统总能量守恒。</p>
</div>
<div class="card">
<h3>1.4 波传播</h3>
<p>原子 1 的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的 <strong>横波</strong>。由于横向振动的几何非线性(弹簧大部分张力在 x 方向,z 方向的有效刚度远小于 1),波的传播速度较慢,且高阶频率成分会在链中产生复杂的非线性动力学行为(类似 FPU 回波现象)。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 2. Algorithm -->
<!-- ============================================================ -->
<section id="algorithm">
<h2>二、数值算法</h2>
<div class="card">
<h3>2.1 蛙跳法(Leapfrog / Velocity-Verlet</h3>
<p>采用能量守恒特性优异的 <strong>蛙跳法</strong>(二阶辛积分器),更新公式为:</p>
<div class="formula">
<strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) = <strong>v</strong>(<em>t</em>) + ½ <strong>a</strong>(<em>t</em>) · Δ<em>t</em><br>
<strong>r</strong>(<em>t</em> + Δ<em>t</em>) = <strong>r</strong>(<em>t</em>) + <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) · Δ<em>t</em><br>
<strong>a</strong>(<em>t</em> + Δ<em>t</em>) = <strong>F</strong>(<strong>r</strong>(<em>t</em> + Δ<em>t</em>), <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>)) / <em>m</em><br>
<strong>v</strong>(<em>t</em> + Δ<em>t</em>) = <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) + ½ <strong>a</strong>(<em>t</em> + Δ<em>t</em>) · Δ<em>t</em>
</div>
<p>蛙跳法在长时间模拟中能量漂移极小(本案例验证 <strong>&lt; 0.004%</strong>),适合无阻尼的保守系统。</p>
</div>
<div class="card">
<h3>2.2 时间步长与采样</h3>
<table>
<tr><th>参数</th><th></th><th>说明</th></tr>
<tr><td>DT</td><td>0.01 s</td><td>积分步长(远小于 1/ω ≈ 0.16 s,满足稳定性条件)</td></tr>
<tr><td>T_total</td><td>100 s</td><td>总模拟时间 → NT = 10000 步</td></tr>
<tr><td>NSTEP</td><td>50</td><td>每 NSTEP 步取一帧用于动画 → 200 帧</td></tr>
<tr><td>method</td><td>leapfrog</td><td>蛙跳法(Velocity-Verlet</td></tr>
</table>
</div>
<div class="card">
<h3>2.3 计算流程</h3>
<div class="flow">
<span class="flow-step">读入 coord.txt<br>connection.txt<br>bond.txt</span>
<span class="flow-arrow"></span>
<span class="flow-step">施加驱动力<br>(驱动原子 1</span>
<span class="flow-arrow"></span>
<span class="flow-step">记录轨迹</span>
<span class="flow-arrow"></span>
<span class="flow-step">蛙跳法<br>更新位置/速度</span>
<span class="flow-arrow"></span>
<span class="flow-step">固定约束<br>x, y 锁定)</span>
<span class="flow-arrow"></span>
<span class="flow-step" style="background:#fef3c7;border-color:#f59e0b;">循环<br>NT 次</span>
</div>
<p style="margin-top:12px;">注意:驱动力在 <strong>每次积分前</strong> 施加,确保受驱原子的位置正确传递给弹簧力计算。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 3. Driving Force -->
<!-- ============================================================ -->
<section id="driver">
<h2>三、驱动力模型</h2>
<div class="card">
<h3>3.1 定义文件</h3>
<p>驱动力由 <code>input/driver.txt</code> 定义,格式如下:</p>
<pre>n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0 0 5 0 0 1 0 0 90 all</pre>
</div>
<div class="card">
<h3>3.2 数学公式</h3>
<p>受驱原子的位置由下式决定(<strong>完全替换</strong> coord.txt 中的初始坐标和固定约束):</p>
<div class="formula">
<strong>r</strong>(<em>t</em>) = <strong>A</strong> · cos(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>速度由解析导数给出:</p>
<div class="formula">
<strong>v</strong>(<em>t</em>) = <strong>A</strong> · 2π<em>f</em> · sin(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>其中 <strong>A</strong> = (amp_x, amp_y, amp_z)<strong>f</strong> = (freq_x, freq_y, freq_z) 为不同方向的驱动频率,<strong>φ</strong> = (phi_x, phi_y, phi_z) 为相位(<strong>角度制</strong>,代码自动转换为弧度)。</p>
</div>
<div class="card">
<h3>3.3 本案例驱动参数</h3>
<table>
<tr><th>参数</th><th></th><th>含义</th></tr>
<tr><td>amp_z</td><td>5.0</td><td>z 方向驱动振幅</td></tr>
<tr><td>freq_z</td><td>1.0 Hz</td><td>驱动频率(周期 1 s</td></tr>
<tr><td>phi_z</td><td>90°</td><td>驱动相位 → z(0) = 5·cos(90°) = 0</td></tr>
<tr><td>period</td><td>all</td><td>全程驱动,永不停止</td></tr>
</table>
<div class="formula">
<em>z</em>(<em>t</em>) = 5.0 · cos(2π · 1.0 · <em>t</em> + 90°)
</div>
</div>
<div class="card">
<h3>3.4 有限周期驱动</h3>
<p><code>period</code> 参数支持三种模式:</p>
<ul>
<li><strong>all</strong> — 全程驱动</li>
<li><strong>数值</strong> — 驱动指定周期数后 <strong>静止</strong>(冻结在最终位置,速度归零)。例如 <code>period: 1</code> 表示驱动 1 个完整周期后停止。</li>
</ul>
</div>
<div class="card">
<h3>3.5 驱动与固定约束的关系</h3>
<p>对于受驱原子(<code>driver.txt</code><code>n</code> 指定的原子),其在 <code>coord.txt</code> 中的初始坐标和 <code>fix_x/fix_y/fix_z</code> 约束被 <strong>完全忽略</strong>。原子的位置和速度完全由驱动力公式决定。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 4. Usage -->
<!-- ============================================================ -->
<section id="usage">
<h2>四、使用方法</h2>
<div class="card">
<h3>4.1 完整运行(模拟 + 动画)</h3>
<pre>cd examples/case06
python run_dynamics.py</pre>
<p>这步会依次执行:物理模拟 → 抽帧 → 打开 VisPy 3D 动画窗口。</p>
</div>
<div class="card">
<h3>4.2 仅查看已有结果</h3>
<p>如果已经跑完模拟且生成了 <code>output/display.txt</code>,可以通过修改 <code>input.txt</code> 跳过计算,只开动画:</p>
<pre>step_simulate: 0 # 跳过模拟
step_sample: 0 # 跳过抽帧
step_animation: 1 # 播放动画</pre>
<p>然后运行:<code>python run_dynamics.py</code></p>
</div>
<div class="card">
<h3>4.3 手动 3D 动画</h3>
<p>也可以单独启动 VisPy 窗口:</p>
<pre>python ../../draw.py output/</pre>
</div>
<div class="card">
<h3>4.4 强制重新计算</h3>
<p>修改参数后需要重新运行模拟时,设置:</p>
<pre>force_calc: 1 # 忽略缓存,强制重新计算</pre>
</div>
<div class="card">
<h3>4.5 动画交互</h3>
<table>
<tr><th>操作</th><th>效果</th></tr>
<tr><td>鼠标拖动</td><td>旋转视角</td></tr>
<tr><td>滚轮</td><td>缩放</td></tr>
<tr><td>W / S 键</td><td>相机沿 Z 轴向前 / 向后移动(靠近/远离场景)</td></tr>
<tr><td>A / D 键</td><td>视角向右 / 向左平移</td></tr>
<tr><td>E / Q 键</td><td>视角上升 / 下降(屏幕方向)</td></tr>
<tr><td>C / X 键</td><td>增大 / 减小步长</td></tr>
<tr><td>V 键</td><td>切换透视 / 正交投影</td></tr>
<tr><td>左上角 <strong>reset</strong> 按钮</td><td>复位视角到初始位置</td></tr>
<tr><td>左上角 <strong>info</strong> 按钮</td><td>切换信息面板显示/隐藏</td></tr>
<tr><td>左上角 <strong>axes</strong> 按钮</td><td>切换坐标轴显示/隐藏</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 5. Parameters -->
<!-- ============================================================ -->
<section id="params">
<h2>五、参数参考</h2>
<div class="card">
<h3>5.1 input.txt 关键参数</h3>
<table>
<tr><th>参数</th><th>默认值</th><th>说明</th></tr>
<tr><td>gravity_field</td><td>0</td><td>均匀重力场(已关闭)</td></tr>
<tr><td>gravity_interaction</td><td>0</td><td>原子间万有引力(已关闭)</td></tr>
<tr><td>elastic_force</td><td>1</td><td>弹簧键力(已开启)</td></tr>
<tr><td>damping_force</td><td>0</td><td>阻尼(已关闭)</td></tr>
<tr><td><strong>driving_force</strong></td><td><strong>1</strong></td><td>驱动力开关(1=开启,需 driver.txt</td></tr>
<tr><td>method</td><td>leapfrog</td><td>数值积分方法</td></tr>
<tr><td>DT</td><td>0.01</td><td>积分步长 (s)</td></tr>
<tr><td>T_total</td><td>100.0</td><td>总模拟时间 (s)</td></tr>
<tr><td>NSTEP</td><td>50</td><td>抽帧步数间隔</td></tr>
<tr><td>engine</td><td>python</td><td>计算引擎(python / c / cpp / fortran</td></tr>
<tr><td>use_marker</td><td>1</td><td>渲染模式(0=Sphere 网格, 1=Marker GPU 实例化)</td></tr>
</table>
</div>
<div class="card">
<h3>5.2 流程控制参数</h3>
<table>
<tr><th>参数</th><th>0</th><th>1</th></tr>
<tr><td>step_simulate</td><td>跳过模拟(加载已有轨迹)</td><td>运行物理模拟</td></tr>
<tr><td>step_sample</td><td>跳过抽帧</td><td>从轨迹抽取显示帧</td></tr>
<tr><td>step_plot</td><td>不生成图表</td><td>生成轨迹/能量图</td></tr>
<tr><td><strong>step_plot_wave</strong></td><td>不生成波形图</td><td>生成波形能量动画 GIF</td></tr>
<tr><td>step_animation</td><td>不启动动画</td><td>自动打开 VisPy 3D 窗口</td></tr>
<tr><td>force_calc</td><td>自动检测缓存</td><td>强制重新计算</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 6. File Structure -->
<!-- ============================================================ -->
<section id="files">
<h2>六、文件结构</h2>
<pre>case06/
├── input/
│ ├── input.txt # 主配置文件(YAML 格式)
│ ├── coord.txt # 原子坐标(120 个原子)
│ ├── connection.txt # 弹簧连接关系(59 条键)
│ ├── bond.txt # 弹簧参数(k=1.0, L₀=1.0
│ └── <strong>driver.txt</strong> # <span class="cm">驱动力定义(本案例新增)</span>
├── output/
│ ├── trajectory.txt # 全量轨迹数据(50000 步 × 120 原子)
│ ├── display.txt # 抽帧后的动画数据(500 帧 × 120 原子)
│ ├── dynamics.log # 计算日志
│ ├── animation.log # 动画启动日志(闪退时排查用)
│ └── wave_animation.gif # 波形能量动画(step_plot_wave=1 时生成)
├── doc/
│ └── index.html # <span class="cm">本文档</span>
├── Readme.md # 案例简介
└── run_dynamics.py # 案例运行入口</pre>
</section>
<!-- ============================================================ -->
<!-- 7. Troubleshooting -->
<!-- ============================================================ -->
<section id="troubleshoot">
<h2>七、常见问题</h2>
<div class="card">
<h3>7.1 动画窗口闪退</h3>
<p>如果 VisPy 窗口一闪就消失,请检查:</p>
<ul>
<li><code>output/animation.log</code> 中是否有错误信息</li>
<li><code>output/display.txt</code> 是否存在(需先跑 <code>step_sample: 1</code></li>
</ul>
</div>
<div class="card">
<h3>7.2 原子不振动</h3>
<p>可能原因:</p>
<ul>
<li><strong>NSTEP 过大</strong>:抽帧间隔大于驱动周期的一半时,动画会丢失振动细节。建议 NSTEP ≤ 1/(freq · DT · 10)</li>
<li><strong>相位 φ 使采样点落在零值</strong>:试试 <code>phi_z: 0</code> 让原子在 t=0 处于振幅峰值</li>
<li>确认 <code>driving_force: 1</code><code>driver.txt</code> 中 amp_z 不为 0</li>
</ul>
</div>
<div class="card">
<h3>7.3 渲染性能慢</h3>
<p>原子数多时动画卡顿:</p>
<ul>
<li>设置 <code>use_marker: 1</code>(使用 GPU 实例化渲染替代独立网格球体)</li>
<li>增大 <code>NSTEP</code> 减少动画帧数</li>
</ul>
</div>
</section>
<hr style="border:none;border-top:1px solid var(--border);margin:40px 0;">
<footer style="text-align:center;color:var(--muted);font-size:0.85rem;margin-bottom:40px;">
Dynamics Simulation Framework &nbsp;·&nbsp; 生成于 2026-06-10
</footer>
</div>
</body>
</html>
+2
View File
@@ -0,0 +1,2 @@
bond_name k rest_length
h 100.0 1.0
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
+102
View File
@@ -0,0 +1,102 @@
n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0.5 0 0 0.05 0 0 90 90 90 all
102 0.5 0 0 0.05 0 0 90 90 90 all
203 0.5 0 0 0.05 0 0 90 90 90 all
304 0.5 0 0 0.05 0 0 90 90 90 all
405 0.5 0 0 0.05 0 0 90 90 90 all
506 0.5 0 0 0.05 0 0 90 90 90 all
607 0.5 0 0 0.05 0 0 90 90 90 all
708 0.5 0 0 0.05 0 0 90 90 90 all
809 0.5 0 0 0.05 0 0 90 90 90 all
910 0.5 0 0 0.05 0 0 90 90 90 all
1011 0.5 0 0 0.05 0 0 90 90 90 all
1112 0.5 0 0 0.05 0 0 90 90 90 all
1213 0.5 0 0 0.05 0 0 90 90 90 all
1314 0.5 0 0 0.05 0 0 90 90 90 all
1415 0.5 0 0 0.05 0 0 90 90 90 all
1516 0.5 0 0 0.05 0 0 90 90 90 all
1617 0.5 0 0 0.05 0 0 90 90 90 all
1718 0.5 0 0 0.05 0 0 90 90 90 all
1819 0.5 0 0 0.05 0 0 90 90 90 all
1920 0.5 0 0 0.05 0 0 90 90 90 all
2021 0.5 0 0 0.05 0 0 90 90 90 all
2122 0.5 0 0 0.05 0 0 90 90 90 all
2223 0.5 0 0 0.05 0 0 90 90 90 all
2324 0.5 0 0 0.05 0 0 90 90 90 all
2425 0.5 0 0 0.05 0 0 90 90 90 all
2526 0.5 0 0 0.05 0 0 90 90 90 all
2627 0.5 0 0 0.05 0 0 90 90 90 all
2728 0.5 0 0 0.05 0 0 90 90 90 all
2829 0.5 0 0 0.05 0 0 90 90 90 all
2930 0.5 0 0 0.05 0 0 90 90 90 all
3031 0.5 0 0 0.05 0 0 90 90 90 all
3132 0.5 0 0 0.05 0 0 90 90 90 all
3233 0.5 0 0 0.05 0 0 90 90 90 all
3334 0.5 0 0 0.05 0 0 90 90 90 all
3435 0.5 0 0 0.05 0 0 90 90 90 all
3536 0.5 0 0 0.05 0 0 90 90 90 all
3637 0.5 0 0 0.05 0 0 90 90 90 all
3738 0.5 0 0 0.05 0 0 90 90 90 all
3839 0.5 0 0 0.05 0 0 90 90 90 all
3940 0.5 0 0 0.05 0 0 90 90 90 all
4041 0.5 0 0 0.05 0 0 90 90 90 all
4142 0.5 0 0 0.05 0 0 90 90 90 all
4243 0.5 0 0 0.05 0 0 90 90 90 all
4344 0.5 0 0 0.05 0 0 90 90 90 all
4445 0.5 0 0 0.05 0 0 90 90 90 all
4546 0.5 0 0 0.05 0 0 90 90 90 all
4647 0.5 0 0 0.05 0 0 90 90 90 all
4748 0.5 0 0 0.05 0 0 90 90 90 all
4849 0.5 0 0 0.05 0 0 90 90 90 all
4950 0.5 0 0 0.05 0 0 90 90 90 all
5051 0.5 0 0 0.05 0 0 90 90 90 all
5152 0.5 0 0 0.05 0 0 90 90 90 all
5253 0.5 0 0 0.05 0 0 90 90 90 all
5354 0.5 0 0 0.05 0 0 90 90 90 all
5455 0.5 0 0 0.05 0 0 90 90 90 all
5556 0.5 0 0 0.05 0 0 90 90 90 all
5657 0.5 0 0 0.05 0 0 90 90 90 all
5758 0.5 0 0 0.05 0 0 90 90 90 all
5859 0.5 0 0 0.05 0 0 90 90 90 all
5960 0.5 0 0 0.05 0 0 90 90 90 all
6061 0.5 0 0 0.05 0 0 90 90 90 all
6162 0.5 0 0 0.05 0 0 90 90 90 all
6263 0.5 0 0 0.05 0 0 90 90 90 all
6364 0.5 0 0 0.05 0 0 90 90 90 all
6465 0.5 0 0 0.05 0 0 90 90 90 all
6566 0.5 0 0 0.05 0 0 90 90 90 all
6667 0.5 0 0 0.05 0 0 90 90 90 all
6768 0.5 0 0 0.05 0 0 90 90 90 all
6869 0.5 0 0 0.05 0 0 90 90 90 all
6970 0.5 0 0 0.05 0 0 90 90 90 all
7071 0.5 0 0 0.05 0 0 90 90 90 all
7172 0.5 0 0 0.05 0 0 90 90 90 all
7273 0.5 0 0 0.05 0 0 90 90 90 all
7374 0.5 0 0 0.05 0 0 90 90 90 all
7475 0.5 0 0 0.05 0 0 90 90 90 all
7576 0.5 0 0 0.05 0 0 90 90 90 all
7677 0.5 0 0 0.05 0 0 90 90 90 all
7778 0.5 0 0 0.05 0 0 90 90 90 all
7879 0.5 0 0 0.05 0 0 90 90 90 all
7980 0.5 0 0 0.05 0 0 90 90 90 all
8081 0.5 0 0 0.05 0 0 90 90 90 all
8182 0.5 0 0 0.05 0 0 90 90 90 all
8283 0.5 0 0 0.05 0 0 90 90 90 all
8384 0.5 0 0 0.05 0 0 90 90 90 all
8485 0.5 0 0 0.05 0 0 90 90 90 all
8586 0.5 0 0 0.05 0 0 90 90 90 all
8687 0.5 0 0 0.05 0 0 90 90 90 all
8788 0.5 0 0 0.05 0 0 90 90 90 all
8889 0.5 0 0 0.05 0 0 90 90 90 all
8990 0.5 0 0 0.05 0 0 90 90 90 all
9091 0.5 0 0 0.05 0 0 90 90 90 all
9192 0.5 0 0 0.05 0 0 90 90 90 all
9293 0.5 0 0 0.05 0 0 90 90 90 all
9394 0.5 0 0 0.05 0 0 90 90 90 all
9495 0.5 0 0 0.05 0 0 90 90 90 all
9596 0.5 0 0 0.05 0 0 90 90 90 all
9697 0.5 0 0 0.05 0 0 90 90 90 all
9798 0.5 0 0 0.05 0 0 90 90 90 all
9899 0.5 0 0 0.05 0 0 90 90 90 all
10000 0.5 0 0 0.05 0 0 90 90 90 all
10101 0.5 0 0 0.05 0 0 90 90 90 all
+73
View File
@@ -0,0 +1,73 @@
# 物理模拟参数配置
# case15 — 双缝干涉(x方向振动)
# 所有粒子 y/z 固定,仅 x 方向振动
# 左边界 x 方向驱动 → 双缝势垒 → x方向波干涉
step_simulate: 1
step_sample: 0
step_plot: 0
step_animation: 1
step_plot_wave: 0
force_calc: 1
save_trajectory: 0
engine: c
box_a: 120.0
coord_file: input/coord.txt
connection_file: input/connection.txt
bond_file: input/bond.txt
driver_file: input/driver.txt
plot_atom: 51
G: [0.000, 0.000, 0.000]
B: [0.000, 0.000, 0.000]
gravity_field: 0
gravity_interaction: 0
elastic_force: 1
damping_force: 0
driving_force: 1
gravity_strength: 1.0
method: leapfrog
warmup_steps: 0
T_total: 100.0
NSTEP: 100
DT: 0.001
sample_start: null
sample_end: null
use_marker: 1
display_color: {
x : [1, [255, 0, 0]],
y : [0, [ 0, 255, 0]],
z : [0, [ 0, 0, 255]],
xy : [0, [255, 255, 0]],
yz : [0, [ 0, 255, 255]],
zx : [0, [255, 0, 255]],
xyz : [0, [255, 255, 255]],
}
alpha: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
ball_color_r: 0.90
ball_color_g: 0.20
ball_color_b: 0.20
box_color_r: 0.80
box_color_g: 0.80
box_color_b: 0.85
camera_distance: 120.0
camera_elevation: 60.0
camera_azimuth: -45.0
camera_center_x: 0.0
camera_center_y: 0.0
camera_center_z: 0.0
move_camera: 0
display_amp: [1.0, 1.0, 1.0]
+2
View File
@@ -0,0 +1,2 @@
0 0 50
0 0 80
+54
View File
@@ -0,0 +1,54 @@
"""
Case runner for Dynamics case15 2D grid (61x61 atomic mesh).
This script keeps program and data separated:
- program: ../../dynamics.py
- input: ./input
- output: ./output
"""
from __future__ import annotations
import argparse
import importlib.util
from pathlib import Path
CASE_DIR = Path(__file__).resolve().parent
DYNAMICS_PATH = Path("..") / ".." / "dynamics.py"
INPUT_DIR = Path("input")
OUTPUT_DIR = Path("output")
CONFIG_FILE = INPUT_DIR / "input.txt"
def load_dynamics_module(module_path: Path):
spec = importlib.util.spec_from_file_location("dynamics_module", module_path)
if spec is None or spec.loader is None:
raise ImportError(f"无法加载 dynamics.py: {module_path}")
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
return module
def main():
parser = argparse.ArgumentParser(description="运行 Dynamics 示例案例 case15")
parser.add_argument("--no-plot", action="store_true", help="跳过 matplotlib 绘图")
args = parser.parse_args()
dynamics_path = (CASE_DIR / DYNAMICS_PATH).resolve()
input_dir = (CASE_DIR / INPUT_DIR).resolve()
output_dir = (CASE_DIR / OUTPUT_DIR).resolve()
config_path = (CASE_DIR / CONFIG_FILE).resolve()
module = load_dynamics_module(dynamics_path)
module.run_case(
config_path=config_path,
runtime_base=CASE_DIR,
input_dir=input_dir,
output_dir=output_dir,
no_plot=args.no_plot,
)
if __name__ == "__main__":
main()
+40
View File
@@ -0,0 +1,40 @@
# case06: 一维原子链横波模拟
60 个原子沿 x 轴排列,相邻原子用弹簧连接。原子 1 受 z 方向驱动力作用,产生沿链传播的横波。
## 物理设定
| 参数 | 值 |
|---|---|
| 原子数 | 120 |
| 排列 | 沿 x 轴等间距排列,间距为 1 |
| 约束 | 原子**沿 z 方向自由振动**fix_x=1, fix_y=1, fix_z=0),x, y 锁定 |
| 弹簧 | 劲度系数 k=1.0,原长 L₀=1.0 |
| 重力 | 无 |
| 万有引力 | 无 |
| 阻尼 | 无 |
| 驱动力 | 原子 1(z 方向驱动) |
| 算法 | leapfrog(蛙跳法,能量守恒) |
## 驱动力
原子 1 的位置由 `input/driver.txt` 中的驱动力公式决定:
```math
z(t) = A_z \cdot \cos(2\pi f_z t + \phi_z)
```
当前参数:A_z = 0.5, f_z = 0.1 Hz, φ_z = 90°, period = all(全程驱动)。
## 动力学行为
原子 1 沿 z 方向的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的**横波**。由于 z 方向的振动是横向的,弹簧大部分张力在 x 方向,z 方向的有效刚度是非线性的——等效于一个三次方恢复力(FPU 型非线性),因此波速较慢。
## 使用方法
```bash
cd examples/case06
python run_dynamics.py
```
配置参数详见 `input/input.txt`,驱动力定义见 `input/driver.txt`,完整文档见 `doc/index.html`
+477
View File
@@ -0,0 +1,477 @@
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>case06 — 一维原子链驱动力学模拟 | 物理原理 &amp; 使用文档</title>
<style>
:root {
--bg: #f8f9fa;
--card: #fff;
--text: #1a1a2e;
--accent: #2563eb;
--accent-light: #dbeafe;
--code-bg: #1e293b;
--code-text: #e2e8f0;
--border: #e2e8f0;
--muted: #64748b;
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Noto Sans SC", sans-serif;
background: var(--bg);
color: var(--text);
line-height: 1.7;
}
/* ── Header ── */
.hero {
background: linear-gradient(135deg, #1e293b 0%, #334155 100%);
color: #fff;
padding: 56px 24px 48px;
text-align: center;
}
.hero h1 { font-size: 2rem; font-weight: 700; letter-spacing: -0.02em; }
.hero .subtitle {
margin-top: 10px;
font-size: 1.05rem;
opacity: 0.8;
}
.hero .badge {
display: inline-block;
margin-top: 14px;
padding: 4px 14px;
border-radius: 999px;
background: rgba(255,255,255,0.12);
font-size: 0.82rem;
}
/* ── Layout ── */
.container { max-width: 820px; margin: 0 auto; padding: 32px 20px; }
section { margin-bottom: 44px; }
h2 {
font-size: 1.35rem;
font-weight: 600;
margin-bottom: 16px;
padding-bottom: 8px;
border-bottom: 2px solid var(--accent);
display: inline-block;
}
h3 {
font-size: 1.05rem;
font-weight: 600;
margin: 20px 0 10px;
}
p, li { margin-bottom: 10px; }
ul, ol { padding-left: 22px; }
strong { color: var(--accent); }
/* ── Cards ── */
.card {
background: var(--card);
border-radius: 12px;
padding: 20px 24px;
margin-bottom: 16px;
border: 1px solid var(--border);
box-shadow: 0 1px 3px rgba(0,0,0,0.04);
}
/* ── Formula / Code blocks ── */
.formula {
background: var(--card);
border-left: 4px solid var(--accent);
padding: 14px 20px;
margin: 14px 0;
font-family: "Times New Roman", "STIX", serif;
font-size: 1.05rem;
overflow-x: auto;
border-radius: 0 8px 8px 0;
}
code {
background: var(--accent-light);
padding: 2px 7px;
border-radius: 4px;
font-family: "JetBrains Mono", "Fira Code", monospace;
font-size: 0.88em;
}
pre {
background: var(--code-bg);
color: var(--code-text);
padding: 16px 20px;
border-radius: 10px;
overflow-x: auto;
font-size: 0.85rem;
line-height: 1.5;
margin: 14px 0;
}
pre .cm { color: #94a3b8; font-style: italic; } /* comment */
/* ── Table ── */
table {
width: 100%;
border-collapse: collapse;
margin: 14px 0;
font-size: 0.92rem;
}
th, td {
padding: 8px 12px;
text-align: left;
border-bottom: 1px solid var(--border);
}
th { background: var(--accent-light); font-weight: 600; }
/* ── TOC ── */
.toc { counter-reset: toc; }
.toc li { counter-increment: toc; list-style: none; margin-bottom: 6px; }
.toc li::before { content: counter(toc) ". "; font-weight: 600; color: var(--accent); }
.toc a { color: var(--accent); text-decoration: none; }
.toc a:hover { text-decoration: underline; }
/* ── Flow diagram ── */
.flow { display: flex; flex-wrap: wrap; gap: 8px; align-items: center; justify-content: center; margin: 16px 0; }
.flow-step {
background: var(--accent-light);
border: 1px solid var(--accent);
border-radius: 8px;
padding: 8px 16px;
font-size: 0.88rem;
font-weight: 500;
}
.flow-arrow { color: var(--muted); font-size: 1.2rem; }
@media (max-width: 600px) {
.hero h1 { font-size: 1.5rem; }
.flow { flex-direction: column; }
.flow-arrow { transform: rotate(90deg); }
}
</style>
</head>
<body>
<!-- ============================================================ -->
<!-- Header -->
<!-- ============================================================ -->
<header class="hero">
<h1>一维原子链驱动力学模拟</h1>
<p class="subtitle">120 个原子沿 x 轴排列 · 弹簧连接 · z 方向受迫振动</p>
<span class="badge">case06 · examples/case06</span>
</header>
<div class="container">
<!-- ============================================================ -->
<!-- TOC -->
<!-- ============================================================ -->
<section>
<h2>目录</h2>
<ol class="toc">
<li><a href="#physics">物理原理</a></li>
<li><a href="#algorithm">数值算法</a></li>
<li><a href="#driver">驱动力模型</a></li>
<li><a href="#usage">使用方法</a></li>
<li><a href="#params">参数参考</a></li>
<li><a href="#files">文件结构</a></li>
<li><a href="#troubleshoot">常见问题</a></li>
</ol>
</section>
<!-- ============================================================ -->
<!-- 1. Physics -->
<!-- ============================================================ -->
<section id="physics">
<h2>一、物理原理</h2>
<div class="card">
<h3>1.1 一维原子链</h3>
<p>120 个原子沿 <strong>x 轴</strong> 等间距排列,原子间距为 1。相邻原子之间用 <strong>理想弹簧</strong> 连接,弹簧的劲度系数 <em>k</em> = 1.0,原长 <em>L</em>₀ = 1.0(与原子间距一致,初始状态弹簧无拉伸)。</p>
<p>每个原子被限制在 <strong>z 方向</strong> 自由振动,x 和 y 方向锁定(<code>fix_x=1, fix_y=1, fix_z=0</code>)。</p>
</div>
<div class="card">
<h3>1.2 弹簧力(胡克定律)</h3>
<p>当原子 <em>i</em><em>j</em> 之间有弹簧连接时,原子 <em>i</em> 受到的弹簧力为:</p>
<div class="formula">
<strong>F</strong> = <em>k</em> · (<em>d</em> <em>L</em>₀) · <strong>u</strong><sub><em>ij</em></sub>
</div>
<p>其中 <em>d</em> = |<strong>r</strong><sub><em>j</em></sub> <strong>r</strong><sub><em>i</em></sub>| 为两原子间距离,<strong>u</strong><sub><em>ij</em></sub> 为从 <em>i</em> 指向 <em>j</em> 的单位向量。由于原子只在 z 方向振动,弹簧在 z 方向的分量是 <strong>几何非线性</strong> 的——对于小振幅近似,z 方向等效于一个三次方恢复力(FPU 型非线性)。</p>
</div>
<div class="card">
<h3>1.3 运动方程</h3>
<p>对于第 <em>i</em> 个自由原子(非受驱),牛顿第二定律给出:</p>
<div class="formula">
<em>m</em> · <strong>a</strong><sub><em>i</em></sub> = <strong>F</strong><sub><em>i</em></sub><sup>spring</sup> + <strong>F</strong><sub><em>i</em></sub><sup>driving</sup>
</div>
<p>本案例中 <strong>唯一的外力</strong> 来自驱动力(仅施加于原子 1)。无重力、无万有引力、无阻尼,系统总能量守恒。</p>
</div>
<div class="card">
<h3>1.4 波传播</h3>
<p>原子 1 的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的 <strong>横波</strong>。由于横向振动的几何非线性(弹簧大部分张力在 x 方向,z 方向的有效刚度远小于 1),波的传播速度较慢,且高阶频率成分会在链中产生复杂的非线性动力学行为(类似 FPU 回波现象)。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 2. Algorithm -->
<!-- ============================================================ -->
<section id="algorithm">
<h2>二、数值算法</h2>
<div class="card">
<h3>2.1 蛙跳法(Leapfrog / Velocity-Verlet</h3>
<p>采用能量守恒特性优异的 <strong>蛙跳法</strong>(二阶辛积分器),更新公式为:</p>
<div class="formula">
<strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) = <strong>v</strong>(<em>t</em>) + ½ <strong>a</strong>(<em>t</em>) · Δ<em>t</em><br>
<strong>r</strong>(<em>t</em> + Δ<em>t</em>) = <strong>r</strong>(<em>t</em>) + <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) · Δ<em>t</em><br>
<strong>a</strong>(<em>t</em> + Δ<em>t</em>) = <strong>F</strong>(<strong>r</strong>(<em>t</em> + Δ<em>t</em>), <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>)) / <em>m</em><br>
<strong>v</strong>(<em>t</em> + Δ<em>t</em>) = <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) + ½ <strong>a</strong>(<em>t</em> + Δ<em>t</em>) · Δ<em>t</em>
</div>
<p>蛙跳法在长时间模拟中能量漂移极小(本案例验证 <strong>&lt; 0.004%</strong>),适合无阻尼的保守系统。</p>
</div>
<div class="card">
<h3>2.2 时间步长与采样</h3>
<table>
<tr><th>参数</th><th></th><th>说明</th></tr>
<tr><td>DT</td><td>0.01 s</td><td>积分步长(远小于 1/ω ≈ 0.16 s,满足稳定性条件)</td></tr>
<tr><td>T_total</td><td>100 s</td><td>总模拟时间 → NT = 10000 步</td></tr>
<tr><td>NSTEP</td><td>50</td><td>每 NSTEP 步取一帧用于动画 → 200 帧</td></tr>
<tr><td>method</td><td>leapfrog</td><td>蛙跳法(Velocity-Verlet</td></tr>
</table>
</div>
<div class="card">
<h3>2.3 计算流程</h3>
<div class="flow">
<span class="flow-step">读入 coord.txt<br>connection.txt<br>bond.txt</span>
<span class="flow-arrow"></span>
<span class="flow-step">施加驱动力<br>(驱动原子 1</span>
<span class="flow-arrow"></span>
<span class="flow-step">记录轨迹</span>
<span class="flow-arrow"></span>
<span class="flow-step">蛙跳法<br>更新位置/速度</span>
<span class="flow-arrow"></span>
<span class="flow-step">固定约束<br>x, y 锁定)</span>
<span class="flow-arrow"></span>
<span class="flow-step" style="background:#fef3c7;border-color:#f59e0b;">循环<br>NT 次</span>
</div>
<p style="margin-top:12px;">注意:驱动力在 <strong>每次积分前</strong> 施加,确保受驱原子的位置正确传递给弹簧力计算。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 3. Driving Force -->
<!-- ============================================================ -->
<section id="driver">
<h2>三、驱动力模型</h2>
<div class="card">
<h3>3.1 定义文件</h3>
<p>驱动力由 <code>input/driver.txt</code> 定义,格式如下:</p>
<pre>n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0 0 5 0 0 1 0 0 90 all</pre>
</div>
<div class="card">
<h3>3.2 数学公式</h3>
<p>受驱原子的位置由下式决定(<strong>完全替换</strong> coord.txt 中的初始坐标和固定约束):</p>
<div class="formula">
<strong>r</strong>(<em>t</em>) = <strong>A</strong> · cos(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>速度由解析导数给出:</p>
<div class="formula">
<strong>v</strong>(<em>t</em>) = <strong>A</strong> · 2π<em>f</em> · sin(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>其中 <strong>A</strong> = (amp_x, amp_y, amp_z)<strong>f</strong> = (freq_x, freq_y, freq_z) 为不同方向的驱动频率,<strong>φ</strong> = (phi_x, phi_y, phi_z) 为相位(<strong>角度制</strong>,代码自动转换为弧度)。</p>
</div>
<div class="card">
<h3>3.3 本案例驱动参数</h3>
<table>
<tr><th>参数</th><th></th><th>含义</th></tr>
<tr><td>amp_z</td><td>5.0</td><td>z 方向驱动振幅</td></tr>
<tr><td>freq_z</td><td>1.0 Hz</td><td>驱动频率(周期 1 s</td></tr>
<tr><td>phi_z</td><td>90°</td><td>驱动相位 → z(0) = 5·cos(90°) = 0</td></tr>
<tr><td>period</td><td>all</td><td>全程驱动,永不停止</td></tr>
</table>
<div class="formula">
<em>z</em>(<em>t</em>) = 5.0 · cos(2π · 1.0 · <em>t</em> + 90°)
</div>
</div>
<div class="card">
<h3>3.4 有限周期驱动</h3>
<p><code>period</code> 参数支持三种模式:</p>
<ul>
<li><strong>all</strong> — 全程驱动</li>
<li><strong>数值</strong> — 驱动指定周期数后 <strong>静止</strong>(冻结在最终位置,速度归零)。例如 <code>period: 1</code> 表示驱动 1 个完整周期后停止。</li>
</ul>
</div>
<div class="card">
<h3>3.5 驱动与固定约束的关系</h3>
<p>对于受驱原子(<code>driver.txt</code><code>n</code> 指定的原子),其在 <code>coord.txt</code> 中的初始坐标和 <code>fix_x/fix_y/fix_z</code> 约束被 <strong>完全忽略</strong>。原子的位置和速度完全由驱动力公式决定。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 4. Usage -->
<!-- ============================================================ -->
<section id="usage">
<h2>四、使用方法</h2>
<div class="card">
<h3>4.1 完整运行(模拟 + 动画)</h3>
<pre>cd examples/case06
python run_dynamics.py</pre>
<p>这步会依次执行:物理模拟 → 抽帧 → 打开 VisPy 3D 动画窗口。</p>
</div>
<div class="card">
<h3>4.2 仅查看已有结果</h3>
<p>如果已经跑完模拟且生成了 <code>output/display.txt</code>,可以通过修改 <code>input.txt</code> 跳过计算,只开动画:</p>
<pre>step_simulate: 0 # 跳过模拟
step_sample: 0 # 跳过抽帧
step_animation: 1 # 播放动画</pre>
<p>然后运行:<code>python run_dynamics.py</code></p>
</div>
<div class="card">
<h3>4.3 手动 3D 动画</h3>
<p>也可以单独启动 VisPy 窗口:</p>
<pre>python ../../draw.py output/</pre>
</div>
<div class="card">
<h3>4.4 强制重新计算</h3>
<p>修改参数后需要重新运行模拟时,设置:</p>
<pre>force_calc: 1 # 忽略缓存,强制重新计算</pre>
</div>
<div class="card">
<h3>4.5 动画交互</h3>
<table>
<tr><th>操作</th><th>效果</th></tr>
<tr><td>鼠标拖动</td><td>旋转视角</td></tr>
<tr><td>滚轮</td><td>缩放</td></tr>
<tr><td>W / S 键</td><td>相机沿 Z 轴向前 / 向后移动(靠近/远离场景)</td></tr>
<tr><td>A / D 键</td><td>视角向右 / 向左平移</td></tr>
<tr><td>E / Q 键</td><td>视角上升 / 下降(屏幕方向)</td></tr>
<tr><td>C / X 键</td><td>增大 / 减小步长</td></tr>
<tr><td>V 键</td><td>切换透视 / 正交投影</td></tr>
<tr><td>左上角 <strong>reset</strong> 按钮</td><td>复位视角到初始位置</td></tr>
<tr><td>左上角 <strong>info</strong> 按钮</td><td>切换信息面板显示/隐藏</td></tr>
<tr><td>左上角 <strong>axes</strong> 按钮</td><td>切换坐标轴显示/隐藏</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 5. Parameters -->
<!-- ============================================================ -->
<section id="params">
<h2>五、参数参考</h2>
<div class="card">
<h3>5.1 input.txt 关键参数</h3>
<table>
<tr><th>参数</th><th>默认值</th><th>说明</th></tr>
<tr><td>gravity_field</td><td>0</td><td>均匀重力场(已关闭)</td></tr>
<tr><td>gravity_interaction</td><td>0</td><td>原子间万有引力(已关闭)</td></tr>
<tr><td>elastic_force</td><td>1</td><td>弹簧键力(已开启)</td></tr>
<tr><td>damping_force</td><td>0</td><td>阻尼(已关闭)</td></tr>
<tr><td><strong>driving_force</strong></td><td><strong>1</strong></td><td>驱动力开关(1=开启,需 driver.txt</td></tr>
<tr><td>method</td><td>leapfrog</td><td>数值积分方法</td></tr>
<tr><td>DT</td><td>0.01</td><td>积分步长 (s)</td></tr>
<tr><td>T_total</td><td>100.0</td><td>总模拟时间 (s)</td></tr>
<tr><td>NSTEP</td><td>50</td><td>抽帧步数间隔</td></tr>
<tr><td>engine</td><td>python</td><td>计算引擎(python / c / cpp / fortran</td></tr>
<tr><td>use_marker</td><td>1</td><td>渲染模式(0=Sphere 网格, 1=Marker GPU 实例化)</td></tr>
</table>
</div>
<div class="card">
<h3>5.2 流程控制参数</h3>
<table>
<tr><th>参数</th><th>0</th><th>1</th></tr>
<tr><td>step_simulate</td><td>跳过模拟(加载已有轨迹)</td><td>运行物理模拟</td></tr>
<tr><td>step_sample</td><td>跳过抽帧</td><td>从轨迹抽取显示帧</td></tr>
<tr><td>step_plot</td><td>不生成图表</td><td>生成轨迹/能量图</td></tr>
<tr><td><strong>step_plot_wave</strong></td><td>不生成波形图</td><td>生成波形能量动画 GIF</td></tr>
<tr><td>step_animation</td><td>不启动动画</td><td>自动打开 VisPy 3D 窗口</td></tr>
<tr><td>force_calc</td><td>自动检测缓存</td><td>强制重新计算</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 6. File Structure -->
<!-- ============================================================ -->
<section id="files">
<h2>六、文件结构</h2>
<pre>case06/
├── input/
│ ├── input.txt # 主配置文件(YAML 格式)
│ ├── coord.txt # 原子坐标(120 个原子)
│ ├── connection.txt # 弹簧连接关系(59 条键)
│ ├── bond.txt # 弹簧参数(k=1.0, L₀=1.0
│ └── <strong>driver.txt</strong> # <span class="cm">驱动力定义(本案例新增)</span>
├── output/
│ ├── trajectory.txt # 全量轨迹数据(50000 步 × 120 原子)
│ ├── display.txt # 抽帧后的动画数据(500 帧 × 120 原子)
│ ├── dynamics.log # 计算日志
│ ├── animation.log # 动画启动日志(闪退时排查用)
│ └── wave_animation.gif # 波形能量动画(step_plot_wave=1 时生成)
├── doc/
│ └── index.html # <span class="cm">本文档</span>
├── Readme.md # 案例简介
└── run_dynamics.py # 案例运行入口</pre>
</section>
<!-- ============================================================ -->
<!-- 7. Troubleshooting -->
<!-- ============================================================ -->
<section id="troubleshoot">
<h2>七、常见问题</h2>
<div class="card">
<h3>7.1 动画窗口闪退</h3>
<p>如果 VisPy 窗口一闪就消失,请检查:</p>
<ul>
<li><code>output/animation.log</code> 中是否有错误信息</li>
<li><code>output/display.txt</code> 是否存在(需先跑 <code>step_sample: 1</code></li>
</ul>
</div>
<div class="card">
<h3>7.2 原子不振动</h3>
<p>可能原因:</p>
<ul>
<li><strong>NSTEP 过大</strong>:抽帧间隔大于驱动周期的一半时,动画会丢失振动细节。建议 NSTEP ≤ 1/(freq · DT · 10)</li>
<li><strong>相位 φ 使采样点落在零值</strong>:试试 <code>phi_z: 0</code> 让原子在 t=0 处于振幅峰值</li>
<li>确认 <code>driving_force: 1</code><code>driver.txt</code> 中 amp_z 不为 0</li>
</ul>
</div>
<div class="card">
<h3>7.3 渲染性能慢</h3>
<p>原子数多时动画卡顿:</p>
<ul>
<li>设置 <code>use_marker: 1</code>(使用 GPU 实例化渲染替代独立网格球体)</li>
<li>增大 <code>NSTEP</code> 减少动画帧数</li>
</ul>
</div>
</section>
<hr style="border:none;border-top:1px solid var(--border);margin:40px 0;">
<footer style="text-align:center;color:var(--muted);font-size:0.85rem;margin-bottom:40px;">
Dynamics Simulation Framework &nbsp;·&nbsp; 生成于 2026-06-10
</footer>
</div>
</body>
</html>
+3
View File
@@ -0,0 +1,3 @@
bond_name k rest_length
h 100.0 1.0
k2 100.0 1.41421356
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
+2
View File
@@ -0,0 +1,2 @@
n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
5101 0 0 2.0 0 0 0.05 90 90 90 all
+89
View File
@@ -0,0 +1,89 @@
# 物理模拟参数配置
# 格式:YAML
# 用法:python run_dynamics.py
# ── 流程控制 ──────────────────────────────────
step_simulate: 1 # 运行物理模拟
step_sample: 0 # 重新抽帧,默认0=不执行
step_plot: 0 # 绘制轨迹/能量图
step_animation: 1 # 自动播放 VisPy 3D 动画窗口
step_plot_wave: 0 # 绘制波形能量动画
force_calc: 1 # 强制重新计算
# ── 文件保存 ──────────────────────────────────
save_trajectory: 0 # 0=不保留完整轨迹文件
# ── 计算引擎 ──────────────────────────────────
engine: c
# ── 盒子 ──────────────────────────────────────
box_a: 120.0
# ── 初始构型 ──────────────────────────────────
coord_file: input/coord.txt
connection_file: input/connection.txt
bond_file: input/bond.txt
driver_file: input/driver.txt
# 绘图/动画展示的原子序号
plot_atom: 5101 # 中心原子 (0,0)
# ── 物理参数 ──────────────────────────────────
G: [0.000, 0.000, 0.000]
B: [0.000, 0.000, 0.000]
# ── 力开关 ────────────────────────────────────
gravity_field: 0
gravity_interaction: 0
elastic_force: 1
damping_force: 0
driving_force: 1
gravity_strength: 1.0
# ── 数值算法 ──────────────────────────────────
method: leapfrog
# ── 步骤控制 ──────────────────────────────────
warmup_steps: 0 # 受迫波动,无需预热
T_total: 100.0
NSTEP: 10
DT: 0.01
sample_start: null
sample_end: null
# ── 渲染方式 ──────────────────────────────────
use_marker: 1
# ── 位移着色 ──────────────────────────────────
display_color: {
x : [0, [255, 0, 0]],
y : [0, [ 0, 255, 0]],
z : [1, [ 0, 0, 255]],
xy : [0, [255, 255, 0]],
yz : [0, [ 0, 255, 255]],
zx : [0, [255, 0, 255]],
xyz : [0, [ 0, 0, 0]],
}
# ── 显示参数 ──────────────────────────────────
alpha: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
ball_color_r: 0.20
ball_color_g: 0.60
ball_color_b: 0.90
box_color_r: 0.80
box_color_g: 0.80
box_color_b: 0.85
# ── 摄像机 ────────────────────────────────────
camera_distance: 120.0
camera_elevation: 60.0
camera_azimuth: -45.0
camera_center_x: 0.0
camera_center_y: 0.0
camera_center_z: 0.0
move_camera: 0
display_amp: [1.0, 1.0, 1.0]
+2
View File
@@ -0,0 +1,2 @@
0 0 50
0 0 80
+54
View File
@@ -0,0 +1,54 @@
"""
Case runner for Dynamics case11 2D grid (61x61 atomic mesh).
This script keeps program and data separated:
- program: ../../dynamics.py
- input: ./input
- output: ./output
"""
from __future__ import annotations
import argparse
import importlib.util
from pathlib import Path
CASE_DIR = Path(__file__).resolve().parent
DYNAMICS_PATH = Path("..") / ".." / "dynamics.py"
INPUT_DIR = Path("input")
OUTPUT_DIR = Path("output")
CONFIG_FILE = INPUT_DIR / "input.txt"
def load_dynamics_module(module_path: Path):
spec = importlib.util.spec_from_file_location("dynamics_module", module_path)
if spec is None or spec.loader is None:
raise ImportError(f"无法加载 dynamics.py: {module_path}")
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
return module
def main():
parser = argparse.ArgumentParser(description="运行 Dynamics 示例案例 case11")
parser.add_argument("--no-plot", action="store_true", help="跳过 matplotlib 绘图")
args = parser.parse_args()
dynamics_path = (CASE_DIR / DYNAMICS_PATH).resolve()
input_dir = (CASE_DIR / INPUT_DIR).resolve()
output_dir = (CASE_DIR / OUTPUT_DIR).resolve()
config_path = (CASE_DIR / CONFIG_FILE).resolve()
module = load_dynamics_module(dynamics_path)
module.run_case(
config_path=config_path,
runtime_base=CASE_DIR,
input_dir=input_dir,
output_dir=output_dir,
no_plot=args.no_plot,
)
if __name__ == "__main__":
main()
+40
View File
@@ -0,0 +1,40 @@
# case06: 一维原子链横波模拟
60 个原子沿 x 轴排列,相邻原子用弹簧连接。原子 1 受 z 方向驱动力作用,产生沿链传播的横波。
## 物理设定
| 参数 | 值 |
|---|---|
| 原子数 | 120 |
| 排列 | 沿 x 轴等间距排列,间距为 1 |
| 约束 | 原子**沿 z 方向自由振动**fix_x=1, fix_y=1, fix_z=0),x, y 锁定 |
| 弹簧 | 劲度系数 k=1.0,原长 L₀=1.0 |
| 重力 | 无 |
| 万有引力 | 无 |
| 阻尼 | 无 |
| 驱动力 | 原子 1(z 方向驱动) |
| 算法 | leapfrog(蛙跳法,能量守恒) |
## 驱动力
原子 1 的位置由 `input/driver.txt` 中的驱动力公式决定:
```math
z(t) = A_z \cdot \cos(2\pi f_z t + \phi_z)
```
当前参数:A_z = 0.5, f_z = 0.1 Hz, φ_z = 90°, period = all(全程驱动)。
## 动力学行为
原子 1 沿 z 方向的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的**横波**。由于 z 方向的振动是横向的,弹簧大部分张力在 x 方向,z 方向的有效刚度是非线性的——等效于一个三次方恢复力(FPU 型非线性),因此波速较慢。
## 使用方法
```bash
cd examples/case06
python run_dynamics.py
```
配置参数详见 `input/input.txt`,驱动力定义见 `input/driver.txt`,完整文档见 `doc/index.html`
+477
View File
@@ -0,0 +1,477 @@
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>case06 — 一维原子链驱动力学模拟 | 物理原理 &amp; 使用文档</title>
<style>
:root {
--bg: #f8f9fa;
--card: #fff;
--text: #1a1a2e;
--accent: #2563eb;
--accent-light: #dbeafe;
--code-bg: #1e293b;
--code-text: #e2e8f0;
--border: #e2e8f0;
--muted: #64748b;
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Noto Sans SC", sans-serif;
background: var(--bg);
color: var(--text);
line-height: 1.7;
}
/* ── Header ── */
.hero {
background: linear-gradient(135deg, #1e293b 0%, #334155 100%);
color: #fff;
padding: 56px 24px 48px;
text-align: center;
}
.hero h1 { font-size: 2rem; font-weight: 700; letter-spacing: -0.02em; }
.hero .subtitle {
margin-top: 10px;
font-size: 1.05rem;
opacity: 0.8;
}
.hero .badge {
display: inline-block;
margin-top: 14px;
padding: 4px 14px;
border-radius: 999px;
background: rgba(255,255,255,0.12);
font-size: 0.82rem;
}
/* ── Layout ── */
.container { max-width: 820px; margin: 0 auto; padding: 32px 20px; }
section { margin-bottom: 44px; }
h2 {
font-size: 1.35rem;
font-weight: 600;
margin-bottom: 16px;
padding-bottom: 8px;
border-bottom: 2px solid var(--accent);
display: inline-block;
}
h3 {
font-size: 1.05rem;
font-weight: 600;
margin: 20px 0 10px;
}
p, li { margin-bottom: 10px; }
ul, ol { padding-left: 22px; }
strong { color: var(--accent); }
/* ── Cards ── */
.card {
background: var(--card);
border-radius: 12px;
padding: 20px 24px;
margin-bottom: 16px;
border: 1px solid var(--border);
box-shadow: 0 1px 3px rgba(0,0,0,0.04);
}
/* ── Formula / Code blocks ── */
.formula {
background: var(--card);
border-left: 4px solid var(--accent);
padding: 14px 20px;
margin: 14px 0;
font-family: "Times New Roman", "STIX", serif;
font-size: 1.05rem;
overflow-x: auto;
border-radius: 0 8px 8px 0;
}
code {
background: var(--accent-light);
padding: 2px 7px;
border-radius: 4px;
font-family: "JetBrains Mono", "Fira Code", monospace;
font-size: 0.88em;
}
pre {
background: var(--code-bg);
color: var(--code-text);
padding: 16px 20px;
border-radius: 10px;
overflow-x: auto;
font-size: 0.85rem;
line-height: 1.5;
margin: 14px 0;
}
pre .cm { color: #94a3b8; font-style: italic; } /* comment */
/* ── Table ── */
table {
width: 100%;
border-collapse: collapse;
margin: 14px 0;
font-size: 0.92rem;
}
th, td {
padding: 8px 12px;
text-align: left;
border-bottom: 1px solid var(--border);
}
th { background: var(--accent-light); font-weight: 600; }
/* ── TOC ── */
.toc { counter-reset: toc; }
.toc li { counter-increment: toc; list-style: none; margin-bottom: 6px; }
.toc li::before { content: counter(toc) ". "; font-weight: 600; color: var(--accent); }
.toc a { color: var(--accent); text-decoration: none; }
.toc a:hover { text-decoration: underline; }
/* ── Flow diagram ── */
.flow { display: flex; flex-wrap: wrap; gap: 8px; align-items: center; justify-content: center; margin: 16px 0; }
.flow-step {
background: var(--accent-light);
border: 1px solid var(--accent);
border-radius: 8px;
padding: 8px 16px;
font-size: 0.88rem;
font-weight: 500;
}
.flow-arrow { color: var(--muted); font-size: 1.2rem; }
@media (max-width: 600px) {
.hero h1 { font-size: 1.5rem; }
.flow { flex-direction: column; }
.flow-arrow { transform: rotate(90deg); }
}
</style>
</head>
<body>
<!-- ============================================================ -->
<!-- Header -->
<!-- ============================================================ -->
<header class="hero">
<h1>一维原子链驱动力学模拟</h1>
<p class="subtitle">120 个原子沿 x 轴排列 · 弹簧连接 · z 方向受迫振动</p>
<span class="badge">case06 · examples/case06</span>
</header>
<div class="container">
<!-- ============================================================ -->
<!-- TOC -->
<!-- ============================================================ -->
<section>
<h2>目录</h2>
<ol class="toc">
<li><a href="#physics">物理原理</a></li>
<li><a href="#algorithm">数值算法</a></li>
<li><a href="#driver">驱动力模型</a></li>
<li><a href="#usage">使用方法</a></li>
<li><a href="#params">参数参考</a></li>
<li><a href="#files">文件结构</a></li>
<li><a href="#troubleshoot">常见问题</a></li>
</ol>
</section>
<!-- ============================================================ -->
<!-- 1. Physics -->
<!-- ============================================================ -->
<section id="physics">
<h2>一、物理原理</h2>
<div class="card">
<h3>1.1 一维原子链</h3>
<p>120 个原子沿 <strong>x 轴</strong> 等间距排列,原子间距为 1。相邻原子之间用 <strong>理想弹簧</strong> 连接,弹簧的劲度系数 <em>k</em> = 1.0,原长 <em>L</em>₀ = 1.0(与原子间距一致,初始状态弹簧无拉伸)。</p>
<p>每个原子被限制在 <strong>z 方向</strong> 自由振动,x 和 y 方向锁定(<code>fix_x=1, fix_y=1, fix_z=0</code>)。</p>
</div>
<div class="card">
<h3>1.2 弹簧力(胡克定律)</h3>
<p>当原子 <em>i</em><em>j</em> 之间有弹簧连接时,原子 <em>i</em> 受到的弹簧力为:</p>
<div class="formula">
<strong>F</strong> = <em>k</em> · (<em>d</em> <em>L</em>₀) · <strong>u</strong><sub><em>ij</em></sub>
</div>
<p>其中 <em>d</em> = |<strong>r</strong><sub><em>j</em></sub> <strong>r</strong><sub><em>i</em></sub>| 为两原子间距离,<strong>u</strong><sub><em>ij</em></sub> 为从 <em>i</em> 指向 <em>j</em> 的单位向量。由于原子只在 z 方向振动,弹簧在 z 方向的分量是 <strong>几何非线性</strong> 的——对于小振幅近似,z 方向等效于一个三次方恢复力(FPU 型非线性)。</p>
</div>
<div class="card">
<h3>1.3 运动方程</h3>
<p>对于第 <em>i</em> 个自由原子(非受驱),牛顿第二定律给出:</p>
<div class="formula">
<em>m</em> · <strong>a</strong><sub><em>i</em></sub> = <strong>F</strong><sub><em>i</em></sub><sup>spring</sup> + <strong>F</strong><sub><em>i</em></sub><sup>driving</sup>
</div>
<p>本案例中 <strong>唯一的外力</strong> 来自驱动力(仅施加于原子 1)。无重力、无万有引力、无阻尼,系统总能量守恒。</p>
</div>
<div class="card">
<h3>1.4 波传播</h3>
<p>原子 1 的受迫振动通过弹簧逐次传递给相邻原子,形成沿链传播的 <strong>横波</strong>。由于横向振动的几何非线性(弹簧大部分张力在 x 方向,z 方向的有效刚度远小于 1),波的传播速度较慢,且高阶频率成分会在链中产生复杂的非线性动力学行为(类似 FPU 回波现象)。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 2. Algorithm -->
<!-- ============================================================ -->
<section id="algorithm">
<h2>二、数值算法</h2>
<div class="card">
<h3>2.1 蛙跳法(Leapfrog / Velocity-Verlet</h3>
<p>采用能量守恒特性优异的 <strong>蛙跳法</strong>(二阶辛积分器),更新公式为:</p>
<div class="formula">
<strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) = <strong>v</strong>(<em>t</em>) + ½ <strong>a</strong>(<em>t</em>) · Δ<em>t</em><br>
<strong>r</strong>(<em>t</em> + Δ<em>t</em>) = <strong>r</strong>(<em>t</em>) + <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) · Δ<em>t</em><br>
<strong>a</strong>(<em>t</em> + Δ<em>t</em>) = <strong>F</strong>(<strong>r</strong>(<em>t</em> + Δ<em>t</em>), <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>)) / <em>m</em><br>
<strong>v</strong>(<em>t</em> + Δ<em>t</em>) = <strong>v</strong>(<em>t</em> + ½Δ<em>t</em>) + ½ <strong>a</strong>(<em>t</em> + Δ<em>t</em>) · Δ<em>t</em>
</div>
<p>蛙跳法在长时间模拟中能量漂移极小(本案例验证 <strong>&lt; 0.004%</strong>),适合无阻尼的保守系统。</p>
</div>
<div class="card">
<h3>2.2 时间步长与采样</h3>
<table>
<tr><th>参数</th><th></th><th>说明</th></tr>
<tr><td>DT</td><td>0.01 s</td><td>积分步长(远小于 1/ω ≈ 0.16 s,满足稳定性条件)</td></tr>
<tr><td>T_total</td><td>100 s</td><td>总模拟时间 → NT = 10000 步</td></tr>
<tr><td>NSTEP</td><td>50</td><td>每 NSTEP 步取一帧用于动画 → 200 帧</td></tr>
<tr><td>method</td><td>leapfrog</td><td>蛙跳法(Velocity-Verlet</td></tr>
</table>
</div>
<div class="card">
<h3>2.3 计算流程</h3>
<div class="flow">
<span class="flow-step">读入 coord.txt<br>connection.txt<br>bond.txt</span>
<span class="flow-arrow"></span>
<span class="flow-step">施加驱动力<br>(驱动原子 1</span>
<span class="flow-arrow"></span>
<span class="flow-step">记录轨迹</span>
<span class="flow-arrow"></span>
<span class="flow-step">蛙跳法<br>更新位置/速度</span>
<span class="flow-arrow"></span>
<span class="flow-step">固定约束<br>x, y 锁定)</span>
<span class="flow-arrow"></span>
<span class="flow-step" style="background:#fef3c7;border-color:#f59e0b;">循环<br>NT 次</span>
</div>
<p style="margin-top:12px;">注意:驱动力在 <strong>每次积分前</strong> 施加,确保受驱原子的位置正确传递给弹簧力计算。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 3. Driving Force -->
<!-- ============================================================ -->
<section id="driver">
<h2>三、驱动力模型</h2>
<div class="card">
<h3>3.1 定义文件</h3>
<p>驱动力由 <code>input/driver.txt</code> 定义,格式如下:</p>
<pre>n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
1 0 0 5 0 0 1 0 0 90 all</pre>
</div>
<div class="card">
<h3>3.2 数学公式</h3>
<p>受驱原子的位置由下式决定(<strong>完全替换</strong> coord.txt 中的初始坐标和固定约束):</p>
<div class="formula">
<strong>r</strong>(<em>t</em>) = <strong>A</strong> · cos(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>速度由解析导数给出:</p>
<div class="formula">
<strong>v</strong>(<em>t</em>) = <strong>A</strong> · 2π<em>f</em> · sin(2π<em>f</em> · <em>t</em> + <strong>φ</strong>)
</div>
<p>其中 <strong>A</strong> = (amp_x, amp_y, amp_z)<strong>f</strong> = (freq_x, freq_y, freq_z) 为不同方向的驱动频率,<strong>φ</strong> = (phi_x, phi_y, phi_z) 为相位(<strong>角度制</strong>,代码自动转换为弧度)。</p>
</div>
<div class="card">
<h3>3.3 本案例驱动参数</h3>
<table>
<tr><th>参数</th><th></th><th>含义</th></tr>
<tr><td>amp_z</td><td>5.0</td><td>z 方向驱动振幅</td></tr>
<tr><td>freq_z</td><td>1.0 Hz</td><td>驱动频率(周期 1 s</td></tr>
<tr><td>phi_z</td><td>90°</td><td>驱动相位 → z(0) = 5·cos(90°) = 0</td></tr>
<tr><td>period</td><td>all</td><td>全程驱动,永不停止</td></tr>
</table>
<div class="formula">
<em>z</em>(<em>t</em>) = 5.0 · cos(2π · 1.0 · <em>t</em> + 90°)
</div>
</div>
<div class="card">
<h3>3.4 有限周期驱动</h3>
<p><code>period</code> 参数支持三种模式:</p>
<ul>
<li><strong>all</strong> — 全程驱动</li>
<li><strong>数值</strong> — 驱动指定周期数后 <strong>静止</strong>(冻结在最终位置,速度归零)。例如 <code>period: 1</code> 表示驱动 1 个完整周期后停止。</li>
</ul>
</div>
<div class="card">
<h3>3.5 驱动与固定约束的关系</h3>
<p>对于受驱原子(<code>driver.txt</code><code>n</code> 指定的原子),其在 <code>coord.txt</code> 中的初始坐标和 <code>fix_x/fix_y/fix_z</code> 约束被 <strong>完全忽略</strong>。原子的位置和速度完全由驱动力公式决定。</p>
</div>
</section>
<!-- ============================================================ -->
<!-- 4. Usage -->
<!-- ============================================================ -->
<section id="usage">
<h2>四、使用方法</h2>
<div class="card">
<h3>4.1 完整运行(模拟 + 动画)</h3>
<pre>cd examples/case06
python run_dynamics.py</pre>
<p>这步会依次执行:物理模拟 → 抽帧 → 打开 VisPy 3D 动画窗口。</p>
</div>
<div class="card">
<h3>4.2 仅查看已有结果</h3>
<p>如果已经跑完模拟且生成了 <code>output/display.txt</code>,可以通过修改 <code>input.txt</code> 跳过计算,只开动画:</p>
<pre>step_simulate: 0 # 跳过模拟
step_sample: 0 # 跳过抽帧
step_animation: 1 # 播放动画</pre>
<p>然后运行:<code>python run_dynamics.py</code></p>
</div>
<div class="card">
<h3>4.3 手动 3D 动画</h3>
<p>也可以单独启动 VisPy 窗口:</p>
<pre>python ../../draw.py output/</pre>
</div>
<div class="card">
<h3>4.4 强制重新计算</h3>
<p>修改参数后需要重新运行模拟时,设置:</p>
<pre>force_calc: 1 # 忽略缓存,强制重新计算</pre>
</div>
<div class="card">
<h3>4.5 动画交互</h3>
<table>
<tr><th>操作</th><th>效果</th></tr>
<tr><td>鼠标拖动</td><td>旋转视角</td></tr>
<tr><td>滚轮</td><td>缩放</td></tr>
<tr><td>W / S 键</td><td>相机沿 Z 轴向前 / 向后移动(靠近/远离场景)</td></tr>
<tr><td>A / D 键</td><td>视角向右 / 向左平移</td></tr>
<tr><td>E / Q 键</td><td>视角上升 / 下降(屏幕方向)</td></tr>
<tr><td>C / X 键</td><td>增大 / 减小步长</td></tr>
<tr><td>V 键</td><td>切换透视 / 正交投影</td></tr>
<tr><td>左上角 <strong>reset</strong> 按钮</td><td>复位视角到初始位置</td></tr>
<tr><td>左上角 <strong>info</strong> 按钮</td><td>切换信息面板显示/隐藏</td></tr>
<tr><td>左上角 <strong>axes</strong> 按钮</td><td>切换坐标轴显示/隐藏</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 5. Parameters -->
<!-- ============================================================ -->
<section id="params">
<h2>五、参数参考</h2>
<div class="card">
<h3>5.1 input.txt 关键参数</h3>
<table>
<tr><th>参数</th><th>默认值</th><th>说明</th></tr>
<tr><td>gravity_field</td><td>0</td><td>均匀重力场(已关闭)</td></tr>
<tr><td>gravity_interaction</td><td>0</td><td>原子间万有引力(已关闭)</td></tr>
<tr><td>elastic_force</td><td>1</td><td>弹簧键力(已开启)</td></tr>
<tr><td>damping_force</td><td>0</td><td>阻尼(已关闭)</td></tr>
<tr><td><strong>driving_force</strong></td><td><strong>1</strong></td><td>驱动力开关(1=开启,需 driver.txt</td></tr>
<tr><td>method</td><td>leapfrog</td><td>数值积分方法</td></tr>
<tr><td>DT</td><td>0.01</td><td>积分步长 (s)</td></tr>
<tr><td>T_total</td><td>100.0</td><td>总模拟时间 (s)</td></tr>
<tr><td>NSTEP</td><td>50</td><td>抽帧步数间隔</td></tr>
<tr><td>engine</td><td>python</td><td>计算引擎(python / c / cpp / fortran</td></tr>
<tr><td>use_marker</td><td>1</td><td>渲染模式(0=Sphere 网格, 1=Marker GPU 实例化)</td></tr>
</table>
</div>
<div class="card">
<h3>5.2 流程控制参数</h3>
<table>
<tr><th>参数</th><th>0</th><th>1</th></tr>
<tr><td>step_simulate</td><td>跳过模拟(加载已有轨迹)</td><td>运行物理模拟</td></tr>
<tr><td>step_sample</td><td>跳过抽帧</td><td>从轨迹抽取显示帧</td></tr>
<tr><td>step_plot</td><td>不生成图表</td><td>生成轨迹/能量图</td></tr>
<tr><td><strong>step_plot_wave</strong></td><td>不生成波形图</td><td>生成波形能量动画 GIF</td></tr>
<tr><td>step_animation</td><td>不启动动画</td><td>自动打开 VisPy 3D 窗口</td></tr>
<tr><td>force_calc</td><td>自动检测缓存</td><td>强制重新计算</td></tr>
</table>
</div>
</section>
<!-- ============================================================ -->
<!-- 6. File Structure -->
<!-- ============================================================ -->
<section id="files">
<h2>六、文件结构</h2>
<pre>case06/
├── input/
│ ├── input.txt # 主配置文件(YAML 格式)
│ ├── coord.txt # 原子坐标(120 个原子)
│ ├── connection.txt # 弹簧连接关系(59 条键)
│ ├── bond.txt # 弹簧参数(k=1.0, L₀=1.0
│ └── <strong>driver.txt</strong> # <span class="cm">驱动力定义(本案例新增)</span>
├── output/
│ ├── trajectory.txt # 全量轨迹数据(50000 步 × 120 原子)
│ ├── display.txt # 抽帧后的动画数据(500 帧 × 120 原子)
│ ├── dynamics.log # 计算日志
│ ├── animation.log # 动画启动日志(闪退时排查用)
│ └── wave_animation.gif # 波形能量动画(step_plot_wave=1 时生成)
├── doc/
│ └── index.html # <span class="cm">本文档</span>
├── Readme.md # 案例简介
└── run_dynamics.py # 案例运行入口</pre>
</section>
<!-- ============================================================ -->
<!-- 7. Troubleshooting -->
<!-- ============================================================ -->
<section id="troubleshoot">
<h2>七、常见问题</h2>
<div class="card">
<h3>7.1 动画窗口闪退</h3>
<p>如果 VisPy 窗口一闪就消失,请检查:</p>
<ul>
<li><code>output/animation.log</code> 中是否有错误信息</li>
<li><code>output/display.txt</code> 是否存在(需先跑 <code>step_sample: 1</code></li>
</ul>
</div>
<div class="card">
<h3>7.2 原子不振动</h3>
<p>可能原因:</p>
<ul>
<li><strong>NSTEP 过大</strong>:抽帧间隔大于驱动周期的一半时,动画会丢失振动细节。建议 NSTEP ≤ 1/(freq · DT · 10)</li>
<li><strong>相位 φ 使采样点落在零值</strong>:试试 <code>phi_z: 0</code> 让原子在 t=0 处于振幅峰值</li>
<li>确认 <code>driving_force: 1</code><code>driver.txt</code> 中 amp_z 不为 0</li>
</ul>
</div>
<div class="card">
<h3>7.3 渲染性能慢</h3>
<p>原子数多时动画卡顿:</p>
<ul>
<li>设置 <code>use_marker: 1</code>(使用 GPU 实例化渲染替代独立网格球体)</li>
<li>增大 <code>NSTEP</code> 减少动画帧数</li>
</ul>
</div>
</section>
<hr style="border:none;border-top:1px solid var(--border);margin:40px 0;">
<footer style="text-align:center;color:var(--muted);font-size:0.85rem;margin-bottom:40px;">
Dynamics Simulation Framework &nbsp;·&nbsp; 生成于 2026-06-10
</footer>
</div>
</body>
</html>
+3
View File
@@ -0,0 +1,3 @@
bond_name k rest_length
h 100.0 1.0
k2 100.0 1.41421356
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
+3
View File
@@ -0,0 +1,3 @@
n amp_x amp_y amp_z freq_x freq_y freq_z phi_x phi_y phi_z period
3081 0 0 2.0 0 0 0.05 90 90 90 all
7121 0 0 2.0 0 0 0.05 90 90 90 all
+83
View File
@@ -0,0 +1,83 @@
# 物理模拟参数配置
# case12 — 二维网格两点干涉(双点源 z 方向驱动)
# 驱动点: (0,-10) 和 (0,10),波从两点向外传播,在中心区域干涉
# ── 流程控制 ──────────────────────────────────
step_simulate: 1 # 运行物理模拟
step_sample: 0 # 重新抽帧,默认0=不执行
step_plot: 0 # 绘制轨迹/能量图
step_animation: 1 # 自动播放 VisPy 3D 动画窗口
step_plot_wave: 0 # 绘制波形能量动画
force_calc: 1 # 强制重新计算
# ── 文件保存 ──────────────────────────────────
save_trajectory: 0
# ── 计算引擎 ──────────────────────────────────
engine: c
# ── 盒子 ──────────────────────────────────────
box_a: 120.0
# ── 初始构型 ──────────────────────────────────
coord_file: input/coord.txt
connection_file: input/connection.txt
bond_file: input/bond.txt
driver_file: input/driver.txt
plot_atom: 5101 # 中心区域用于信息显示
# ── 物理参数 ──────────────────────────────────
G: [0.000, 0.000, 0.000]
B: [0.000, 0.000, 0.000]
gravity_field: 0
gravity_interaction: 0
elastic_force: 1
damping_force: 0
driving_force: 1
gravity_strength: 1.0
method: leapfrog
# ── 步骤控制 ──────────────────────────────────
warmup_steps: 0
T_total: 100.0
NSTEP: 500
DT: 0.001
sample_start: null
sample_end: null
# ── 渲染/着色 ─────────────────────────────────
use_marker: 1
display_color: {
x : [0, [255, 0, 0]],
y : [0, [ 0, 255, 0]],
z : [1, [ 0, 0, 255]],
xy : [1, [255, 0, 0]],
yz : [0, [ 0, 255, 0]],
zx : [0, [ 0, 0, 255]],
xyz : [0, [255, 255, 255]],
}
# ── 显示参数 ──────────────────────────────────
alpha: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
ball_color_r: 0.20
ball_color_g: 0.60
ball_color_b: 0.90
box_color_r: 0.80
box_color_g: 0.80
box_color_b: 0.85
# ── 摄像机 ────────────────────────────────────
camera_distance: 120.0
camera_elevation: 60.0
camera_azimuth: -45.0
camera_center_x: 0.0
camera_center_y: 0.0
camera_center_z: 0.0
move_camera: 0
display_amp: [1.0, 1.0, 1.0]
+2
View File
@@ -0,0 +1,2 @@
0 0 50
0 0 80
+54
View File
@@ -0,0 +1,54 @@
"""
Case runner for Dynamics case12 2D grid dual source interference.
This script keeps program and data separated:
- program: ../../dynamics.py
- input: ./input
- output: ./output
"""
from __future__ import annotations
import argparse
import importlib.util
from pathlib import Path
CASE_DIR = Path(__file__).resolve().parent
DYNAMICS_PATH = Path("..") / ".." / "dynamics.py"
INPUT_DIR = Path("input")
OUTPUT_DIR = Path("output")
CONFIG_FILE = INPUT_DIR / "input.txt"
def load_dynamics_module(module_path: Path):
spec = importlib.util.spec_from_file_location("dynamics_module", module_path)
if spec is None or spec.loader is None:
raise ImportError(f"无法加载 dynamics.py: {module_path}")
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
return module
def main():
parser = argparse.ArgumentParser(description="运行 Dynamics 示例案例 case11")
parser.add_argument("--no-plot", action="store_true", help="跳过 matplotlib 绘图")
args = parser.parse_args()
dynamics_path = (CASE_DIR / DYNAMICS_PATH).resolve()
input_dir = (CASE_DIR / INPUT_DIR).resolve()
output_dir = (CASE_DIR / OUTPUT_DIR).resolve()
config_path = (CASE_DIR / CONFIG_FILE).resolve()
module = load_dynamics_module(dynamics_path)
module.run_case(
config_path=config_path,
runtime_base=CASE_DIR,
input_dir=input_dir,
output_dir=output_dir,
no_plot=args.no_plot,
)
if __name__ == "__main__":
main()

Some files were not shown because too many files have changed in this diff Show More