关于「一维有限差分:均匀热度怎样碎成等距高点」,《图灵斑图与据点间距》给出的定量结论是什么?
结论:隐显式有限差分把扩散与衰减放到下一时刻去解,把产生项留在当前时刻,用托马斯算法解三对角方程组,让纯 Python 在一秒内算完一条 40 公里的走廊。程序 p21 显示,从均匀热度出发,格局经历线性放大、非线性变尖、高点合并三个阶段,最终在默认参数下停在 6 个高点,平均间距 7.20 公里;总热度从 80.00 降到 63.67。程序 p22 显示,格宽与步长在合理范围内变化时,高点个数不变、间距变化不到百分之一;而换一组初始噪声,间距可以变一成。那么,究竟是什么决定了间距?为什么同一组方程可以停在不同的间距上?下一章回到色散关系,把这两个问题算清楚。
数据口径:书中半径、衰减率、城市长度与数据均为示意;全部计算由书中程序生成;本书属非个人化经营方法,不构成投资、法律或选址决策建议。 书中以标注框区分已发表的证据、可运行的 Python 计算(17 段程序)、原典与作者的假说。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 40, 63.67, 7.20, 80.00
| t | peaks | contrast | max a | total activity (a dx) |
|---|---|---|---|---|
| 0 | 0 | 1.010 | 2.01 | 80.00 |
| 10 | 7 | 1.112 | 2.11 | 79.99 |
| 20 | 7 | 8.872 | 3.75 | 75.75 |
| 30 | 6 | 34.673 | 4.38 | 64.18 |
| 40 | 6 | 196.809 | 4.44 | 63.09 |
| 60 | 6 | 121.263 | 4.42 | 63.31 |
| 100 | 6 | 71.495 | 4.39 | 63.49 |
| 200 | 6 | 47.826 | 4.35 | 63.60 |
| 400 | 6 | 44.306 | 4.32 | 63.67 |
| peak positions km: 3.9 11.5 18.9 26.1 33.1 39.9 | ||||
| gaps km: 7.6 7.4 7.2 7.0 6.8 | edge margins 3.9 and 0.1 | |||
| mean gap 7.20 km; onset scale 2 pi sqrt(la lh) = 7.54 km |
查看完整验算程序
# Chapter 2. p21: how evenly spread outposts break into regular peaks.
# Model: Gierer-Meinhardt with mu = 1, nu = 2, la = 0.6 km, lh = 2.4 km
# (reach ratio 4, illustrative) on a 40 km corridor, no-flux ends.
# Scheme: cell width 0.2 km, step 0.05; decay and diffusion implicit
# (Thomas algorithm), production a^2 / h and a^2 explicit. Start from
# the uniform state with 1 per cent seeded noise; print snapshots of
# peak count, contrast and the largest activity, then the final gaps.
import math
import random
MU, NU, LA, LH = 1.0, 2.0, 0.6, 2.4
LENGTH, DX, DT = 40.0, 0.2, 0.05
def factor(n, r, c):
di = [1.0 + c + 2.0 * r] * n
di[0] = di[-1] = 1.0 + c + r
cp, den = [0.0] * n, [0.0] * n
den[0] = di[0]
cp[0] = -r / den[0]
for i in range(1, n):
den[i] = di[i] + r * cp[i - 1]
cp[i] = -r / den[i]
return cp, den, r
def solve(f, rhs):
cp, den, r = f
n = len(rhs)
y = [0.0] * n
y[0] = rhs[0] / den[0]
for i in range(1, n):
y[i] = (rhs[i] + r * y[i - 1]) / den[i]
for i in range(n - 2, -1, -1):
y[i] -= cp[i] * y[i + 1]
return y
def peaks(a):
lo, hi = min(a), max(a)
if hi - lo < 0.01 * hi:
return []
cut, n = lo + 0.5 * (hi - lo), len(a)
return [(i + 0.5) * DX for i in range(n) if a[i] > cut
and (i == 0 or a[i] >= a[i - 1])
and (i == n - 1 or a[i] > a[i + 1])]
n = int(round(LENGTH / DX))
rng = random.Random(20261113)
a = [2.0 * (1.0 + 0.01 * (rng.random() - 0.5)) for _ in range(n)]
h = [2.0] * n
fa = factor(n, DT * MU * LA * LA / DX ** 2, DT * MU)
fh = factor(n, DT * NU * LH * LH / DX ** 2, DT * NU)
marks = (0, 10, 20, 30, 40, 60, 100, 200, 400)
print(" t peaks contrast max a total activity (a dx)")
step = 0
for t_mark in marks:
while step * DT < t_mark - 1e-9:
ra = [a[i] + DT * a[i] * a[i] / h[i] for i in range(n)]
rh = [h[i] + DT * a[i] * a[i] for i in range(n)]
a, h = solve(fa, ra), solve(fh, rh)
step += 1
print("%4d %5d %8.3f %5.2f %8.2f"
% (t_mark, len(peaks(a)), max(a) / min(a), max(a),
DX * math.fsum(a)))
p = peaks(a)
gaps = [p[i + 1] - p[i] for i in range(len(p) - 1)]
print("peak positions km: " + " ".join("%.1f" % x for x in p))
print("gaps km: " + " ".join("%.1f" % g for g in gaps)
+ " edge margins %.1f and %.1f" % (p[0], LENGTH - p[-1]))
print("mean gap %.2f km; onset scale 2 pi sqrt(la lh) = %.2f km"
% (math.fsum(gaps) / len(gaps), 2 * math.pi * math.sqrt(LA * LH)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《图灵反应扩散斑图、小众口碑品牌陌生城市据点涌现与区域垄断扩散比条件母典》第 2 章「一维有限差分:均匀热度怎样碎成等距高点」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-247 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026247,
author = {Fan, Yafu},
title = {Turing Reaction-Diffusion Patterns: How the Outposts of a Niche Word-of-Mouth Brand Emerge in a New City, and the Diffusion-Ratio Condition for Regional Dominance},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}