关于「盒计数:把自然纹理量成一个维数」,《亲生命分形界面与首屏信任》给出的定量结论是什么?
结论:结果有一个系统规律:理论维数越高,估计值低得越多。H 为 0.9 时理论值 1.10,估计均值 1.144,反而略高;H 为 0.5 时理论值 1.50,估计均值只有 1.409;H 为 0.3 时理论值 1.70,估计均值 1.554。原因在于有限点数:曲线越粗糙,越需要更细的盒子才能把起伏数全,而四千多个点撑不起那么细的层级。二十条曲线之间的标准差在 0.024 到 0.041 之间,说明同一参数下不同随机种子也会带来可见差异。
数据口径:核对原文后,本书写下蓝图修正:中等维数受偏爱,但证据来自小样本实验室静态图片,峰位随刺激类型与人群漂移;「降压」只是皮肤电导、脑电 alpha 一类实验室压力标志物的差异,不是血压读数,本书不宣称任何健康功效;「显著提升转化」改写为带推翻条件的中介链假说;「黄金配比」改写为用自家数据拟合的最优区间。 书中参数为示意;全部计算由书中程序生成;本书不是医疗建议,也不是个人化经营建议。
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: 0.024, 0.041, 0.3, 0.5, 0.9, 1.10
| midpoint displacement, 4,097 points, 20 profiles per H | |||||
| H | D=2-H | mean D | sd | min | max |
| 0.9 | 1.10 | 1.144 | 0.033 | 1.087 | 1.193 |
| 0.7 | 1.30 | 1.276 | 0.041 | 1.228 | 1.355 |
| 0.6 | 1.40 | 1.353 | 0.032 | 1.302 | 1.436 |
| 0.5 | 1.50 | 1.409 | 0.026 | 1.360 | 1.454 |
| 0.4 | 1.60 | 1.495 | 0.037 | 1.409 | 1.578 |
| 0.3 | 1.70 | 1.554 | 0.024 | 1.513 | 1.600 |
| H values whose mean D lands in 1.3 to 1.5: 0.6, 0.5, 0.4 |
查看完整验算程序
# Chapter 1. p12: midpoint displacement profiles and their box dimension
# Random midpoint displacement (Fournier, Fussell and Carpenter 1982)
# builds a ridge line with Hurst exponent H; theory says the graph has
# D = 2 - H. We box-count the graph (column min-max cover) and compare.
# Twenty profiles per H; seeds fixed; all values illustrative.
import math
import random
import statistics
def profile(h, levels, rng):
n = 2 ** levels
y = [0.0] * (n + 1)
y[n] = rng.gauss(0.0, 1.0)
step, scale = n, 1.0
while step > 1:
half = step // 2
scale *= 0.5 ** h
for i in range(half, n, step):
y[i] = (y[i - half] + y[i + half]) / 2 + rng.gauss(0.0, scale)
step = half
lo, hi = min(y), max(y)
return [(v - lo) / (hi - lo) for v in y]
def graph_dim(y, kmin=2, kmax=8):
n = len(y) - 1
xs, ys = [], []
for k in range(kmin, kmax + 1):
m = 2 ** k
per = n // m
count = 0
for c in range(m):
seg = y[c * per:(c + 1) * per + 1]
a = min(int(min(seg) * m), m - 1)
b = min(int(max(seg) * m), m - 1)
count += b - a + 1
xs.append(k * math.log(2))
ys.append(math.log(count))
mx, my = statistics.fmean(xs), statistics.fmean(ys)
sxy = math.fsum((u - mx) * (v - my) for u, v in zip(xs, ys))
sxx = math.fsum((u - mx) ** 2 for u in xs)
return sxy / sxx
rng = random.Random(41802)
print("midpoint displacement, 4,097 points, 20 profiles per H")
print("{:>5} {:>9} {:>8} {:>7} {:>7} {:>9}".format(
"H", "D=2-H", "mean D", "sd", "min", "max"))
band = []
for h in (0.9, 0.7, 0.6, 0.5, 0.4, 0.3):
ds = [graph_dim(profile(h, 12, rng)) for _ in range(20)]
m = statistics.fmean(ds)
print("{:>5.1f} {:>9.2f} {:>8.3f} {:>7.3f} {:>7.3f} {:>9.3f}".format(
h, 2 - h, m, statistics.stdev(ds), min(ds), max(ds)))
if 1.3 <= m <= 1.5:
band.append(h)
print("H values whose mean D lands in 1.3 to 1.5:",
", ".join("{:.1f}".format(v) for v in band))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《亲生命自然分形与几何对称:数字界面潜意识降压与首屏信任度工程母典》第 1 章「盒计数:把自然纹理量成一个维数」(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-418 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026418,
author = {Fan, Yafu},
title = {Biophilic Botanical Fractals and Geometric Symmetry in Interfaces: Subconscious Stress Reduction and First-Impression Trust Lift},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}