关于「合并冲突的生日问题:共享文件、热点文件与私有目录」,《团队规模的协调边界》给出的定量结论是什么?
结论:第五章把合并冲突写成生日问题:一个热点文件让四个窗口的冲突概率达到 0.6875,私有目录则为零。
数据口径:书中参数为示意;全部计算由书中程序生成;本书属于非个人化经营建议。 第四章以本出版局每波四卷并行为示意场景:容量峰值在 6.78 个窗口,四个窗口已拿到峰值的 89.6 %,协调者的审稿队列把上限压在四。
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: 0.6875
| P(at least one conflict per wave) | ||||
| n | A uniform | B +hot file | C hot->coord | D private |
| 2 | 0.1175 | 0.3382 | 0.1175 | 0.0000 |
| 3 | 0.3086 | 0.6543 | 0.3086 | 0.0000 |
| 4 | 0.5164 | 0.8489 | 0.5164 | 0.0000 |
| 5 | 0.6965 | 0.9431 | 0.6965 | 0.0000 |
| 6 | 0.8282 | 0.9812 | 0.8282 | 0.0000 |
| 8 | 0.9589 | 0.9986 | 0.9589 | 0.0000 |
| hot file alone, n = 4: 0.6875; n = 2: 0.2500 | ||||
| A with only 40 shared files left, n = 4: 0.1352 |
查看完整验算程序
# Chapter 5. p52: one hot shared file against private directories.
# Each agent touches file j independently with probability p_j. No file
# is touched by two agents with probability
# prod_j [(1 - p_j)^n + n p_j (1 - p_j)^(n - 1)], computed as a math.fsum
# of logarithms. Layouts (illustrative): A, 200 shared files, p = 0.025
# each (about 5 per agent); B, A plus one catalogue file every agent
# edits with p = 0.5; C, B with the catalogue written only by the
# coordinator; D, private directory per agent, nothing shared.
import math
def p_clash(n, probs):
logs = [math.log((1 - p) ** n + n * p * (1 - p) ** (n - 1))
for p in probs]
return 1.0 - math.exp(math.fsum(logs))
LAYOUT = {
"A uniform": [0.025] * 200,
"B +hot file": [0.025] * 200 + [0.5],
"C hot->coord": [0.025] * 200,
"D private": [0.0] * 200,
}
if __name__ == "__main__":
names = ("A uniform", "B +hot file", "C hot->coord", "D private")
print("P(at least one conflict per wave)")
print(" n " + " ".join("{:>12}".format(x) for x in names))
for n in (2, 3, 4, 5, 6, 8):
print(" {:2d} ".format(n) + " ".join(
"{:12.4f}".format(p_clash(n, LAYOUT[x])) for x in names))
hot = [0.5]
print("hot file alone, n = 4: {:.4f}; n = 2: {:.4f}".format(
p_clash(4, hot), p_clash(2, hot)))
shared = [0.025] * 40
print("A with only 40 shared files left, n = 4: {:.4f}".format(
p_clash(4, shared)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《林格曼效应与布鲁克斯定律、团队与智能体军团规模的协调开销最优边界母典》第 5 章「合并冲突的生日问题:共享文件、热点文件与私有目录」(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-322 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026322,
author = {Fan, Yafu},
title = {The Ringelmann Effect and Brooks's Law: the Optimal Size of Teams and Agent Squads under Coordination Overhead},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}