关于「同名同辈,撞名远比想象多」,《族谱合龙与义庄永续》给出的定量结论是什么?
结论:先看 A 段。齐夫指数取 0.8,各字概率平方和为 0.01520。平方和倒数,叫有效字池,只有 65.8 个字。四百个候选字,看着宽,实际起作用,不到七十个。最常用十个字,就占了 0.294。指数取 0.6,有效字池也只有 159.6;取 1.0,缩到 26.3。
数据口径:按字辈起名,同代人只剩一字可变,示意参数下一代 11 人就有一半机会撞名;两本 60 人对 40 人之支谱按全名硬接,接对者只占 0.208。 书中数据全为合成或示意,不含任何真实在世者资料;全部计算由书中程序生成;本书是非个人化分析,不构成法律、理财或经营建议,也不做任何血统高下或族群比较。
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: 0.01520, 0.294, 0.6, 0.8, 1.0, 159.6
| A. one generation, K | 400 free characters |
| Zipf s sum p^2 effective pool top-10 share | |
| 0.6 | 0.00627 159.6 0.174 |
| 0.8 | 0.01520 65.8 0.294 |
| 1.0 | 0.03805 26.3 0.446 |
| B. chance of at least one repeated full name among n men | |
| n uniform s | 0.6 s=0.8 s=1.0 pairs s=0.8 |
| 10 | 0.107 0.238 0.453 0.741 0.68 |
| 20 | 0.383 0.667 0.896 0.991 2.89 |
| 30 | 0.672 0.912 0.991 1.000 6.61 |
| 50 | 0.959 0.999 1.000 1.000 18.62 |
| 80 | 1.000 1.000 1.000 1.000 48.04 |
| C. men needed for a 50% / 90% chance of a repeat | |
| approximation sqrt(2 ln 2 / sum p^2): uniform 23.5, s | 0.8 9.5 |
| uniform | 24 / 43 |
| Zipf s | 0.6: 16 / 30 |
| Zipf s | 0.8: 11 / 21 |
| Zipf s | 1.0: 8 / 14 |
查看完整验算程序
# Chapter 1. p11: same-name collisions inside one generation (birthday).
# Model: men of one surname and one generation share the generation
# character, so a two-character given name differs only in the free
# character. The free character is drawn from K candidates with Zipf
# weights p_i proportional to 1 / i^s. The chance that n men contain at
# least one repeated name is 1 - n! e_n(p), where e_n is the elementary
# symmetric polynomial of the weights (exact dynamic programme). The
# uniform case is 1 - prod(1 - i / K). K = 400 and s = 0.8 are
# illustrative values (s = 0.6 and 1.0 shown as sensitivity), not
# measured from any real genealogy.
import math
K = 400
def zipf(k, s):
raw = [1.0 / (i ** s) for i in range(1, k + 1)]
tot = math.fsum(raw)
return [r / tot for r in raw]
def p_no_repeat(weights, n):
# e[j] = elementary symmetric polynomial of degree j
e = [1.0] + [0.0] * n
for w in weights:
for j in range(n, 0, -1):
e[j] = math.fsum([e[j], w * e[j - 1]])
return math.factorial(n) * e[n]
def p_no_repeat_uniform(k, n):
return math.prod(1.0 - i / k for i in range(n))
def first_n(fun, target=0.5):
n = 1
while 1.0 - fun(n) < target:
n += 1
return n
if __name__ == "__main__":
sets = [(sv, zipf(K, sv)) for sv in (0.6, 0.8, 1.0)]
print("A. one generation, K = {} free characters".format(K))
print(" Zipf s sum p^2 effective pool top-10 share")
for sv, w in sets:
q = math.fsum(x * x for x in w)
print(" {:>6.1f} {:.5f} {:>14.1f} {:>12.3f}".format(
sv, q, 1.0 / q, math.fsum(w[:10])))
print("B. chance of at least one repeated full name among n men")
print(" n uniform s=0.6 s=0.8 s=1.0 pairs s=0.8")
q8 = math.fsum(x * x for x in sets[1][1])
for n in (10, 20, 30, 50, 80):
row = [1.0 - p_no_repeat_uniform(K, n)] + [
1.0 - p_no_repeat(w, n) for _, w in sets]
print(" {:>4} {:>6.3f} {:>6.3f} {:>6.3f} {:>6.3f} {:>8.2f}"
.format(n, *row, math.comb(n, 2) * q8))
print("C. men needed for a 50% / 90% chance of a repeat")
print(" approximation sqrt(2 ln 2 / sum p^2): uniform {:.1f},"
" s=0.8 {:.1f}".format(math.sqrt(2 * math.log(2) * K),
math.sqrt(2 * math.log(2) / q8)))
nu = first_n(lambda n: p_no_repeat_uniform(K, n))
nu9 = first_n(lambda n: p_no_repeat_uniform(K, n), 0.9)
print(" uniform: {} / {}".format(nu, nu9))
for sv, w in sets:
a = first_n(lambda n: p_no_repeat(w, n))
b = first_n(lambda n: p_no_repeat(w, n), 0.9)
print(" Zipf s={:.1f}: {} / {}".format(sv, a, b))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《宗族世系有向无环图合龙算法、字辈同余编码与范氏义庄跨代治理母典》第 1 章「同名同辈,撞名远比想象多」(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-334 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026334,
author = {Fan, Yafu},
title = {Genealogical DAG Merge Algorithms, Generation-Poem Congruence Coding and the Institutional Economics of the Fan Clan Charitable Estate},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}