关于「实践与证伪:给营销文案装一个音律伪精确检测器」,《三分损益、朱载堉密率与拍频粗糙度》给出的定量结论是什么?

结论:第八章给出执行清单、文案检查规则与样本量为 169 的预先登记检验。书中以标注框区分已发表的证据、可运行的 Python 计算(22 段程序)、原典与作者的假说。

数据口径:书中的听者、样本与文案均为示意,全部计算由书中程序生成;本书不是医疗建议,不宣称任何频率、音阶或乐器有身心功效,也不构成购买建议。 书中以标注框区分已发表的证据、可运行的 Python 计算(22 段程序)、原典与作者的假说。

English version

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: 169

true p1first nstable ncritical countpower
0.556206543490.811
0.60158169960.823
0.656976460.826
0.703742270.836
0.752328190.862
n = 200: reject if at least 113 of 200 pick the just third
size of that test: 0.0384
查看完整验算程序
# Chapter 8. p81: how many paired judgements a pre-registered test needs.
# Design: each trial plays one interval twice (say just and equal major
# third on C5, order randomised); the listener names the smoother one.
# Null: p = 0.5. One-sided exact binomial test at alpha = 0.05. For a
# true share p1 we find the smallest n with power >= 0.80 that also
# keeps power >= 0.80 for every n up to n + 40 (the binomial power
# curve is saw-toothed). Critical counts use exact integers.
import math

NMAX = 720


def crit(n):
    # smallest c with sum_{k>=c} C(n,k) <= 0.05 * 2^n, in integers
    acc, c, lim = 0, n + 1, 2 ** n
    while c > 0:
        nxt = acc + math.comb(n, c - 1)
        if 20 * nxt > lim:
            break
        acc, c = nxt, c - 1
    return c, acc


def power(n, c, p1):
    lp, lq = math.log(p1), math.log(1 - p1)
    terms = []
    for k in range(c, n + 1):
        lg = (math.lgamma(n + 1) - math.lgamma(k + 1)
              - math.lgamma(n - k + 1))
        terms.append(math.exp(lg + k * lp + (n - k) * lq))
    return math.fsum(terms)


def main():
    crits = {n: crit(n) for n in range(5, NMAX + 1)}
    print("  true p1   first n   stable n  critical count  power")
    for p1 in (0.55, 0.60, 0.65, 0.70, 0.75):
        pw = {n: power(n, crits[n][0], p1) for n in crits}
        first = min(n for n in pw if pw[n] >= 0.8)
        stable = min(n for n in pw if n + 40 <= NMAX and
                     all(pw[m] >= 0.8 for m in range(n, n + 41)))
        print("{:>9.2f} {:>9} {:>10} {:>15} {:>6.3f}".format(
            p1, first, stable, crits[stable][0], pw[stable]))
    n = 200
    c, acc = crits[n]
    print("n = {}: reject if at least {} of {} pick the just third".format(
        n, c, n))
    print("size of that test: {:.4f}".format(acc / 2 ** n))


main()

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《三分损益律旋宫不归、朱载堉十二平均律开方逼近与纯律泛音拍频粗糙度》第 8 章「实践与证伪:给营销文案装一个音律伪精确检测器」(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-159 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026159,
  author    = {Fan, Yafu},
  title     = {The Unclosed Circle of Pythagorean Tuning, Zhu Zaiyu's Root Approximation of Equal Temperament and the Beat Roughness of Just-Intonation Overtones},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论