关于「交界前后一刻钟:翻转概率的精确计算」,《出生时刻误差与交界翻转》给出的定量结论是什么?

结论:第二章核实立项命题:分界前后一刻钟内,凭记忆报告的平均翻转概率是 0.4401,医院记录只有 0.1135,恰好报在分界上的时刻不论来源都是一半。

数据口径:本书只分析排盘规则的结构、误差传播与概率,不做任何个人占断,不宣称任何体系有预测效力;书中参数为示意,全部计算由书中程序生成。 第一章说明,示意模型中凭记忆报告的时刻有 0.2782 恰好落在时辰分界上,不取整时只有 0.0083。

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: 0.1135, 0.4401

flip probability by signed distance d = report - 11:00 (minutes)
dmemoryrecordno-round(s=15)
-300.03770.00000.0228
-200.09210.00000.0912
-150.16520.00010.1587
-100.25330.00090.2525
-50.36990.05850.3694
-10.47340.36940.4734
+00.50000.50000.5000
+10.47340.36940.4734
+50.36990.05850.3694
+100.25330.00090.2525
+150.16520.00010.1587
+200.09210.00000.0912
+300.03770.00000.0228
one 120-minute cycle (repeats all day)
sourceP(|d|<=15)flip in windowflip overall
memory0.36960.44010.1736
record0.27540.11350.0313
no-round0.25830.31060.0998
查看完整验算程序
# Chapter 2. p22: flip probability against the distance of the report
# from the nearest double-hour boundary, and its average inside the
# quarter-hour window on either side of a boundary.
# Model and illustrative weights as in p21. With a uniform prior the
# pattern repeats every 120 minutes, so one cycle around the 11:00
# boundary is enough; the posterior is evaluated on t within 240
# minutes of r, beyond which the likelihood is below 1e-12. Report
# frequencies P(r) are exact marginals of the model.
import math

GRAN = [1, 5, 15, 30, 60]
SOURCES = {"memory": ([0.10, 0.15, 0.10, 0.25, 0.40], 15.0),
           "record": ([0.70, 0.20, 0.05, 0.03, 0.02], 3.0),
           "no-round": ([1.0, 0.0, 0.0, 0.0, 0.0], 15.0)}
B = 660


def cdf(x):
    return 0.5 * (1.0 + math.erf(x / math.sqrt(2.0)))


def hour_index(t):
    return int(((t + 60) % 1440) // 120)


def lik(r, t, g, s):
    if r % g:
        return 0.0
    d = r - (t + 0.5)
    h = g / 2.0
    return cdf((d + h) / s) - cdf((d - h) / s)


def flip_and_mass(r, weights, s):
    terms, wrong = [], []
    k = hour_index(r)
    for t in range(r - 240, r + 241):
        w = math.fsum(p * lik(r, t, g, s) for p, g in zip(weights, GRAN))
        terms.append(w)
        if hour_index(t) != k:
            wrong.append(w)
    z = math.fsum(terms)
    return math.fsum(wrong) / z, z / 1440.0


print("flip probability by signed distance d = report - 11:00 (minutes)")
print("   d    memory   record   no-round(s=15)")
for d in (-30, -20, -15, -10, -5, -1, 0, 1, 5, 10, 15, 20, 30):
    row = [flip_and_mass(B + d, *SOURCES[k])[0]
           for k in ("memory", "record", "no-round")]
    print("%+4d  " % d + "".join("%9.4f" % v for v in row))
print("one 120-minute cycle (repeats all day)")
print("source     P(|d|<=15)  flip in window  flip overall")
for k in ("memory", "record", "no-round"):
    win_m, win_f, all_m, all_f = [], [], [], []
    for r in range(B - 60, B + 60):
        f, m = flip_and_mass(r, *SOURCES[k])
        all_m.append(m)
        all_f.append(f * m)
        if abs(r - B) <= 15:
            win_m.append(m)
            win_f.append(f * m)
    pw, pa = math.fsum(win_m), math.fsum(all_m)
    print("%-9s  %9.4f  %13.4f  %12.4f"
          % (k, pw / pa, math.fsum(win_f) / pw, math.fsum(all_f) / pa))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《出生时刻记忆误差传播与鲁棒定盘、交界时辰盘面翻转概率与最少前事反推题数母典》第 2 章「交界前后一刻钟:翻转概率的精确计算」(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-284 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026284,
  author    = {Fan, Yafu},
  title     = {Error Propagation from Remembered Birth Times and Robust Chart Fixing: Flip Probabilities at Double-Hour Boundaries and the Fewest Past-Event Questions},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论