关于「报告时刻的堆积:整点、半点与取整混合模型」,《出生时刻误差与交界翻转》给出的定量结论是什么?
结论: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。
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。
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: 0.0083, 0.2782
| minute-of-hour class shares, 60000 reports per source | |||||
| class | expected | memory | ratio | record | ratio |
| :00 | 0.0167 | 0.5662 | 33.97 | 0.0764 | 4.58 |
| :30 | 0.0167 | 0.1625 | 9.75 | 0.0575 | 3.45 |
| :15/:45 | 0.0333 | 0.0784 | 2.35 | 0.0792 | 2.38 |
| other x5 | 0.1333 | 0.1136 | 0.85 | 0.2278 | 1.71 |
| rest | 0.8000 | 0.0794 | 0.10 | 0.5590 | 0.70 |
| memory reports exactly on an odd hour (a boundary): 0.2782 | |||||
| same share with no rounding at all: 0.0083 |
查看完整验算程序
# Chapter 1. p11: a rounding-mixture model makes reported times heap.
# Model: the true clock time T is uniform over the day (minutes); the
# report adds a recall error e ~ Normal(0, s) and then rounds T + e to
# the nearest g minutes, where g is drawn from a mixture over
# {1, 5, 15, 30, 60}. Two sources, both with illustrative weights (not
# measured): a parent's memory and a hospital record. Fixed-seed Monte
# Carlo; we print the share of reports by minute-of-hour class and a
# heaping ratio (observed share / share expected with no rounding).
import math
import random
GRAN = [1, 5, 15, 30, 60]
# 示意参数:父母回忆与医院记录两种来源的取整权重与回忆误差(分钟)
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),
}
N = 60000
def report(rng, weights, s):
t = rng.random() * 1440.0
g = rng.choices(GRAN, weights)[0]
x = t + rng.gauss(0.0, s)
r = int(math.floor(x / g + 0.5)) * g
return t, r % 1440
def minute_class(m):
if m == 0:
return ":00"
if m == 30:
return ":30"
if m in (15, 45):
return ":15/:45"
if m % 5 == 0:
return "other x5"
return "rest"
# 无取整时各类分钟的理论份额
EXPECT = {":00": 1 / 60, ":30": 1 / 60, ":15/:45": 2 / 60,
"other x5": 8 / 60, "rest": 48 / 60}
CLASSES = [":00", ":30", ":15/:45", "other x5", "rest"]
rng = random.Random(20261113)
print("minute-of-hour class shares, %d reports per source" % N)
print("class expected memory ratio record ratio")
counts = {}
for name in ("memory", "record"):
w, s = SOURCES[name]
c = dict.fromkeys(CLASSES, 0)
for _ in range(N):
_, r = report(rng, w, s)
c[minute_class(r % 60)] += 1
counts[name] = c
for k in CLASSES:
a = counts["memory"][k] / N
b = counts["record"][k] / N
print("%-9s %7.4f %6.4f %6.2f %6.4f %6.2f"
% (k, EXPECT[k], a, a / EXPECT[k], b, b / EXPECT[k]))
odd = 0
for _ in range(N):
_, r = report(rng, *SOURCES["memory"])
if r % 120 == 60:
odd += 1
print("memory reports exactly on an odd hour (a boundary): %.4f" % (odd / N))
print("same share with no rounding at all: %.4f" % (1 / 120))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 1 章「报告时刻的堆积:整点、半点与取整混合模型」(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}
}