关于「出生时刻的估计理论:取整误差、序贯检验与「事后对上」」,《最少题数的信息论》给出的定量结论是什么?

结论:第六章把出生时刻校准当作估计问题:取整到整点的记录有 0.2500 的机会落在错的时段;判别力为 0.7 对 0.3 时,序贯检验在相邻两时段间平均仍需 9.35 题;题目毫无判别力时,十题八中的标准在十二个候选中偶然达标的概率是 0.4908。

数据口径:书中的价格、预算、题库与判别力参数均为示意;全部计算由书中程序生成;本书是非个人化经营建议;涉及出生时刻的部分只讨论估计理论与问卷设计,不评价任何体系的预测效度,本书不做任何个人占断。 第四章的模拟中,自适应五题的误差 0.3866 优于固定八题的 0.3969,按精度停止平均只需 4.33 题;信道容量给出任何问法都绕不过的题数下界。

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: 0.2500

rounding g (min):6030155
recall sd 0min:0.25000.12500.06250.0208
recall sd 10min:0.25000.12570.07960.0679
recall sd 20min:0.25130.15910.13930.1337
recall sd 40min:0.31750.27810.26880.2660
share of hourly records on a block boundary: 0.5000
查看完整验算程序
# Chapter 6. p61: how often a rounded or recalled clock time falls in
# a different two-hour block. Exact interval arithmetic, illustrative.
# 一天分 12 个两小时时段,分界在奇数整点(23 点、1 点、3 点等)。
# 真实时刻在一天内均匀;记录 = 四舍五入到 g 分钟,再加回忆误差
# N(0, s^2)(按 1 分钟格离散到正负 4 s)。落在分界线上的记录按惯例
# 归入后一时段。输出:记录所在时段与真实时段不同的概率。
import math
from statistics import NormalDist

DAY = 1440


def block(minute):
    return int(((minute + 60) % DAY) // 120)


def overlap(lo, hi, b):
    # 区间 [lo, hi) 与第 b 时段 [120b - 60, 120b + 60) 的重叠长度
    parts = []
    for shift in (-DAY, 0, DAY):
        a0, a1 = 120 * b - 60 + shift, 120 * b + 60 + shift
        parts.append(max(0.0, min(hi, a1) - max(lo, a0)))
    return math.fsum(parts)


def p_mismatch(g, s):
    if s == 0:
        offs = [(0, 1.0)]
    else:
        nd = NormalDist(0.0, s)
        span = int(4 * s)
        offs = [(d, nd.cdf(d + 0.5) - nd.cdf(d - 0.5))
                for d in range(-span, span + 1)]
        z = math.fsum(w for _, w in offs)
        offs = [(d, w / z) for d, w in offs]
    terms = []
    for k in range(DAY // g):
        rec = k * g
        lo, hi = rec - g / 2.0, rec + g / 2.0
        for d, w in offs:
            right = overlap(lo, hi, block(rec + d))
            terms.append(w * (g - right))
    return math.fsum(terms) / DAY


GS = (60, 30, 15, 5)
SS = (0, 10, 20, 40)
print("rounding g (min):   " + "".join("%-9d" % g for g in GS))
for s in SS:
    print("recall sd %-3d min:  " % s
          + "".join("%-9.4f" % p_mismatch(g, s) for g in GS))
print("share of hourly records on a block boundary: %.4f" % (12 / 24))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《费希尔信息矩阵、克拉美罗方差下界与预算及出生时刻探询最少题数母典》第 6 章「出生时刻的估计理论:取整误差、序贯检验与「事后对上」」(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-235 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026235,
  author    = {Fan, Yafu},
  title     = {Fisher Information Matrices and the Cramér-Rao Bound: the Fewest Questions Needed to Estimate a Budget or a Birth Time},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论