关于「硬的代价:模式宽度、误拦与被关掉的门禁」,《硬门禁与软提示衰减》给出的定量结论是什么?

结论:第六章讲误拦的代价:覆盖 0.97 的宽规则集一年只活 15.52 周,全年漏网 354.64 次。

数据口径:软提醒按比例起作用,各层独立时残余违规率是漏过率之积,示意参数下四层之后仍是 0.0048,不是零;只有确定性的代码门禁能在它自己定义的模式类别之内做到零。 书中参数均为示意;全部计算由书中程序生成。

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: 0.97, 15.52, 354.64

never switched off, escapes per year:
  S1  c 0.60  wrong blocks/week   1.99  escapes 199.68
  S2  c 0.80  wrong blocks/week   7.96  escapes  99.84
  S3  c 0.92  wrong blocks/week  29.86  escapes  39.94
  S4  c 0.97  wrong blocks/week 119.42  escapes  14.98
beta = 0.0005:
  S1  h/week 0.0010  weeks on 50.65  escapes 207.44
  S2  h/week 0.0040  weeks on 46.88  escapes 139.18
  S3  h/week 0.0148  weeks on 35.90  escapes 182.17
  S4  h/week 0.0580  weeks on 15.52  escapes 354.64
  fewest escapes: S2
beta = 0.0020:
  S1  h/week 0.0040  weeks on 46.88  escapes 229.19
  S2  h/week 0.0158  weeks on 35.08  escapes 229.77
  S3  h/week 0.0580  weeks on 15.52  escapes 362.10
  S4  h/week 0.2125  weeks on  3.71  escapes 464.68
  fewest escapes: S1
查看完整验算程序
# Chapter 6. p62: a gate that blocks too much gets switched off.
# Author's model (a hypothesis): each week the people running the work
# disable or loosen the gate with probability h = 1 - exp(-beta x F),
# where F is the number of wrong blocks that week; once off it stays
# off for the rest of the year. N = 2000 saves a week, r = 0.0048,
# beta = 0.0005 or 0.002 (illustrative); (c, f) of the sets from p61.
# Expected yearly escapes = sum over 52 weeks of
#   P(on) x N r (1 - c) + P(off) x N r.
import math

N, R, W = 2000, 0.0048, 52
SETS = [("S1", 0.60, 0.001), ("S2", 0.80, 0.004), ("S3", 0.92, 0.015),
        ("S4", 0.97, 0.060)]
print("never switched off, escapes per year:")
for name, c, f in SETS:
    print("  %s  c %.2f  wrong blocks/week %6.2f  escapes %6.2f" % (
        name, c, N * (1 - R) * f, W * N * R * (1 - c)))
for beta in (0.0005, 0.002):
    print("beta = %.4f:" % beta)
    best = None
    for name, c, f in SETS:
        F = N * (1 - R) * f
        h = 1 - math.exp(-beta * F)
        on = [(1 - h) ** (w + 1) for w in range(W)]
        esc = math.fsum(p * N * R * (1 - c) + (1 - p) * N * R
                        for p in on)
        if best is None or esc < best[0]:
            best = (esc, name)
        print("  %s  h/week %.4f  weeks on %5.2f  escapes %6.2f" % (
            name, h, math.fsum(on), esc))
    print("  fewest escapes: %s" % best[1])

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《客户端生命周期硬门禁工程、软提示遵从率轮次衰减与保存前物理拦截状态机母典》第 6 章「硬的代价:模式宽度、误拦与被关掉的门禁」(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-256 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026256,
  author    = {Fan, Yafu},
  title     = {Hard Gates across the Client Lifecycle: the Decay of Soft-Prompt Compliance over Turns and State Machines for Interception before Publication},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论