关于「规则即区间:阈值冲突的机械判定」,《冲突多半是冗余》给出的定量结论是什么?

结论:第二章说明一维阈值一次扫描即可判定,两量约束须做负环检测,浮点会在 0.1151 的边界配对上造出假冲突。

数据口径:书中规则库与数值均为示意;全部计算由书中程序生成;本书属非个人化经营方法,不构成法律、合规或经营决策意见,规则的增删须由主事者人工放行。 第一章正面陈述并量化命题:一千条示意规则里,标红配对中真矛盾只占 0.0392,裁决后 971 条是冗余,只剩 18 条约束性规则;越界规则占一成时真矛盾比例也只有 0.2667,但无解指标已遍及全部。

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: 0.1151

A. lower bound a + b against upper bound c = a + b, 9801 pairs
float says conflict (false)1128 share 0.1151
float says slack below c978 share 0.0998
exact Fraction says conflict0
B. percent / 100 differs from the written decimal261 of 999
C. three-term bounds 0.01 to 0.49, 117649 triples
naive float addition disagrees31633 share 0.2689
math.fsum of the floats19085 share 0.1622
查看完整验算程序
# Chapter 2. p23: binary floating point invents and hides conflicts
# Model: two rules meet exactly on a boundary, "x >= a + b" (a bound
# computed from two written decimals) and "x <= c" with c = a + b in
# decimal. Exact arithmetic says the pair is feasible (a single
# point). Binary floats may say a + b > c (a false conflict). Part A
# enumerates every pair a, b of two-decimal values from 0.01 to 0.99.
# Part B converts one-decimal percentages 0.1 to 99.9 by dividing by
# 100 and compares with the decimal written directly, the way two
# kits write the same threshold. Part C: three-term bounds a + b + c
# on a coarse grid, naive left-to-right float addition against
# math.fsum of the same three floats. Exhaustive, no randomness;
# Fraction is the judge.

import math
from fractions import Fraction as F

false_conf = 0
false_slack = 0
total = 0
for i in range(1, 100):
    for j in range(1, 100):
        total += 1
        a, b = "0.%02d" % i, "0.%02d" % j
        c = F(i + j, 100)
        fa = float(a) + float(b)
        fc = float(c)
        if fa > fc:
            false_conf += 1
        elif fa < fc:
            false_slack += 1
print("A. lower bound a + b against upper bound c = a + b, %d pairs"
      % total)
print("   float says conflict (false)      %5d  share %.4f"
      % (false_conf, false_conf / total))
print("   float says slack below c         %5d  share %.4f"
      % (false_slack, false_slack / total))
print("   exact Fraction says conflict         0")

diff = 0
for k in range(1, 1000):
    pct = "%d.%d" % divmod(k, 10)
    via_pct = float(pct) / 100
    direct = float(F(k, 1000))
    if via_pct != direct:
        diff += 1
print("B. percent / 100 differs from the written decimal: %d of 999"
      % diff)

bad3 = 0
badf = 0
n3 = 0
for i in range(1, 50):
    for j in range(1, 50):
        for k in range(1, 50):
            n3 += 1
            t = [float(F(i, 100)), float(F(j, 100)), float(F(k, 100))]
            want = float(F(i + j + k, 100))
            if (t[0] + t[1]) + t[2] != want:
                bad3 += 1
            if math.fsum(t) != want:
                badf += 1
print("C. three-term bounds 0.01 to 0.49, %d triples" % n3)
print("   naive float addition disagrees %6d  share %.4f"
      % (bad3, bad3 / n3))
print("   math.fsum of the floats        %6d  share %.4f"
      % (badf, badf / n3))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《跨卷规则消解与公理归约、千条机器规则阈值冲突检测与最小原语集母典》第 2 章「规则即区间:阈值冲突的机械判定」(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-258 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026258,
  author    = {Fan, Yafu},
  title     = {Cross-Volume Rule Resolution and Axiom Reduction: Detecting Threshold Conflicts among a Thousand Machine Rules and Finding a Minimal Primitive Set},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论