关于「必要多样性:剪枝的信息预算何时盖过立项噪音」,《母机收敛判据》给出的定量结论是什么?

结论:第二章逐一枚举艾什比的调节博弈,并把收敛条件写成噪音流入不超过剪枝容量,敏感度扫描中模拟与理论 56 格全部吻合。

数据口径:书中母机参数均为示意,不涉及任何服务器、域名、密钥或生产信息;全部计算由书中程序生成;本书是非个人化的经营分析,不构成经营建议。 第四章用卡恩算法排序、用强连通分量查环修环,三百卷合成引用图中可比对为 18,828 对,占 0.420。

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。 与结论吻合的打印值: 56

eps = 0.25, g = 0.5; o = converges, X = grows (80 quarters)
  K \ a    8  12  16  20  24  32  40  48
     2     o   X   X   X   X   X   X   X
     3     o   o   X   X   X   X   X   X
     4     o   o   o   X   X   X   X   X
     6     o   o   o   o   o   X   X   X
     8     o   o   o   o   o   o   X   X
    10     o   o   o   o   o   o   o   X
    12     o   o   o   o   o   o   o   o
simulation agrees with eps*a <= K in 56 of 56 cells
a = 24, K = 6, g = 0.5: noise share sweep
   eps  inflow  N@40    N@80   verdict
  0.15     3.6    7.2     7.2   converges
  0.20     4.8    9.6     9.6   converges
  0.25     6.0   12.0    12.0   converges
  0.30     7.2   57.0   105.0   grows
  0.40     9.6  151.2   295.2   grows
查看完整验算程序
# Chapter 2. p23: sensitivity scan of the convergence boundary.
# Model: the stock N of noise rules (wrong rules still live) obeys
#   N[t+1] = N[t] + eps * a - min(K, g * N[t])
# a proposals per quarter, noise share eps, pruning gain g (share of
# the noise stock found per quarter), pruning capacity K (rules that
# can be retired per quarter). Theory: N converges to eps * a / g iff
# eps * a <= K; otherwise it grows by eps * a - K every quarter.
# Grid values are illustrative.


def simulate(a, eps, g, k, quarters):
    n, path = 0.0, []
    for _ in range(quarters):
        n = n + eps * a - min(k, g * n)
        path.append(n)
    return path


def bounded(path):
    return abs(path[-1] - path[-2]) < 1e-6


if __name__ == "__main__":
    eps, g = 0.25, 0.5
    aa = (8, 12, 16, 20, 24, 32, 40, 48)
    ks = (2, 3, 4, 6, 8, 10, 12)
    print("eps = {}, g = {}; o = converges, X = grows (80 quarters)"
          .format(eps, g))
    print("  K \ a " + "".join("{:>4}".format(a) for a in aa))
    agree, cells = 0, 0
    for k in ks:
        row = ""
        for a in aa:
            ok = bounded(simulate(a, eps, g, k, 80))
            row += "{:>4}".format("o" if ok else "X")
            agree += int(ok == (eps * a <= k))
            cells += 1
        print("  {:>4}  {}".format(k, row))
    print("simulation agrees with eps*a <= K in {} of {} cells".format(
        agree, cells))
    print("a = 24, K = 6, g = 0.5: noise share sweep")
    print("   eps  inflow  N@40    N@80   verdict")
    for e in (0.15, 0.20, 0.25, 0.30, 0.40):
        p = simulate(24, e, g, 6, 80)
        print("  {:.2f}  {:>6.1f}  {:>5.1f}  {:>6.1f}   {}".format(
            e, e * 24, p[39], p[79],
            "converges" if bounded(p) else "grows"))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《三百卷定理依赖图拓扑排序与母机收敛判据、立项噪音与剪枝增益的闭环停机条件母典》第 2 章「必要多样性:剪枝的信息预算何时盖过立项噪音」(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-332 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026332,
  author    = {Fan, Yafu},
  title     = {Topological Ordering of the Theorem-Dependency Graph of Three Hundred Volumes and a Convergence Criterion for the Self-Evolving Engine: Halting Conditions between Proposal Noise and Pruning Gain},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论