关于「格上的超模函数:托普基斯单调比较静态与单项试点的陷阱」,《存量资产的超模串联》给出的定量结论是什么?
结论:第二章讲格上的超模函数与托普基斯单调比较静态:三项拉新资产造成 239 对超模违反,替代关系会让最优组合随条件「换人」;整合成本较高的互补接入会形成山谷,单项试点各自亏损,联合接入多赚 963。
数据口径:在示意模型里,六项资产两两配对的十五对中有十二对互补、三对替代,全部打开时月度价值 23,104,各自相加只有 13,668。 书中参数为示意,不对应任何真实业务、客户、达人或成交;全部计算由书中程序生成。
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: 239
| A. supermodularity audit (incomparable pairs only) | |
| ground set subsets pairs violations worst gap | |
| all six | 64 1,351 239 -3,251 |
| drop D | 32 285 26 -1,947 |
| drop A | 32 285 22 -1,113 |
| drop A and D | 16 55 0 0 |
| B. increasing differences in the order rate c0 | |
| smallest change in a marginal value when c0 rises | 76.8 |
| (> | 0 means every marginal value rises with c0) |
查看完整验算程序
# Chapter 2. p21: is the asset value function supermodular on the lattice?
# Lattice: all 64 subsets of the six assets of p11, ordered by inclusion,
# join = union, meet = intersection. f is supermodular when
# f(S | T) + f(S & T) >= f(S) + f(T) for every pair S, T.
# We count violations over all unordered incomparable pairs, then repeat
# on sublattices that drop one or both of the reach substitutes A and D.
# Model and parameters are the illustrative ones of p11.
import itertools
import math
N, V, C0 = 40000, 30.0, 0.01
R = [0.30, 0.10, 0.35, 0.02, 0.20, 0.08]
M = [1.10, 1.35, 1.00, 1.60, 1.00, 1.15]
CODES = "LKAFDT"
def f(mask):
idx = [i for i in range(6) if mask >> i & 1]
if not idx:
return 0.0
miss = math.prod(1.0 - R[i] for i in idx)
return N * V * C0 * (1.0 - miss) * math.prod(M[i] for i in idx)
def audit(ground):
sets = []
for k in range(len(ground) + 1):
for c in itertools.combinations(ground, k):
m = 0
for i in c:
m |= 1 << i
sets.append(m)
pairs = bad = 0
worst = 0.0
for a, b in itertools.combinations(sets, 2):
if a & b in (a, b):
continue
pairs += 1
gap = f(a | b) + f(a & b) - f(a) - f(b)
if gap < -1e-9:
bad += 1
worst = min(worst, gap)
return len(sets), pairs, bad, worst
if __name__ == "__main__":
print("A. supermodularity audit (incomparable pairs only)")
print(" ground set subsets pairs violations worst gap")
cases = [("all six", range(6)), ("drop D", (0, 1, 2, 3, 5)),
("drop A", (0, 1, 3, 4, 5)), ("drop A and D", (0, 1, 3, 5))]
for name, g in cases:
n, p, b, w = audit(tuple(g))
print(" {:<16} {:>8} {:>6,} {:>11} {:>10,.0f}".format(
name, n, p, b, w))
print("B. increasing differences in the order rate c0")
worst = None
for lo, hi in ((0.008, 0.010), (0.010, 0.012)):
for a in range(64):
for i in range(6):
if a >> i & 1:
continue
d = (f(a | 1 << i) - f(a)) / C0
gain = d * (hi - lo)
worst = gain if worst is None else min(worst, gain)
print(" smallest change in a marginal value when c0 rises: {:,.1f}"
.format(worst))
print(" (>= 0 means every marginal value rises with c0)")唯一数据源与引用
范雅福(Fan, Y.). (2026). 《存量武器库超模组合代数、邻近可能联想图谱与跨业务火力串联母典》第 2 章「格上的超模函数:托普基斯单调比较静态与单项试点的陷阱」(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-375 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026375,
author = {Fan, Yafu},
title = {Supermodular Weapon-Arsenals, Adjacent-Possible Recombination Graphs and Cross-Asset Cashflow Chaining},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}