What does Anti-Collapse Engineering with Two Anchors conclude, in numbers, about “Replace or Accumulate: the Counter-Evidence of Gerstgrasser and the Bound of pi^2/6”?
Answer: Chapter 3 sets out the contrary evidence of Gerstgrasser and colleagues: when real data are accumulated rather than replaced, the error of the mean in the Gaussian twin is capped by pi^2/6, that is 1.644934.
Scope of the numbers: Parameters are illustrative; all calculations are generated by the book's programs, which call no model interface and use no reader's or customer's data. Marked boxes separate published evidence, runnable Python calculations (22 programs), classical sources and the author's hypotheses.
Data table
The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. standard-library Python onlyprinted values that match the answer: 1.644934
| n | replace | accumulate | ratio |
|---|---|---|---|
| 0 | 1.0000 | 1.000000 | 1.0 |
| 1 | 2.0000 | 1.250000 | 1.6 |
| 10 | 11.0000 | 1.558032 | 7.1 |
| 100 | 101.0000 | 1.635082 | 61.8 |
| 1000 | 1001.0000 | 1.643936 | 608.9 |
| 10000 | 10001.0000 | 1.644834 | 6080.2 |
| accumulate limit pi^2/6 = 1.644934 | |||
| accumulate share of the limit at n = 10: 0.9472 | |||
| pool size under accumulate after n = 100: 101 x N |
Show the full program
# Chapter 3. p31: replace versus accumulate, exact mean-squared drift.
# Model: a normal law with known variance s^2 = 1 and unknown mean;
# each generation adds N points drawn from the current fit. Replace:
# the fit uses only the newest N points, so the error e_n of the mean
# is a random walk and E[e_n^2] = (n + 1) / N. Accumulate: the fit
# uses every point drawn so far; then e_n = e_{n-1} + eps_n / (n + 1)
# and E[e_n^2] = (1/N) * sum_{k=1}^{n+1} 1/k^2 < (pi^2/6) / N. This is
# the Gaussian twin of the linear-regression bound of Gerstgrasser et
# al. (2024). Values are N * E[e_n^2], free of N.
import math
GENS = (0, 1, 10, 100, 1000, 10000)
def replace(n):
return float(n + 1)
def accumulate(n):
return math.fsum(1.0 / (k * k) for k in range(1, n + 2))
print("n replace accumulate ratio")
for n in GENS:
r, a = replace(n), accumulate(n)
print("%-7d %-11.4f %-11.6f %.1f" % (n, r, a, r / a))
print("accumulate limit pi^2/6 = %.6f" % (math.pi ** 2 / 6.0))
print("accumulate share of the limit at n = 10: %.4f"
% (accumulate(10) / (math.pi ** 2 / 6.0)))
print("pool size under accumulate after n = 100: %d x N" % 101)Single source and citation
Fan, Y. (2026). Anti-Collapse Engineering with Two Anchors of Truth: the Variance-Collapse Rate of Self-Generated Training Data and Immunity through Standard-Library Recomputation and Textual Collation (Version 1.0.0), Chapter 3: Replace or Accumulate: the Counter-Evidence of Gerstgrasser and the Bound of pi^2/6. 5D Library. DOI pending
DOI pending · monograph page goes live after publication · 5DL-B-265 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026265,
author = {Fan, Yafu},
title = {Anti-Collapse Engineering with Two Anchors of Truth: the Variance-Collapse Rate of Self-Generated Training Data and Immunity through Standard-Library Recomputation and Textual Collation},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}Other chapter answers in this book
- Chapter 1 · Variance Collapsing Generation by Generation: the Expectation Formula of a Gaussian Feedback Chain and Almost-Sure Extinction
- Chapter 2 · The Tails Die First: Generational Extinction in a Discrete Law and the Absorbing States of Wright–Fisher
- Chapter 4 · The Threshold of Mixing in Real Sources: How Much Variance a Fixed Share Preserves, and How Much of the Tail It Misses
- Chapter 5 · Examining for Collapse: Why Inward-Looking Metrics Go Blind, and How Much Data an Outward-Looking Anchor Needs
- Chapter 6 · The First Anchor: Standard-Library Recomputation, Deterministic Fingerprints and the Correlated Failure of Two Anchors
- Chapter 7 · The Classics Examined: Duan Yucai on "the Difficulty of Settling Right and Wrong" and Erasmus's Return to the Original in 1516
- Chapter 8 · The Empirical Loop: a Two-Anchor Procedure for Every Volume, the Feedback Collapse Rate Calculator and a Pre-Registered Test