What does Elasticities across Scale conclude, in numbers, about “Estimating a Scaling Exponent from Scattered Points: Intervals, a Curvature Test and the Zeros that Get Dropped”?

Answer: Chapter 2 covers intervals, a curvature test and the bias from dropped zeros, which pushes a true exponent of 1 down by 0.135.

Scope of the numbers: Marked boxes separate published evidence, runnable Python calculations (21 programs), classical sources and the author's hypotheses.

中文版

Data table

The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. standard-library Python only

niches 30, sites in all 149, smallest 1, largest 15
beta_hat0.877 se = 0.051 R2 = 0.912 (true 0.85)
t interval0.772 to 0.982
bootstrap0.798 to 0.974 (2000 resamples)
per-site conversions at n16 versus n = 1: 0.711
elasticity 1 inside t interval: no

Key formula

\[\ln Y = a + \beta \ln n + \varepsilon ,\]
Show the full program
# Chapter 2. p21: the scaling exponent by log-log least squares.
# Model (illustrative): one owner runs 30 niches (keyword pools); niche
# j holds n_j sites, n_j drawn log-uniform from 1 to 16. Monthly
# conversions of the niche are Y = 40 n^0.85 exp(e), e ~ N(0, 0.25).
# We fit ln Y = a + beta ln n by least squares, report the standard
# error, a t interval (t = 2.048 for 28 degrees of freedom) and a
# bootstrap percentile interval from 2000 resamples. Seed 20261113.
import math
import random

rng = random.Random(20261113)
BETA, SIG, M = 0.85, 0.25, 30


def ols(xs, ys):
    n = len(xs)
    mx, my = math.fsum(xs) / n, math.fsum(ys) / n
    sxx = math.fsum((x - mx) ** 2 for x in xs)
    sxy = math.fsum((x - mx) * (y - my) for x, y in zip(xs, ys))
    b = sxy / sxx
    a = my - b * mx
    rss = math.fsum((y - a - b * x) ** 2 for x, y in zip(xs, ys))
    syy = math.fsum((y - my) ** 2 for y in ys)
    se = math.sqrt(rss / (n - 2) / sxx)
    return a, b, se, 1 - rss / syy


sites = [max(1, round(math.exp(rng.uniform(0, math.log(16)))))
         for _ in range(M)]
conv = [40 * n ** BETA * math.exp(rng.gauss(0, SIG)) for n in sites]
xs = [math.log(n) for n in sites]
ys = [math.log(y) for y in conv]
a, b, se, r2 = ols(xs, ys)
print("niches %d, sites in all %d, smallest %d, largest %d"
      % (M, int(math.fsum(sites)), min(sites), max(sites)))
print("beta_hat = %.3f  se = %.3f  R2 = %.3f  (true 0.85)"
      % (b, se, r2))
print("t interval    %.3f to %.3f" % (b - 2.048 * se, b + 2.048 * se))
boot = []
idx = list(range(M))
for _ in range(2000):
    pick = [rng.choice(idx) for _ in idx]
    bx = [xs[i] for i in pick]
    if max(bx) - min(bx) < 1e-9:
        continue
    boot.append(ols(bx, [ys[i] for i in pick])[1])
boot.sort()
lo, hi = boot[int(0.025 * len(boot))], boot[int(0.975 * len(boot))]
print("bootstrap     %.3f to %.3f  (%d resamples)" % (lo, hi, len(boot)))
print("per-site conversions at n = 16 versus n = 1: %.3f"
      % (16 ** (b - 1)))
print("elasticity 1 inside t interval: %s"
      % ("yes" if b + 2.048 * se >= 1 else "no"))

Single source and citation

Fan, Y. (2026). Renormalisation Group and Cross-Scale Scaling Laws: Which Conversion Elasticities Survive from a One-Person Workshop to a Hundred-Site Matrix (Version 1.0.0), Chapter 2: Estimating a Scaling Exponent from Scattered Points: Intervals, a Curvature Test and the Zeros that Get Dropped. 5D Library. DOI pending

DOI pending · monograph page goes live after publication · 5DL-B-248 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026248,
  author    = {Fan, Yafu},
  title     = {Renormalisation Group and Cross-Scale Scaling Laws: Which Conversion Elasticities Survive from a One-Person Workshop to a Hundred-Site Matrix},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

Other chapter answers in this book