关于「从下单时间戳到序参量:折叠、瑞利检验与小样本偏差」,《直播间集体下单的临界耦合》给出的定量结论是什么?

结论:把下单时刻按口令周期折叠成相位,平均向量的长度就是数据里的序参量,方向就是订单相对口令的平均滞后。示意直播里,有节奏的甲场序参量 0.3249、滞后 48.3 秒,瑞利统计量 19.00;无节奏的乙场 0.0434,统计量 0.34;一百八十笔订单的噪声底是 0.0661。小样本会让序参量系统性偏高,二十笔订单时毫无节奏的直播平均也有 0.195,偏差校正 \(R^2_{adj}\) 能把它扣回零。扫描周期时要警惕谐波,并为十七个候选付出族错误率门槛 5.83 的代价。数据里测到的整齐,到底是口令推出来的,还是观众彼此带出来的?下一章把口令写成外力,把两者分开。

数据口径:书中的观众、节奏、耦合、口令强度、库存与订单均为示意参数;全部计算由书中程序生成;本书是非个人化经营建议,只讨论真实订单,紧迫感只能来自真实库存与真实截止,节律部分只讲同步数学,不作任何身心功效断言。 第八章给出统一倒数的 A/B 闭环、小工具「直播间同步序参量计算器」与预注册方案(示意参数下每组 26 场,模拟检验力 0.907)。

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: 0.0434, 0.0661, 0.3249, 0.34, 19.00, 48.3

first orders in A (s)6 26 33 46 55 62 86 87
session n R lag_s Z p
A call180 0.3249 48.3 19.00 3.50e-09
B free180 0.0434 5.7 0.34 7.13e-01
noise floor for n180: E[R] ~ 0.0661, p95 ~ 0.1290
design value of R in A0.35 * I1(4)/I0(4) = 0.3022
查看完整验算程序
# Chapter 4. p41: From order timestamps to an order parameter.
# Model: a 60-minute session in which the host gives the same call every
# 300 s. Each order time t (whole seconds) becomes a phase
#   theta = 2 pi (t mod 300) / 300,
# and R = |mean of exp(i theta)| is the sample order parameter. Rayleigh
# Z = n R^2; p-value by Zar's approximation
#   p = exp(sqrt(1 + 4n + 4(n^2 - (nR)^2)) - (1 + 2n)).
# Illustrative data, fixed seed: session A puts 35 % of 180 orders in a
# von Mises cluster (kappa 4) about 45 s after each call, the rest
# uniform; session B is 180 uniform orders (no common rhythm).
import cmath
import math
import random

PERIOD, LEN = 300, 3600


def session(n, share, rng):
    out = []
    for _ in range(n):
        if rng.random() < share:
            cyc = rng.randrange(LEN // PERIOD)
            ph = rng.vonmisesvariate(2 * math.pi * 45 / PERIOD, 4.0)
            t = cyc * PERIOD + PERIOD * ph / (2 * math.pi)
        else:
            t = rng.uniform(0, LEN)
        out.append(int(t) % LEN)
    return sorted(out)


def stats(times):
    n = len(times)
    zs = [cmath.exp(2j * math.pi * (t % PERIOD) / PERIOD) for t in times]
    m = complex(math.fsum(z.real for z in zs) / n,
                math.fsum(z.imag for z in zs) / n)
    r = abs(m)
    lag = (cmath.phase(m) % (2 * math.pi)) * PERIOD / (2 * math.pi)
    z = n * r * r
    p = math.exp(math.sqrt(1 + 4 * n + 4 * (n * n - (n * r) ** 2))
                 - (1 + 2 * n))
    return n, r, lag, z, min(1.0, p)


def bessel(order, x):
    terms = [(x / 2) ** (2 * k + order)
             / (math.factorial(k) * math.factorial(k + order))
             for k in range(40)]
    return math.fsum(terms)


rng = random.Random(20261117)
a = session(180, 0.35, rng)
b = session(180, 0.0, rng)
print("first orders in A (s):", " ".join(str(t) for t in a[:8]))
print("session  n    R       lag_s   Z       p")
for name, ts in (("A call", a), ("B free", b)):
    n, r, lag, z, p = stats(ts)
    print("%-7s  %d  %.4f  %5.1f   %6.2f  %.2e" % (name, n, r, lag, z, p))
print("noise floor for n = 180: E[R] ~ %.4f, p95 ~ %.4f"
      % (math.sqrt(math.pi / (4 * 180)), math.sqrt(math.log(20) / 180)))
print("design value of R in A: 0.35 * I1(4)/I0(4) = %.4f"
      % (0.35 * bessel(1, 4.0) / bessel(0, 4.0)))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《藏本耦合振子同步、百人直播间集体下单相变与声频节律夹带临界耦合母典》第 4 章「从下单时间戳到序参量:折叠、瑞利检验与小样本偏差」(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-239 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026239,
  author    = {Fan, Yafu},
  title     = {Kuramoto Coupled-Oscillator Synchrony: the Collective-Ordering Phase Transition in Hundred-Viewer Livestreams and the Critical Coupling of Rhythmic Entrainment},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论