关于「双轨转换:英文可追溯率约束下的最大留白」,《字源、意合与中英双轨》给出的定量结论是什么?
结论:全部写明与全部并置之间,存在一条可以计算的前沿。在可追溯率不低于 0.90 的约束下,温庭筠一联最多保留 0.314 的留白,《论语》开篇一章保留 0.371。分句的效果 s 决定前沿的位置,s 取 0.7 时留白可达 0.514。最优方案正好压在红线上,扰动之下只有 0.598 的概率守住红线,规划时要留余量。
数据口径:以今律古按示意参数每千字有 1.307 次期望误读,错句只有三句时抽检须读全稿的 0.632。 书中参数为示意,全部计算由书中程序生成(23 段程序)。
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: 0.314, 0.514, 0.598, 0.7, 0.90
| threshold | s=0.3 | s=0.5 | s=0.7 |
| 0.80 | 0.571 | 0.629 | 0.771 |
| 0.85 | 0.429 | 0.486 | 0.657 |
| 0.90 | 0.286 | 0.314 | 0.514 |
| 0.95 | 0.086 | 0.086 | 0.171 |
| nominal plan CSSCCJC under perturbation: P(trace>=0.90) 0.598 | |||
| trace 5th pct 0.874, median 0.905, 95th pct 0.934 |
查看完整验算程序
# Chapter 5. p53: the trace-gap frontier and how fragile it is
# Part A: for the Wen couplet modes (p41), the largest gap kept at each
# trace threshold, for three values of the semicolon uplift s (the share
# of residual confusion an independent-clause rendering removes; p51
# uses 0.5). Part B: perturb every modal share by a multiplicative
# log-normal factor (sigma 0.15, capped at 0.95), 2,000 fixed-seed
# draws, and count how often the nominal 0.90 plan still reaches 0.90.
import math
import random
from itertools import product
MODES = [0.45, 0.60, 0.70, 0.65, 0.60, 0.80, 0.55]
GAP = {"J": 1.0, "S": 0.6, "C": 0.0}
def trace(dev, m, s):
if dev == "J":
return m
if dev == "S":
return m + s * (1 - m)
return 0.97
def frontier(modes, tmin, s):
best = None
for plan in product("JSC", repeat=len(modes)):
t = math.fsum(trace(d, m, s) for d, m in zip(plan, modes))
t /= len(modes)
if t + 1e-12 < tmin:
continue
g = math.fsum(GAP[d] for d in plan) / len(modes)
if best is None or g > best[0] + 1e-12:
best = (g, "".join(plan))
return best
print("threshold s=0.3 s=0.5 s=0.7")
for tmin in (0.80, 0.85, 0.90, 0.95):
cells = []
for s in (0.3, 0.5, 0.7):
g, _ = frontier(MODES, tmin, s)
cells.append("{:.3f}".format(g))
print(" {:.2f} {}".format(tmin, " ".join(cells)))
rng = random.Random(38253)
plan = "CSSCCJC"
ok = 0
ts = []
for _ in range(2000):
mm = [min(0.95, m * math.exp(rng.gauss(0.0, 0.15))) for m in MODES]
t = math.fsum(trace(d, m, 0.5) for d, m in zip(plan, mm)) / 7
ts.append(t)
if t >= 0.90:
ok += 1
ts.sort()
print("nominal plan {} under perturbation: P(trace>=0.90) {:.3f}"
.format(plan, ok / 2000))
print("trace 5th pct {:.3f}, median {:.3f}, 95th pct {:.3f}".format(
ts[99], ts[999], ts[1899]))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《古汉语字源本义训诂、意合留白拓扑与中西意境张力转化母典》第 5 章「双轨转换:英文可追溯率约束下的最大留白」(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-382 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026382,
author = {Fan, Yafu},
title = {Classical Chinese Etymological Philology, Paratactic Imagery Topology and East-West Aesthetic Asymmetry in High-Authority Writing},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}