forked from bai/curriculum-project-hub
sjfhsjfh
c73a2c903f
Real-content fixture for the end-to-end pipeline. TH-141 (39 parts: 22 segment, 15 lemma, 2 example) migrated from the prototype's typst `#let parts` manifest to ADR-0008: declarative manifest.toml (project+info+ordered parts+targets) + per-element element.toml (kind + scalars; examples carry `source`). Content .typ files copied byte-identical (no math corruption); per-element main.typ + meta.toml dropped (wiring is now generated). Part order matches source exactly; 5 lemmas have no proof.typ (optional); no cross-file imports / paralearn refs / figs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 lines
513 B
Typst
4 lines
513 B
Typst
= 收束：各模型横向对比
|
|
|
|
本章沿着"假设由少到多"的方向给出了多种模型：量纲分析、Stefan 极简估算、缺键模型一般式、Eötvös 规则、$sigma(T, p)$ 多项式拟合。模型的输入复杂度不同，对常温水 $sigma_(L G)$ 的预测精度也不同。把几种典型预测并排列出，与实测 $0.072 thin "N/m"$ 对照，可以看到一个略反预期的现象——最简单的微观模型并不是最准的。下面给出对照表并逐一讨论偏差来源。
|