新皇极经世 (Neo Huangji Jingshi): Structural Priors from Ancient Chinese Mathematics — Twelve Laws Verified Through Deep Learning

Seven ancient Chinese texts (I Ching, Dao De Jing, Huangdi Neijing, Wuxing Dayi, Hetu Luoshu, Ziwei Doushu, Huangji Jingshi) contain mathematical structures that, when translated into neural network architectures, improve performance across classification, causal discovery, and time series prediction. We extracted 12 laws, ran 24 controlled experiments, and validated 12/12 (100%). Core finding: architecture-level constraints consistently outperform loss-level regularization. Includes: Nature-format manuscript (v3, 27 refs, 5 figures), 24 reproducible experiment scripts, 4 reusable Python modules (daoshu), cover letter, reporting summary, and suggested reviewers.