MuLabPKU · PhotonYan · Nov 29, 2025 · Nov 29, 2025 · Dec 3, 2025 · Dec 18, 2025
diff --git a/.gitignore b/.gitignore
@@ -14,4 +14,9 @@ notes/2025/tikz/4
 notes/2025/mvp/chapters/4-nn.tex
 notes/2025/mvp/chapters/4-nn.pdf
 
-0-ac*
+0-ac*
+
+logs/
+movs/
+
+*.png
diff --git a/docs/superpowers/specs/2026-07-11-a5-chinese-notes-design.md b/docs/superpowers/specs/2026-07-11-a5-chinese-notes-design.md
@@ -0,0 +1,25 @@
+# A5 中文讲义副本设计
+
+## 目标
+
+在不修改现有 A4 中文版 `notes/2025/mvp-cn` 的前提下，创建独立的 A5 竖版副本 `notes/2025/mvp-cn-a5`。A5 版必须保留完整内容、公式、插图、标签与交叉引用，并形成可直接阅读和打印的 PDF。
+
+## 选定版式
+
+采用用户确认的“均衡版”：纸张尺寸为 A5（148 × 210 mm），正文约 10 pt，页边距约 13–14 mm。目标是在小开本上保持舒适阅读，同时避免紧凑版的拥挤感和大字版过多的分页。
+
+## 实现方式
+
+完整复制当前中文版本，然后只在新副本中调整主文档的纸张和页面几何参数。正文必须由 LaTeX 针对 A5 页面重新换行和分页，不采用把 A4 PDF 等比例缩小的方式。封面缩放到 A5 页面；章节源码和本地化插图默认保持不变。
+
+如果完整构建发现超宽公式、表格、算法或插图，仅在 A5 副本中做局部、最小化的版式修正。修正不得删改内容、改变数学含义或反向影响 A4 中文版。
+
+## 文件边界
+
+源版本为 `notes/2025/mvp-cn`，目标版本为 `notes/2025/mvp-cn-a5`。A4 中文版、英文版、2024 notes 和 slides 均不在修改范围内。
+
+## 验证标准
+
+使用 XeLaTeX/latexmk 完整构建 A5 主文档。最终 PDF 的页面尺寸必须由 `pdfinfo` 验证为 A5；构建日志不得包含 LaTeX 错误、未定义引用、缺字或超出版心。渲染并视觉检查封面、目录、各章首页、公式密集页、算法页与插图页，确保无裁切、重叠、过小文字或孤立标题。
+
+源版与 A5 版的 LaTeX 环境、标签、引用和插图集合必须保持一致；允许的差异仅限页面几何和必要的 A5 局部排版命令。
diff --git a/notes/2024/linear-regression.md → notes/2024/1-linear-regression.md b/notes/2024/linear-regression.md → notes/2024/1-linear-regression.md
diff --git a/notes/2024/logistic-regression.md → notes/2024/2-logistic-regression.md b/notes/2024/logistic-regression.md → notes/2024/2-logistic-regression.md
diff --git a/notes/2024/svm.md → notes/2024/3-svm.md b/notes/2024/svm.md → notes/2024/3-svm.md
diff --git a/notes/2024/representer-theorem.md → notes/2024/4-representer-theorem.md b/notes/2024/representer-theorem.md → notes/2024/4-representer-theorem.md
diff --git a/notes/2024/learning-theory.md → notes/2024/5-learning-theory.md b/notes/2024/learning-theory.md → notes/2024/5-learning-theory.md
diff --git a/notes/2024/gaussian-process.md → notes/2024/6-gaussian-process.md b/notes/2024/gaussian-process.md → notes/2024/6-gaussian-process.md
diff --git a/notes/2024/tree-models.md → notes/2024/7-tree-models.md b/notes/2024/tree-models.md → notes/2024/7-tree-models.md
diff --git a/notes/2024/ensemble-learning.md → notes/2024/8-ensemble-learning.md b/notes/2024/ensemble-learning.md → notes/2024/8-ensemble-learning.md
diff --git a/notes/2024/unsupervised-learning.md → notes/2024/9-unsupervised-learning.md b/notes/2024/unsupervised-learning.md → notes/2024/9-unsupervised-learning.md
diff --git a/notes/2025/fig/unet.pdf b/notes/2025/fig/unet.pdf
diff --git a/notes/2025/mvp/appendix/1-la.tex b/notes/2025/mvp/appendix/1-la.tex
@@ -0,0 +1,4 @@
+\documentclass[../main]{subfiles}
+\begin{document}
+\chapter{Linear Algebra}
+\end{document}
diff --git a/notes/2025/mvp/appendix/2-info.tex b/notes/2025/mvp/appendix/2-info.tex
@@ -0,0 +1,4 @@
+\documentclass[../main]{subfiles}
+\begin{document}
+\chapter{Information Theory}
+\end{document}
diff --git a/notes/2025/mvp/appendix/3-convex.tex b/notes/2025/mvp/appendix/3-convex.tex
@@ -0,0 +1,4 @@
+\documentclass[../main]{subfiles}
+\begin{document}
+\chapter{Convex Analysis}
+\end{document}
diff --git a/notes/2025/mvp/appendix/4-diffgeo.tex b/notes/2025/mvp/appendix/4-diffgeo.tex
@@ -0,0 +1,4 @@
+\documentclass[../main]{subfiles}
+\begin{document}
+\chapter{Differential Geometry}
+\end{document}
diff --git a/notes/2025/mvp/appendix/5-lie.tex b/notes/2025/mvp/appendix/5-lie.tex
@@ -0,0 +1,4 @@
+\documentclass[../main]{subfiles}
+\begin{document}
+\chapter{Lie Groups and Lie Algebras}
+\end{document}
diff --git a/notes/2025/mvp/appendix/6-mt.tex b/notes/2025/mvp/appendix/6-mt.tex
@@ -0,0 +1,4 @@
+\documentclass[../main]{subfiles}
+\begin{document}
+\chapter{Measure Theory}
+\end{document}
diff --git a/notes/2025/mvp/appendix/7-pt.tex b/notes/2025/mvp/appendix/7-pt.tex
@@ -0,0 +1,4 @@
+\documentclass[../main]{subfiles}
+\begin{document}
+\chapter{Probability Theory}
+\end{document}
diff --git a/notes/2025/mvp/appendix/8-sde.tex b/notes/2025/mvp/appendix/8-sde.tex
@@ -0,0 +1,4 @@
+\documentclass[../main]{subfiles}
+\begin{document}
+\chapter{Stochastic Optimization and Stochastic Differential Equations}
+\end{document}
diff --git a/notes/2025/mvp/appendix/9-ie.pdf b/notes/2025/mvp/appendix/9-ie.pdf
diff --git a/notes/2025/mvp/appendix/9-ie.tex b/notes/2025/mvp/appendix/9-ie.tex
@@ -0,0 +1,4 @@
+\documentclass[../main]{subfiles}
+\begin{document}
+\chapter{Concentration Inequalities}
+\end{document}
diff --git a/notes/2025/mvp/chapters/1-lr.pdf b/notes/2025/mvp/chapters/1-lr.pdf
diff --git a/notes/2025/mvp/chapters/1-lr.tex b/notes/2025/mvp/chapters/1-lr.tex
@@ -15,12 +15,10 @@ \section{Basic Knowledge of Linear Regression}
   \item Dataset: $D=\{(x_i,y_i)\}_{i=1}^n$, where $x_i\in\mathbb{R}^d$ and $y_i\in\mathbb{R}$.  
   Here, $y_i$ denotes the regression target, while $x_i$ represents the input features used to predict $y_i$.
 
-  \item Linear Model: $f(x)=w^{\top}x+b$, with weight $w\in\mathbb{R}^d$ and bias $b\in\mathbb{R}$.  
   \item Linear Model: $f(x)=w^{\top}x+b$, with weight $w\in\mathbb{R}^d$ and bias $b\in\mathbb{R}$.  
   \begin{note}
     This definition is equivalent to an inner product:  
     $\hat y = w^{\top}x+b.$
-    $\hat y = w^{\top}x+b.$
   \end{note}
 
   \begin{definition}[Learnable / Trainable Parameters]

diff --git a/notes/2025/mvp/chapters/10-vae.pdf b/notes/2025/mvp/chapters/10-vae.pdf
-Original file line number
+Diff line change
@@ Expand Up / @@ -14,4 +14,9 @@ notes/2025/tikz/4 @@
     notes/2025/mvp/chapters/4-nn.tex
     notes/2025/mvp/chapters/4-nn.pdf
--ac*
+-ac*
+    logs/
+    movs/
+    *.png