math-skill 3.0.1 → 3.1.0

package/knowledge-base/lie-theory/lie-algebra.en.md

@@ -45,7 +45,22 @@ High GPU friendliness. The core advantage of the Lie algebra is that it is a "li
 
 ## Further References
 
-- Distillation notes: references/books/micro-lie-theory.md (Section II-C Tangent Space and Lie Algebra, hat/vee operators)
-- Distillation notes: references/books/differential-geometry.md (Ch 5 Lie Groups, Lie algebra section)
+- Distillation notes: ../../references/books/micro-lie-theory.en.md (Section II-C Tangent Space and Lie Algebra, hat/vee operators)
+- Distillation notes: ../../references/books/differential-geometry.en.md (Ch 5 Lie Groups, Lie algebra section)
 - Original text: Joan Sola et al., *A micro Lie theory*, Section II-C (Lie algebra definition and hat/vee), Section II-D (exp/log bridge)
 - Original text: Jeffrey M. Lee, *Manifolds and Differential Geometry*, Ch 5 (Lie groups and Lie algebras)
+
+
+## Routing Extensions
+- If the corresponding global group is needed -> `lie-group.en.md` (integrating Lie algebra yields Lie group)
+- If algebra representations are needed -> `representation.en.md` (representation theory of Lie algebras)
+- If acting as the tangent space of a group -> `../differential-geometry/tangent-space.md` (Lie algebra structure on tangent space)
+
+## Extensible Directions
+- Structure constants: components of Lie bracket in a basis
+- Jacobi identity: fundamental axiom of Lie algebras
+- Ideal / subalgebra: substructures of Lie algebras
+- Nilpotent / solvable / semisimple classification: structure theorems for Lie algebras
+- Killing form: invariant bilinear form of Lie algebras
+- Cartan subalgebra: maximal toral subalgebra of semisimple Lie algebras
+- Root system: root system classification of semisimple Lie algebras

package/knowledge-base/lie-theory/lie-algebra.md

@@ -45,7 +45,22 @@ GPU 友好度高。李代数的核心优势是"线性空间"：
 
 ## 深入参考
 
-- 蒸馏稿：references/books/micro-lie-theory.md（§II-C 切空间与李代数, hat/vee 算子）
-- 蒸馏稿：references/books/differential-geometry.md（Ch 5 Lie Groups, 李代数部分）
+- 蒸馏稿：../../references/books/micro-lie-theory.md（§II-C 切空间与李代数, hat/vee 算子）
+- 蒸馏稿：../../references/books/differential-geometry.md（Ch 5 Lie Groups, 李代数部分）
 - 原书：Joan Sola et al., *A micro Lie theory*, §II-C（李代数定义与 hat/vee）、§II-D（exp/log 桥梁）
 - 原书：Jeffrey M. Lee, *Manifolds and Differential Geometry*, Ch 5（李群与李代数）
+
+
+## 路由扩展
+- 若需要对应的全局群 → `lie-group.md`（李代数积分得到李群）
+- 若需要代数表示 → `representation.md`（李代数的表示理论）
+- 若作为群的切空间 → `../differential-geometry/tangent-space.md`（切空间的李代数结构）
+
+## 可扩展方向
+- 结构常数（structure constants）：李括号在基下的分量
+- Jacobi 恒等式：李代数的基本公理
+- 理想/子代数（ideal / subalgebra）：李代数的子结构
+- 幂零/可解/半单分类（nilpotent / solvable / semisimple）：李代数的结构定理
+- Killing 型（Killing form）：李代数的不变双线性型
+- Cartan 子代数（Cartan subalgebra）：半单李代数的极大环面子代数
+- 根系（root system）：半单李代数的根系分类

package/knowledge-base/lie-theory/lie-group.en.md

@@ -45,7 +45,21 @@ GPU friendliness: the key factor is whether exp/log has a closed form.
 
 ## Further References
 
-- Distillation notes: references/books/micro-lie-theory.md (Section II-A Lie Group Definition, Section II-D Exponential Map, Section II-E Plus/Minus Operators)
-- Distillation notes: references/books/differential-geometry.md (Ch 5 Lie Groups)
-- Distillation notes: references/books/smooth-manifolds.md (Ch 7 Lie Groups)
+- Distillation notes: ../../references/books/micro-lie-theory.en.md (Section II-A Lie Group Definition, Section II-D Exponential Map, Section II-E Plus/Minus Operators)
+- Distillation notes: ../../references/books/differential-geometry.en.md (Ch 5 Lie Groups)
+- Distillation notes: ../../references/books/smooth-manifolds.en.md (Ch 7 Lie Groups)
 - Original text: Joan Sola et al., *A micro Lie theory*, Section II-A through Section II-F (complete Lie group toolkit)
+
+
+## Routing Extensions
+- If infinitesimal structure is needed -> `lie-algebra.en.md` (Lie algebra is the tangent space of a Lie group)
+- If linear representations are needed -> `representation.en.md` (finite-dimensional representations of Lie groups)
+- If optimization on Lie groups is needed -> `../optimization/riemannian-optimization.md` (Riemannian optimization on Lie groups)
+
+## Extensible Directions
+- Simply connected Lie group: universal covering group
+- Covering group: covering maps between Lie groups
+- Maximal torus: maximal abelian subgroup in compact Lie groups
+- Compact / semisimple / solvable: structural classification of Lie groups
+- Nilpotent: nilpotent Lie groups and nilpotent Lie algebras
+- Exponential map properties: properties of the map from Lie algebra to Lie group

package/knowledge-base/lie-theory/lie-group.md

@@ -45,7 +45,21 @@ GPU 友好度：关键看 exp/log 是否有闭式。
 
 ## 深入参考
 
-- 蒸馏稿：references/books/micro-lie-theory.md（§II-A 李群定义, §II-D 指数映射, §II-E 加减算子）
-- 蒸馏稿：references/books/differential-geometry.md（Ch 5 Lie Groups）
-- 蒸馏稿：references/books/smooth-manifolds.md（Ch 7 Lie Groups）
+- 蒸馏稿：../../references/books/micro-lie-theory.md（§II-A 李群定义, §II-D 指数映射, §II-E 加减算子）
+- 蒸馏稿：../../references/books/differential-geometry.md（Ch 5 Lie Groups）
+- 蒸馏稿：../../references/books/smooth-manifolds.md（Ch 7 Lie Groups）
 - 原书：Joan Sola et al., *A micro Lie theory*, §II-A ~ §II-F（完整李群工具链）
+
+
+## 路由扩展
+- 若需要无穷小结构 → `lie-algebra.md`（李代数是李群的切空间）
+- 若需要线性表示 → `representation.md`（李群的有限维表示）
+- 若需要在李群上优化 → `../optimization/riemannian-optimization.md`（李群上的黎曼优化）
+
+## 可扩展方向
+- 单连通李群（simply connected Lie group）：万有覆叠群
+- 覆叠群（covering group）：李群之间的覆叠映射
+- 极大环面（maximal torus）：紧李群中的极大交换子群
+- 紧/半单/可解（compact / semisimple / solvable）：李群的结构分类
+- 幂零（nilpotent）：幂零李群与幂零李代数
+- 指数映射性质（exponential map properties）：李代数到李群的映射特性

package/knowledge-base/lie-theory/representation.en.md

@@ -46,7 +46,22 @@ GPU friendliness depends on the group size and representation dimension:
 
 ## Further References
 
-- Distillation notes: references/books/micro-lie-theory.md (Section II-F Adjoint $\text{Ad}_X$ and Adjoint Matrix)
-- Distillation notes: references/books/differential-geometry.md (Ch 5 Lie Groups, adjoint representation section)
+- Distillation notes: ../../references/books/micro-lie-theory.en.md (Section II-F Adjoint $\text{Ad}_X$ and Adjoint Matrix)
+- Distillation notes: ../../references/books/differential-geometry.en.md (Ch 5 Lie Groups, adjoint representation section)
 - Original text: Joan Sola et al., *A micro Lie theory*, Section II-F (adjoint representation, equations 30--35)
 - Original text: Jeffrey M. Lee, *Manifolds and Differential Geometry*, Ch 5 (representations of Lie groups)
+
+
+## Routing Extensions
+- If equivariant network design is needed -> `equivariance.en.md` (representation theory drives equivariant network construction)
+- If the specific form of group action is needed -> `group-action.en.md` (representations are linear group actions)
+- If irreducible decomposition is needed -> `../matrix-analysis/spectral-decomposition.md` (analogous to spectral decomposition of matrices)
+
+## Extensible Directions
+- Irreducible representation: basic building blocks of representations
+- Character: trace function and classification of representations
+- Schur's lemma: morphisms between irreducible representations
+- Peter-Weyl theorem: decomposition of regular representation for compact groups
+- Induced representation: constructing representations of large groups from subgroups
+- Tensor product of representations: combining multi-particle systems
+- Clebsch-Gordan coefficients: transformation coefficients for tensor product decomposition into irreducibles

package/knowledge-base/lie-theory/representation.md

@@ -46,7 +46,22 @@ GPU 友好度取决于群的规模和表示维数：
 
 ## 深入参考
 
-- 蒸馏稿：references/books/micro-lie-theory.md（§II-F 伴随 Ad_X 与伴随矩阵）
-- 蒸馏稿：references/books/differential-geometry.md（Ch 5 Lie Groups, 伴随表示部分）
+- 蒸馏稿：../../references/books/micro-lie-theory.md（§II-F 伴随 Ad_X 与伴随矩阵）
+- 蒸馏稿：../../references/books/differential-geometry.md（Ch 5 Lie Groups, 伴随表示部分）
 - 原书：Joan Sola et al., *A micro Lie theory*, §II-F（伴随表示，式 30-35）
 - 原书：Jeffrey M. Lee, *Manifolds and Differential Geometry*, Ch 5（李群的表示）
+
+
+## 路由扩展
+- 若需要等变网络设计 → `equivariance.md`（表示理论驱动等变网络构造）
+- 若需要群作用的具体形式 → `group-action.md`（表示是线性群作用）
+- 若需要不可约分解 → `../matrix-analysis/spectral-decomposition.md`（类比为矩阵的谱分解）
+
+## 可扩展方向
+- 不可约表示（irreducible representation）：表示的基本构建块
+- 特征标（character）：表示的迹函数与分类
+- Schur 引理：不可约表示间的态射
+- Peter-Weyl 定理：紧群的正则表示分解
+- 诱导表示（induced representation）：从子群构造大群表示
+- 表示的张量积（tensor product of representations）：多粒子系统的组合
+- Clebsch-Gordan 系数：张量积分解为不可约的变换系数

package/knowledge-base/matrix-analysis/index.en.md

@@ -0,0 +1,49 @@
+# Matrix Analysis Activation Index
+
+## Domain Signals
+Activate this domain direction when the problem involves:
+- Dimensionality reduction / compression: need to compress high-dimensional data or parameters into a low-dimensional subspace
+- Low-rank approximation: need to approximate large matrices with low-rank ones
+- Condition number / ill-conditioning: numerical stability issues in matrix computations
+- Orthogonal decomposition: need to decompose space into orthogonal subspaces
+- Spectral structure: need to analyze eigenvalue / singular value distributions
+- Subspace separation: need to measure or control relationships between subspaces
+
+## Core Anchors
+- `projection.en.md` — Projection operators and subspace projection
+- `spectral-decomposition.en.md` — Spectral decomposition and eigendecomposition
+- `low-rank-approximation.en.md` — Low-rank approximation and truncated SVD
+- `positive-semidefinite.en.md` — Positive semidefinite matrices and PSD cone
+- `matrix-perturbation.en.md` — Matrix perturbation theory and error bounds
+
+## Extended Concepts
+When core anchors are insufficient, the following concepts may need temporary activation:
+- SVD variants (truncated SVD, randomized SVD): fast decomposition methods for large-scale matrices
+- PCA / kernel PCA: principal component analysis and its kernelized version
+- condition number: computation and control of condition numbers
+- pseudoinverse: Moore-Penrose generalized inverse and its applications
+- matrix equation (Sylvester / Lyapunov): solution methods and stability for matrix equations
+- Schur decomposition: Schur decomposition and invariant subspaces
+- Jordan form: Jordan canonical form and generalized eigenspaces
+- matrix function: definition and computation of matrix functions
+- Kronecker product / vectorization: tensor product and vectorization operations
+- randomized linear algebra: randomized linear algebra methods
+- CUR decomposition: column/row sampling-based matrix approximation
+- Nystrom approximation: low-rank approximation of kernel matrices
+- matrix concentration inequalities: concentration inequalities for random matrices
+
+## Reference Book Directions
+- `../../references/books/matrix-analysis.en.md`: comprehensive coverage of matrix analysis, especially spectral decomposition, perturbation theory, and matrix function chapters
+
+## AI Translation Directions
+- projection → subspace attention / conflict removal / shared-private split
+- spectral decomposition → low-rank KV cache / token pruning / stability monitor
+- low-rank approximation → LoRA / adapter modules / memory-efficient attention
+- positive-semidefinite → covariance-aware regularization / PSD-constrained loss
+- matrix perturbation → condition number monitoring / robustness-aware training
+
+## Temporary Activation Rules
+When the problem requires mathematics not in the core anchors:
+1. First check whether extended concepts contain a match
+2. If yes, generate a temporary knowledge card based on the lens
+3. If no, enter the Knowledge Gap Protocol

package/knowledge-base/matrix-analysis/index.md

@@ -0,0 +1,49 @@
+# 矩阵分析 激活索引 / Matrix Analysis Activation Index
+
+## 领域信号
+当问题涉及以下信号时，激活本领域方向：
+- 降维/压缩：需要将高维数据或参数压缩到低维子空间
+- 低秩近似：需要用低秩矩阵近似大规模矩阵
+- 条件数/病态：矩阵运算的数值稳定性出现问题
+- 正交分解：需要将空间分解为正交子空间
+- 谱结构：需要分析矩阵的特征值/奇异值分布
+- 子空间分离：需要度量或控制子空间之间的关系
+
+## 核心锚点
+- `projection.md` — 投影算子与子空间投影
+- `spectral-decomposition.md` — 谱分解与特征分解
+- `low-rank-approximation.md` — 低秩近似与截断 SVD
+- `positive-semidefinite.md` — 半正定矩阵与 PSD 锥
+- `matrix-perturbation.md` — 矩阵扰动理论与误差界
+
+## 扩展概念
+当核心锚点不够时，以下概念可能需要临时激活：
+- SVD 变体（truncated SVD, randomized SVD）：大规模矩阵的快速分解方法
+- PCA / kernel PCA：主成分分析及其核化版本
+- condition number：条件数的计算与控制
+- pseudoinverse：Moore-Penrose 广义逆及其应用
+- matrix equation（Sylvester / Lyapunov）：矩阵方程的解法与稳定性
+- Schur decomposition：Schur 分解与不变子空间
+- Jordan form：Jordan 标准形与广义特征空间
+- matrix function：矩阵函数的定义与计算
+- Kronecker product / vectorization：张量积与向量化运算
+- randomized linear algebra：随机化线性代数方法
+- CUR decomposition：基于列/行采样的矩阵近似
+- Nystrom approximation：核矩阵的低秩近似
+- matrix concentration inequalities：随机矩阵的集中不等式
+
+## 参考书方向
+- `../../references/books/matrix-analysis.md`：矩阵分析全面覆盖，特别是谱分解、扰动理论和矩阵函数章节
+
+## AI 翻译方向
+- projection → subspace attention / conflict removal / shared-private split
+- spectral decomposition → low-rank KV cache / token pruning / stability monitor
+- low-rank approximation → LoRA / adapter modules / memory-efficient attention
+- positive-semidefinite → covariance-aware regularization / PSD-constrained loss
+- matrix perturbation → condition number monitoring / robustness-aware training
+
+## 临时激活规则
+当问题需要的数学不在核心锚点中时：
+1. 先检查扩展概念中是否有匹配
+2. 若有，根据透镜生成临时知识卡
+3. 若无，进入 Knowledge Gap Protocol

package/knowledge-base/matrix-analysis/low-rank-approximation.en.md

@@ -2,7 +2,7 @@
 
 ## Minimal Definition
 
-Given a matrix $A \in \mathbb{R}^{m \times n}$, find a matrix $B$ with rank at most $k$ that minimizes $\|A - B\|$. The Eckart-Young-Mirsky theorem guarantees that the truncated SVD provides the unique optimal solution under both the Frobenius norm and the spectral norm: $B_k = \sum_{i=1}^k \sigma_i u_i v_i^H$.
+Given a matrix $A \in \mathbb{R}^{m \times n}$, find a matrix $B$ with rank at most $k$ that minimizes $\|A - B\|$. The Eckart-Young-Mirsky theorem guarantees that the truncated SVD provides an optimal solution under both the Frobenius norm and the spectral norm; the principal subspace is unique when $\sigma_k > \sigma_{k+1}$: $B_k = \sum_{i=1}^k \sigma_i u_i v_i^H$.
 
 ## Core Formulas
 
@@ -16,7 +16,7 @@ Given a matrix $A \in \mathbb{R}^{m \times n}$, find a matrix $B$ with rank at m
 ## Applicable Problems
 
 - LoRA weight compression: $W \approx W_0 + BA$, $B \in \mathbb{R}^{d \times r}, A \in \mathbb{R}^{r \times d}$, $r \ll d$
-- KV-Cache compression: projecting Key/Value caches into a low-dimensional subspace, reducing memory from $O(n)$ to $O(k)$
+- KV-Cache compression: projecting Key/Value caches into a low-rank factor format, reducing memory from $O(Ld)$ to $O(Lk + kd)$; standard softmax still has sequence length $L$
 - PCA / whitening: the top $k$ principal components of the data covariance matrix correspond to the truncated SVD
 - Gradient compression: the effective rank of gradient matrices in large models is often much lower than the nominal rank, allowing safe truncation
 - Recommender systems / matrix completion: low-rank factorization $R \approx UV^H$
@@ -25,16 +25,16 @@ Given a matrix $A \in \mathbb{R}^{m \times n}$, find a matrix $B$ with rank at m
 
 - **LoRA (Low-Rank Adaptation)**: Freeze $W_0$, train $\Delta W = BA$ ($r \ll d$), and merge $W = W_0 + BA$ at inference time. Forward pass = two matmul operations ($x \to Ax \to BAx$), reducing trainable parameters from $O(d^2)$ to $O(dr)$. Implemented via `torch.mm(B, torch.mm(A, x))` or merged into a single matmul.
 - **Randomized SVD operator**: For a large matrix $A \in \mathbb{R}^{m \times n}$, first sample $Y = A\Omega$ ($\Omega \in \mathbb{R}^{n \times (k+p)}$ random Gaussian), compute QR decomposition $Y = QR$, then form $B = Q^HA$ (small matrix $O(k \times n)$), and perform SVD on $B$. Total complexity $O(mnk)$ instead of $O(mn^2)$; all core operations are matmul.
-- **KV-Cache low-rank reduction**: Maintain $K_k = K P_k$ ($P_k$ being the projection onto the top $k$ principal components), performing incremental PCA or streaming SVD updates for each new token. Attention computation $\text{softmax}(Q K_k^H / \sqrt{d}) V_k$ involves three matmul operations, reducing the sequence dimension from $L$ to $k$.
+- **KV-Cache low-rank reduction**: Maintain $K$ in factored low-rank form $K \approx U_k \Sigma_k V_k^H$ (storing $U_k \in \mathbb{R}^{L \times k}$ and $\Sigma_k V_k^H \in \mathbb{R}^{k \times d}$, totaling $O(Lk + kd)$ instead of $O(Ld)$). Perform incremental PCA or streaming SVD updates for each new token. **Note**: for standard softmax attention, low-rank factors cannot be treated as $k$ compressed tokens; softmax is still normalized over length $L$. However, one can compute $qK^T = (q(\Sigma_k V_k^H)^T)U_k^T$ in factored form, avoiding materialized $L \times d$ reconstruction and reducing the QK inner dimension from $d$ to $k$. Only in linear attention, and only when compressing additive statistics such as $\phi(K)^T V$ and $\phi(K)^T\mathbf{1}$, can the historical state truly shrink from $L$ tokens to $k$ statistical factors.
 - **Nuclear norm regularization**: $\mathcal{L} = \mathcal{L}_{\text{task}} + \lambda \|W\|_*$ promotes low-rank solutions. However, nuclear norm computation requires full SVD ($O(n^3)$). Alternatives: (1) approximate with truncated SVD; (2) factorize $\|W\|_* = \min_{W=UV^H} \frac{1}{2}(\|U\|_F^2 + \|V\|_F^2)$, converting to Frobenius regularization on $U, V$.
-- **Gradient low-rank compression (distributed training)**: Truncate gradient $G$ to $G_k$ (top-$k$ SVD) before all-reduce, reducing communication from $O(d)$ to $O(kd)$. Use randomized SVD locally on each device, then merge.
+- **Gradient low-rank compression (distributed training)**: For a gradient matrix $G \in \mathbb{R}^{m \times n}$, transmit top-$k$ SVD factors before all-reduce, reducing communication from $O(mn)$ to $O(k(m+n))$ (if $m \approx n \approx d$, this is $O(d^2) \to O(kd)$). Use randomized SVD locally on each device, then merge.
 
 ## Engineering Feasibility
 
-- **Primary operations**: matmul + small-matrix SVD. LoRA forward = two matmul operations; randomized SVD = three matmul operations + one small QR; truncated SVD = $O(k/n)$ fraction of full SVD (using Lanczos).
+- **Primary operations**: matmul + small-matrix SVD. LoRA forward = two matmul operations; randomized SVD = several matmuls + one thin QR + one small SVD, typically $O(mnk)$ (also depending on oversampling, power iterations, and spectral gaps). Lanczos / randomized truncated SVD is usually far cheaper than full SVD, but should not be summarized as an $O(k/n)$ fraction of full SVD.
 - **GPU friendliness**: Extremely high. LoRA forward/backward are all tensor core matmul operations; the dominant cost of randomized SVD is also matmul. Small-matrix SVD has cuSOLVER batched implementations.
 - **Complexity**: LoRA forward $O(dk)$ per sample vs. $O(d^2)$ full rank; randomized SVD $O(mnk)$; full SVD $O(\min(m^2n, mn^2))$.
-- **Memory**: LoRA storage $O(dr)$ vs. $O(d^2)$; KV-Cache low-rank reduction $O(Lk)$ vs. $O(Ld)$.
+- **Memory**: LoRA storage $O(dr)$ vs. $O(d^2)$; KV-Cache low-rank factor storage $O(Lk + kd)$ vs. $O(Ld)$.
 
 ## Risks and Failure Conditions
 
@@ -45,5 +45,18 @@ Given a matrix $A \in \mathbb{R}^{m \times n}$, find a matrix $B$ with rank at m
 
 ## Further References
 
-- Distilled notes: references/books/matrix-analysis.md (Section 2.6 SVD, Section 7.4 Polar Decomposition and SVD, nuclear norm-spectral norm duality Section 5.5)
+- Distilled notes: ../../references/books/matrix-analysis.en.md (Section 2.6 SVD, Section 7.4 Polar Decomposition and SVD, nuclear norm-spectral norm duality Section 5.5)
 - Original text: Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 2 Section 2.6 (SVD) + Chapter 7 Section 7.3-7.4 (Polar Decomposition & SVD)
+
+
+## Routing Extensions
+- If subspace projection implementation is needed -> `projection.en.md` (projection operators)
+- If detailed decomposition tools are needed -> `spectral-decomposition.en.md` (SVD/EVD)
+- If the goal is information-preserving compression -> `../probability/information-bottleneck.en.md` (information bottleneck theory)
+
+## Extensible Directions
+- Tensor decomposition (CP / Tucker / TT): low-rank decomposition of higher-order tensors
+- Structured low-rank (Toeplitz / Hankel): structure-preserving low-rank approximation
+- Online / streaming low-rank: incrementally updated low-rank estimation
+- Matrix completion: recovering low-rank matrices from partial observations
+- Robust PCA: low-rank + sparse decomposition

package/knowledge-base/matrix-analysis/low-rank-approximation.md

@@ -2,7 +2,7 @@
 
 ## 最小定义
 
-给定矩阵 $A \in \mathbb{R}^{m \times n}$，寻找秩不超过 $k$ 的矩阵 $B$ 使得 $\|A - B\|$ 最小。Eckart-Young-Mirsky 定理保证截断 SVD 给出 Frobenius 范数和谱范数下的唯一最优解：$B_k = \sum_{i=1}^k \sigma_i u_i v_i^H$。
+给定矩阵 $A \in \mathbb{R}^{m \times n}$，寻找秩不超过 $k$ 的矩阵 $B$ 使得 $\|A - B\|$ 最小。Eckart-Young-Mirsky 定理保证截断 SVD 给出 Frobenius 范数和谱范数下的一个最优解；当 $\sigma_k > \sigma_{k+1}$ 时，对应主子空间唯一：$B_k = \sum_{i=1}^k \sigma_i u_i v_i^H$。
 
 ## 核心公式
 
@@ -16,7 +16,7 @@
 ## 适用问题
 
 - LoRA 权重压缩：$W \approx W_0 + BA$，$B \in \mathbb{R}^{d \times r}, A \in \mathbb{R}^{r \times d}$，$r \ll d$
-- KV-Cache 压缩：将 Key/Value 缓存投影到低维子空间，显存 $O(n) \to O(k)$
+- KV-Cache 压缩：将 Key/Value 缓存投影到低秩因子格式，显存从 $O(Ld)$ 变为 $O(Lk + kd)$；标准 softmax 下序列长度仍为 $L$
 - PCA / 白化：数据协方差的前 $k$ 个主成分即截断 SVD
 - 梯度压缩：大模型梯度矩阵的有效秩通常远低于名义秩，可安全截断
 - 推荐系统 / 矩阵补全：低秩因子分解 $R \approx UV^H$
@@ -25,16 +25,16 @@
 
 - **LoRA (Low-Rank Adaptation)**：冻结 $W_0$，训练 $\Delta W = BA$（$r \ll d$），推理时合并 $W = W_0 + BA$。前向传播 = 两次 matmul（$x \to Ax \to BAx$），训练参数量从 $O(d^2)$ 降到 $O(dr)$。用 `torch.mm(B, torch.mm(A, x))` 或合并为单次 matmul。
 - **随机化 SVD 算子**：对大矩阵 $A \in \mathbb{R}^{m \times n}$，先采样 $Y = A\Omega$（$\Omega \in \mathbb{R}^{n \times (k+p)}$ 随机高斯），QR 分解 $Y = QR$，再算 $B = Q^HA$（小矩阵 $O(k \times n)$），对 $B$ 做 SVD。总复杂度 $O(mnk)$ 而非 $O(mn^2)$，核心操作全是 matmul。
-- **KV-Cache 低秩化**：维护 $K_k = K P_k$（$P_k$ 为投影到前 $k$ 主成分），每新到 token 做增量 PCA 或 streaming SVD 更新。Attention 计算 $\text{softmax}(Q K_k^H / \sqrt{d}) V_k$，三次 matmul，序列维度从 $L$ 降到 $k$。
+- **KV-Cache 低秩化**：维护 $K$ 的低秩因子形式 $K \approx U_k \Sigma_k V_k^H$（存储 $U_k \in \mathbb{R}^{L \times k}$ 和 $\Sigma_k V_k^H \in \mathbb{R}^{k \times d}$，共 $O(Lk + kd)$ 而非 $O(Ld)$）。每新到 token 做增量 PCA 或 streaming SVD 更新。**注意**：对标准 softmax attention，低秩因子不能当作 $k$ 个压缩 token；softmax 仍在长度 $L$ 上归一化。但可用因子化计算 $qK^T = (q(\Sigma_k V_k^H)^T)U_k^T$，避免物化 $L \times d$ 重构，并将 QK 内维从 $d$ 降到 $k$。仅在线性注意力中，且压缩 $\phi(K)^T V$、$\phi(K)^T\mathbf{1}$ 等可加统计量时，历史状态才可真正从 $L$ 降至 $k$ 个统计因子。
 - **核范数正则化**：$\mathcal{L} = \mathcal{L}_{\text{task}} + \lambda \|W\|_*$ 促进低秩解。但核范数计算需完整 SVD（$O(n^3)$），替代方案：(1) 用截断 SVD 近似；(2) 因子化 $\|W\|_* = \min_{W=UV^H} \frac{1}{2}(\|U\|_F^2 + \|V\|_F^2)$ 转为对 $U, V$ 的 Frobenius 正则。
-- **梯度低秩压缩 (分布式训练)**：将梯度 $G$ 截断为 $G_k$（top-$k$ SVD）后 all-reduce 通信量从 $O(d)$ 降到 $O(kd)$。用随机化 SVD 在每卡本地算，再合并。
+- **梯度低秩压缩 (分布式训练)**：对梯度矩阵 $G \in \mathbb{R}^{m \times n}$ 做 top-$k$ SVD 后传输因子，all-reduce 通信量从 $O(mn)$ 降到 $O(k(m+n))$（若 $m \approx n \approx d$，即 $O(d^2) \to O(kd)$）。用随机化 SVD 在每卡本地算，再合并。
 
 ## 工程可行性
 
-- **主要操作**：matmul + 小矩阵 SVD。LoRA 前向 = 两次 matmul；随机化 SVD = 三次 matmul + 一次小 QR；截断 SVD = 完整 SVD 的 $O(k/n)$ 倍（用 Lanczos）。
+- **主要操作**：matmul + 小矩阵 SVD。LoRA 前向 = 两次 matmul；随机化 SVD = 若干次 matmul + 一次 thin QR + 小矩阵 SVD，典型复杂度 $O(mnk)$（还依赖 oversampling、power iteration 和谱间隙）；Lanczos/随机化截断 SVD 通常远低于完整 SVD，但不能简单写成完整 SVD 的 $O(k/n)$ 倍。
 - **GPU 友好度**：极高。LoRA 前向/反向全是 tensor core matmul；随机化 SVD 的主要开销也是 matmul。小矩阵 SVD 有 cuSOLVER 的 batched 实现。
 - **复杂度**：LoRA 前向 $O(dk)$ per sample vs. $O(d^2)$ 全秩；随机化 SVD $O(mnk)$；完整 SVD $O(\min(m^2n, mn^2))$。
-- **显存**：LoRA 存储 $O(dr)$ vs. $O(d^2)$；KV-Cache 低秩化 $O(Lk)$ vs. $O(Ld)$。
+- **显存**：LoRA 存储 $O(dr)$ vs. $O(d^2)$；KV-Cache 低秩因子存储 $O(Lk + kd)$ vs. $O(Ld)$。
 
 ## 风险与失效条件
 
@@ -45,5 +45,18 @@
 
 ## 深入参考
 
-- 蒸馏稿：references/books/matrix-analysis.md（§2.6 SVD、§7.4 极分解与 SVD、核范数-谱范数对偶 §5.5）
+- 蒸馏稿：../../references/books/matrix-analysis.md（§2.6 SVD、§7.4 极分解与 SVD、核范数-谱范数对偶 §5.5）
 - 原书：Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 2 §2.6 (SVD) + Chapter 7 §7.3-7.4 (Polar Decomposition & SVD)
+
+
+## 路由扩展
+- 若需要子空间投影的具体实现 → `projection.md`（投影算子）
+- 若需要分解工具的详细分析 → `spectral-decomposition.md`（SVD/EVD）
+- 若目标是信息保持压缩 → `../probability/information-bottleneck.md`（信息瓶颈理论）
+
+## 可扩展方向
+- 张量分解（CP / Tucker / TT）：高阶张量的低秩分解
+- 结构化低秩（Toeplitz / Hankel）：保持结构的低秩近似
+- 在线/流式低秩（online / streaming low-rank）：增量更新的低秩估计
+- 矩阵补全（matrix completion）：从部分观测恢复低秩矩阵
+- 鲁棒 PCA（robust PCA）：低秩+稀疏分解

package/knowledge-base/matrix-analysis/matrix-perturbation.en.md

@@ -45,5 +45,17 @@ Studies how the eigenvalues, singular values, and eigenspaces of a matrix $A$ ch
 
 ## Further References
 
-- Distilled notes: references/books/matrix-analysis.md (Section 4.3 Eigenvalue Inequalities Weyl/Interlacing, Section 6.1-6.3 Geršgorin Discs and Perturbation Theorems, Section 5.8 Condition Numbers)
+- Distilled notes: ../../references/books/matrix-analysis.en.md (Section 4.3 Eigenvalue Inequalities Weyl/Interlacing, Section 6.1-6.3 Geršgorin Discs and Perturbation Theorems, Section 5.8 Condition Numbers)
 - Original text: Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 4 Section 4.3 (Eigenvalue Inequalities) + Chapter 6 (Location and Perturbation of Eigenvalues Section 6.1-6.3)
+
+
+## Routing Extensions
+- If focusing on eigenvalue sensitivity -> `spectral-decomposition.en.md` (perturbation analysis of spectral decomposition)
+- If random perturbation bounds are involved -> `../probability/concentration-inequality.md` (random matrix concentration inequalities)
+
+## Extensible Directions
+- Pseudospectra: spectral sensitivity analysis for non-normal matrices
+- Structured perturbation: perturbation analysis preserving matrix structure
+- Davis-Kahan theorem variants: multiple bounds for subspace perturbation
+- Eigenvalue interlacing: relationship between submatrix and eigenvalues
+- Wilkinson polynomial: condition numbers and sensitivity of polynomial roots

package/knowledge-base/matrix-analysis/matrix-perturbation.md

@@ -45,5 +45,17 @@
 
 ## 深入参考
 
-- 蒸馏稿：references/books/matrix-analysis.md（§4.3 特征值不等式 Weyl/交错、§6.1-6.3 Ger\v{s}gorin 圆盘与扰动定理、§5.8 条件数）
+- 蒸馏稿：../../references/books/matrix-analysis.md（§4.3 特征值不等式 Weyl/交错、§6.1-6.3 Ger\v{s}gorin 圆盘与扰动定理、§5.8 条件数）
 - 原书：Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 4 §4.3 (Eigenvalue Inequalities) + Chapter 6 (Location and Perturbation of Eigenvalues §6.1-6.3)
+
+
+## 路由扩展
+- 若关注特征值灵敏度 → `spectral-decomposition.md`（谱分解的扰动分析）
+- 若涉及随机扰动界 → `../probability/concentration-inequality.md`（随机矩阵集中不等式）
+
+## 可扩展方向
+- 伪谱（pseudospectra）：非正规矩阵的谱灵敏度分析
+- 结构化扰动（structured perturbation）：保持矩阵结构的扰动分析
+- Davis-Kahan 定理变体：子空间扰动的多种界
+- 特征值交错（eigenvalue interlacing）：主子矩阵与特征值关系
+- Wilkinson 多项式：多项式根的条件数与扰动灵敏度

package/knowledge-base/matrix-analysis/positive-semidefinite.en.md

@@ -2,12 +2,12 @@
 
 ## Minimal Definition
 
-A Hermitian matrix $A$ is positive semidefinite (PSD) if $x^HAx \geq 0$ for all nonzero vectors $x$, denoted $A \succeq 0$. Equivalent conditions: all eigenvalues $\geq 0$; there exists $B$ such that $A = B^HB$ (Gram representation); all leading principal minors $\geq 0$. Positive definite (PD) requires strict $> 0$, denoted $A \succ 0$.
+A Hermitian matrix $A$ is positive semidefinite (PSD) if $x^HAx \geq 0$ for all nonzero vectors $x$, denoted $A \succeq 0$. Equivalent conditions: all eigenvalues $\geq 0$; there exists $B$ such that $A = B^HB$ (Gram representation); **all principal minors** $\geq 0$ (note: ALL principal minors, not just the leading ones). Positive definite (PD) requires strict $> 0$, denoted $A \succ 0$.
 
 ## Core Formulas
 
-- PSD equivalence: $A \succeq 0 \iff \lambda_i(A) \geq 0 \ \forall i \iff A = B^HB$
-- PD equivalence: $A \succ 0 \iff \lambda_i(A) > 0 \ \forall i \iff$ all leading principal minors $> 0$ (Sylvester's criterion)
+- PSD equivalence: $A \succeq 0 \iff \lambda_i(A) \geq 0 \ \forall i \iff A = B^HB \iff$ all principal minors $\geq 0$ (note: **all** principal minors, not just the leading ones)
+- PD equivalence (Sylvester's criterion): $A \succ 0 \iff \lambda_i(A) > 0 \ \forall i \iff$ all leading principal minors $> 0$ (checking only leading principal minors suffices; this is necessary and sufficient for PD)
 - Cholesky decomposition: $A \succ 0 \implies A = LL^H$, $L$ lower triangular
 - Loewner partial order: $A \succeq B \iff A - B \succeq 0$
 - Schur product theorem: $A \succeq 0, B \succeq 0 \implies A \circ B \succeq 0$ (Hadamard product preserves PSD)
@@ -20,12 +20,12 @@ A Hermitian matrix $A$ is positive semidefinite (PSD) if $x^HAx \geq 0$ for all
 - Covariance matrices: $\Sigma = \mathbb{E}[xx^H] \succeq 0$; PCA/whitening relies on positive definiteness
 - Second-order optimization preconditioning: The PSD structure of the Hessian/Fisher information matrix guarantees descent directions
 - Semidefinite programming (SDP): optimizing a linear objective subject to $X \succeq 0$ constraints
-- Attention matrix analysis: Gram structure of the row-stochastic matrix output by softmax
+- Attention matrix analysis: a row-stochastic softmax attention matrix is generally not a Gram/PSD matrix; PSD tools apply only after explicitly constructing a symmetric PSD kernel (e.g. $K_{ij}=k(x_i,x_j)$) or symmetrizing and projecting to the PSD cone.
 
 ## AI Design Translation
 
 - **PSD kernel engineering (learnable kernels)**: Use the Schur product theorem to combine multiple PSD kernels: $K = K_1 \circ K_2 \circ \cdots$ (Hadamard product), guaranteeing the result remains PSD. Implemented as elementwise tensor multiplication `K = K1 * K2`, $O(n^2)$ elementwise, extremely GPU-friendly. Can parameterize $K_\theta(x,y) = \exp(-\|f_\theta(x)-f_\theta(y)\|^2)$ to guarantee PSD.
-- **Covariance whitening**: $\hat{x} = \Sigma^{-1/2}x$, where $\Sigma^{-1/2}$ is approximated via Newton-Schulz iteration (pure matmul). Newton-Schulz: $X_{k+1} = \frac{1}{2}X_k(3I - AX_k)$, two matmul operations per step, converging in 5-6 steps. BatchNorm can be viewed as an approximation to diagonal whitening.
+- **Covariance whitening**: $\hat{x} = \Sigma^{-1/2}x$, where $\Sigma^{-1/2}$ can be approximated via scaled Newton-Schulz iteration (pure matmul). A common coupled form uses $Y_0=A/\alpha, Z_0=I$, $T_k=\frac{1}{2}(3I-Z_kY_k)$, $Y_{k+1}=Y_kT_k, Z_{k+1}=T_kZ_k$, with $A^{-1/2}\approx Z_k/\sqrt{\alpha}$; an equivalent single-variable form is $X_{k+1}=\frac{1}{2}X_k(3I-AX_k^2)$. Convergence requires spectral scaling and positive definiteness; in low precision, residual checks are needed. 5-6 steps is an engineering budget, not a universal guarantee. BatchNorm can be viewed as an approximation to diagonal whitening.
 - **Cholesky preconditioner**: For a PSD Hessian $H$, use the $H = LL^H$ decomposition and solve $L^{-1}L^{-H}g$ instead of $H^{-1}g$. cuSOLVER provides batched Cholesky `potrf`. In K-FAC, the inversion of each Kronecker factor proceeds via Cholesky.
 - **Nearest PSD approximation (Higham)**: Given a symmetric matrix $A$, find the nearest PSD matrix $A_+ = \arg\min_{X \succeq 0} \|A - X\|_F$. Solution: EVD $A = U\Lambda U^H$, clamp negative values in $\Lambda$ to zero, $A_+ = U\Lambda_+ U^H$. Used to correct loss of positive definiteness in covariance matrices due to floating-point errors.
 - **Jitter / diagonal loading**: $A_{\text{stable}} = A + \epsilon I$ ($\epsilon \sim 10^{-6}$), ensuring numerical positive definiteness. Standard practice in Gaussian processes, kernel methods, and Cholesky decomposition. Implemented as `A + eps * torch.eye(n)`, a zero-cost operation.
@@ -46,5 +46,19 @@ A Hermitian matrix $A$ is positive semidefinite (PSD) if $x^HAx \geq 0$ for all
 
 ## Further References
 
-- Distilled notes: references/books/matrix-analysis.md (Ch 7 Positive Definite and Semidefinite Matrices, Section 7.5 Schur Product Theorem, Section 7.7 Loewner Partial Order)
+- Distilled notes: ../../references/books/matrix-analysis.en.md (Ch 7 Positive Definite and Semidefinite Matrices, Section 7.5 Schur Product Theorem, Section 7.7 Loewner Partial Order)
 - Original text: Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 7 (Positive Definite and Semidefinite Matrices Section 7.1-7.8)
+
+
+## Routing Extensions
+- If solving SDP problems -> `../optimization/convex-optimization.md` (semidefinite programming as convex optimization)
+- If PSD matrix conditioning and perturbation are involved -> `matrix-perturbation.en.md` (eigenvalue perturbation bounds)
+- If used for Fisher information matrix -> `../probability/fisher-information.md` (PSD property of Fisher information)
+
+## Extensible Directions
+- Semidefinite programming (SDP): solution methods and applications
+- PSD completion: completing partially known PSD matrices
+- Matrix square root: unique square root of PSD matrices
+- Lowner order: partial order on the PSD cone
+- Operator monotone functions: Loewner-Heinz theorem
+- Completely positive matrices: CP decomposition and cone structure

package/knowledge-base/matrix-analysis/positive-semidefinite.md

@@ -2,12 +2,12 @@
 
 ## 最小定义
 
-Hermitian 矩阵 $A$ 若对所有非零向量 $x$ 满足 $x^HAx \geq 0$，则称半正定（PSD），记 $A \succeq 0$。等价条件：所有特征值 $\geq 0$；存在 $B$ 使 $A = B^HB$（Gram 表示）；所有顺序主子式 $\geq 0$。正定（PD）要求严格 $> 0$，记 $A \succ 0$。
+Hermitian 矩阵 $A$ 若对所有非零向量 $x$ 满足 $x^HAx \geq 0$，则称半正定（PSD），记 $A \succeq 0$。等价条件：所有特征值 $\geq 0$；存在 $B$ 使 $A = B^HB$（Gram 表示）；**所有主子式** $\geq 0$（注意：是全部主子式，不仅是顺序主子式）。正定（PD）要求严格 $> 0$，记 $A \succ 0$。
 
 ## 核心公式
 
-- PSD 等价条件：$A \succeq 0 \iff \lambda_i(A) \geq 0 \ \forall i \iff A = B^HB$
-- 正定等价：$A \succ 0 \iff \lambda_i(A) > 0 \ \forall i \iff$ 所有顺序主子式 $> 0$（Sylvester 准则）
+- PSD 等价条件：$A \succeq 0 \iff \lambda_i(A) \geq 0 \ \forall i \iff A = B^HB \iff$ 所有主子式 $\geq 0$（注意：是 **所有** 主子式，不仅是顺序主子式）
+- 正定等价（Sylvester 准则）：$A \succ 0 \iff \lambda_i(A) > 0 \ \forall i \iff$ 所有顺序主子式 $> 0$（仅需检验顺序主子式，这是 PD 的充分必要条件）
 - Cholesky 分解：$A \succ 0 \implies A = LL^H$，$L$ 下三角
 - Loewner 偏序：$A \succeq B \iff A - B \succeq 0$
 - Schur 积定理：$A \succeq 0, B \succeq 0 \implies A \circ B \succeq 0$（Hadamard 积保 PSD）
@@ -20,12 +20,12 @@ Hermitian 矩阵 $A$ 若对所有非零向量 $x$ 满足 $x^HAx \geq 0$，则称
 - 协方差矩阵：$\Sigma = \mathbb{E}[xx^H] \succeq 0$，PCA/白化依赖其正定性
 - 二阶优化预条件：Hessian/Fisher 信息矩阵的 PSD 结构保证下降方向
 - 半定规划 (SDP)：约束 $X \succeq 0$ 下的线性目标优化
-- Attention 矩阵分析：softmax 输出的行随机矩阵的 Gram 结构
+- Attention 矩阵分析：softmax 输出的行随机矩阵一般不是 Gram/PSD；只有显式构造对称 PSD kernel（如 $K_{ij}=k(x_i,x_j)$）或对称化并投影到 PSD cone 后，才能使用 PSD 工具。
 
 ## AI 设计翻译
 
 - **PSD 核工程 (可学习核)**：用 Schur 积定理组合多个 PSD 核：$K = K_1 \circ K_2 \circ \cdots$（Hadamard 积），保证结果始终 PSD。实现为 elementwise 张量乘 `K = K1 * K2`，$O(n^2)$ elementwise，极 GPU 友好。可参数化 $K_\theta(x,y) = \exp(-\|f_\theta(x)-f_\theta(y)\|^2)$ 保证 PSD。
-- **协方差白化 (Whitening)**：$\hat{x} = \Sigma^{-1/2}x$，其中 $\Sigma^{-1/2}$ 通过 Newton-Schulz 迭代近似（纯 matmul）。Newton-Schulz：$X_{k+1} = \frac{1}{2}X_k(3I - AX_k)$，每步两次 matmul，5-6 步收敛。BatchNorm 可视为对角白化的近似。
+- **协方差白化 (Whitening)**：$\hat{x} = \Sigma^{-1/2}x$，其中 $\Sigma^{-1/2}$ 可通过缩放后的 Newton-Schulz 迭代近似（纯 matmul）。常用 coupled 形式：$Y_0=A/\alpha, Z_0=I$，$T_k=\frac{1}{2}(3I-Z_kY_k)$，$Y_{k+1}=Y_kT_k, Z_{k+1}=T_kZ_k$，最后 $A^{-1/2}\approx Z_k/\sqrt{\alpha}$；或单变量形式 $X_{k+1}=\frac{1}{2}X_k(3I-AX_k^2)$。收敛需要谱缩放和正定条件，低精度下需残差监控；5-6 步只是常见工程预算。BatchNorm 可视为对角白化的近似。
 - **Cholesky 预条件子**：对 PSD Hessian $H$，用 $H = LL^H$ 分解后解 $L^{-1}L^{-H}g$ 代替 $H^{-1}g$。cuSOLVER 有 batched Cholesky `potrf`。K-FAC 中每个 Kronecker 因子的求逆即走 Cholesky。
 - **最近 PSD 近似 (Higham)**：给定对称矩阵 $A$，求最近 PSD 矩阵 $A_+ = \arg\min_{X \succeq 0} \|A - X\|_F$。解法：EVD $A = U\Lambda U^H$，将 $\Lambda$ 中负值截零，$A_+ = U\Lambda_+ U^H$。用于修正浮点误差导致的协方差矩阵失去正定性。
 - **Jitter / 对角加载**：$A_{\text{stable}} = A + \epsilon I$（$\epsilon \sim 10^{-6}$），保证数值正定性。高斯过程、核方法、Cholesky 分解中标准做法。实现为 `A + eps * torch.eye(n)`，零成本操作。
@@ -46,5 +46,19 @@ Hermitian 矩阵 $A$ 若对所有非零向量 $x$ 满足 $x^HAx \geq 0$，则称
 
 ## 深入参考
 
-- 蒸馏稿：references/books/matrix-analysis.md（Ch 7 Positive Definite and Semidefinite Matrices、§7.5 Schur 积定理、§7.7 Loewner 偏序）
+- 蒸馏稿：../../references/books/matrix-analysis.md（Ch 7 Positive Definite and Semidefinite Matrices、§7.5 Schur 积定理、§7.7 Loewner 偏序）
 - 原书：Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 7 (Positive Definite and Semidefinite Matrices §7.1-7.8)
+
+
+## 路由扩展
+- 若需要求解 SDP 问题 → `../optimization/convex-optimization.md`（半正定规划作为凸优化）
+- 若涉及 PSD 矩阵的条件与扰动 → `matrix-perturbation.md`（特征值扰动界）
+- 若用于 Fisher 信息矩阵 → `../probability/fisher-information.md`（Fisher 信息的 PSD 性质）
+
+## 可扩展方向
+- 半正定规划（semidefinite programming）：SDP 的求解方法与应用
+- PSD 补全（PSD completion）：部分已知 PSD 矩阵的补全
+- 矩阵平方根（matrix square root）：PSD 矩阵的唯一平方根
+- Lowner 序（Lowner order）：PSD 锥上的偏序关系
+- 算子单调函数（operator monotone functions）：Loewner-Heinz 定理
+- 完全正矩阵（completely positive matrices）：CP 分解与锥结构

package/knowledge-base/matrix-analysis/projection.en.md

@@ -43,5 +43,18 @@ Maps a vector $v$ onto a subspace $\mathcal{S}$ such that the residual $v - Pv$
 
 ## Further References
 
-- Distilled notes: references/books/matrix-analysis.md (Section 2.1 QR Decomposition, Section 4.2 Courant-Fischer Variational Characterization, Section 2.6 SVD)
+- Distilled notes: ../../references/books/matrix-analysis.en.md (Section 2.1 QR Decomposition, Section 4.2 Courant-Fischer Variational Characterization, Section 2.6 SVD)
 - Original text: Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 2 (Unitary Similarity) + Chapter 4 (Hermitian Matrices Section 4.2 Variational Characterizations)
+
+
+## Routing Extensions
+- If the goal is compression / dimensionality reduction -> `low-rank-approximation.en.md` (truncated SVD implementation)
+- If projection constraints are needed on a manifold -> `../optimization/riemannian-optimization.md` (constrained optimization on Riemannian manifolds)
+- If shared vs. private subspace separation is involved -> `shared-private-decomposition` (design pattern layer)
+
+## Extensible Directions
+- Oblique projection: non-orthogonal projection operators
+- Alternating projection: Von Neumann alternating projection convergence theorem
+- Projection onto convex sets (POCS): projecting onto intersections of convex sets
+- Randomized projection: Johnson-Lindenstrauss lemma and random projections
+- Subspace tracking: online subspace estimation methods

package/knowledge-base/matrix-analysis/projection.md

@@ -43,5 +43,18 @@
 
 ## 深入参考
 
-- 蒸馏稿：references/books/matrix-analysis.md（§2.1 QR 分解、§4.2 Courant-Fischer 变分刻画、§2.6 SVD）
+- 蒸馏稿：../../references/books/matrix-analysis.md（§2.1 QR 分解、§4.2 Courant-Fischer 变分刻画、§2.6 SVD）
 - 原书：Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 2 (Unitary Similarity) + Chapter 4 (Hermitian Matrices §4.2 Variational Characterizations)
+
+
+## 路由扩展
+- 若目标是压缩/降维 → `low-rank-approximation.md`（截断 SVD 实现）
+- 若需要在流形上做投影约束 → `../optimization/riemannian-optimization.md`（黎曼流形上的约束优化）
+- 若涉及共享与私有子空间分离 → `shared-private-decomposition`（设计模式层）
+
+## 可扩展方向
+- 斜投影（oblique projection）：非正交投影算子
+- 交替投影（alternating projection）：Von Neumann 交替投影收敛定理
+- 凸集投影（POCS）：投影到多个凸集交集的方法
+- 随机投影（randomized projection）：Johnson-Lindenstrauss 引理与随机投影
+- 子空间跟踪（subspace tracking）：在线更新的子空间估计方法

package/knowledge-base/matrix-analysis/spectral-decomposition.en.md

@@ -44,5 +44,18 @@ Decomposes a matrix into a linear combination of eigenvalue-eigenvector pairs. F
 
 ## Further References
 
-- Distilled notes: references/books/matrix-analysis.md (Ch 1 Eigenvalues and Similarity, Section 2.4-2.5 Schur Triangularization and Normal Matrices, Section 4.2 Courant-Fischer)
+- Distilled notes: ../../references/books/matrix-analysis.en.md (Ch 1 Eigenvalues and Similarity, Section 2.4-2.5 Schur Triangularization and Normal Matrices, Section 4.2 Courant-Fischer)
 - Original text: Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 1 (Eigenvalues, Eigenvectors, Similarity) + Chapter 2 (Unitary Similarity Section 2.4-2.5)
+
+
+## Routing Extensions
+- If truncation approximation is needed -> `low-rank-approximation.en.md` (SVD-based low-rank approximation)
+- If used for attention mechanism design -> `spectral-attention` (design pattern layer)
+- If spectral concentration bounds are needed -> `../probability/concentration-inequality.md` (concentration inequalities for random matrix spectra)
+
+## Extensible Directions
+- SVD variants (truncated / randomized SVD): fast decomposition for large-scale matrices
+- CUR decomposition: interpretable matrix approximation via column/row sampling
+- Nystrom method: low-rank approximation of kernel matrices
+- Spectral graph theory (Laplacian eigenvalues): graph Laplacian eigenvalues and graph structure analysis
+- Random matrix theory: spectral distributions of large random matrices

package/knowledge-base/matrix-analysis/spectral-decomposition.md

@@ -44,5 +44,18 @@
 
 ## 深入参考
 
-- 蒸馏稿：references/books/matrix-analysis.md（Ch 1 特征值与相似、§2.4-2.5 Schur 三角化与正规矩阵、§4.2 Courant-Fischer）
+- 蒸馏稿：../../references/books/matrix-analysis.md（Ch 1 特征值与相似、§2.4-2.5 Schur 三角化与正规矩阵、§4.2 Courant-Fischer）
 - 原书：Horn & Johnson, *Matrix Analysis* 2nd Ed., Chapter 1 (Eigenvalues, Eigenvectors, Similarity) + Chapter 2 (Unitary Similarity §2.4-2.5)
+
+
+## 路由扩展
+- 若需要截断近似 → `low-rank-approximation.md`（基于 SVD 的低秩近似）
+- 若用于注意力机制设计 → `spectral-attention`（设计模式层）
+- 若需要谱的集中界 → `../probability/concentration-inequality.md`（随机矩阵谱的集中不等式）
+
+## 可扩展方向
+- SVD 变体（truncated / randomized SVD）：大规模矩阵的快速分解
+- CUR 分解：基于列/行采样的可解释矩阵近似
+- Nystrom 方法：核矩阵的低秩近似
+- 谱图理论（Laplacian eigenvalues）：图拉普拉斯特征值与图结构分析
+- 随机矩阵理论：大维随机矩阵的谱分布