Files
lnet_tutor/learning_plans/linear_algebra/01_KNOWLEDGE_GRAPH.md
2025-10-22 20:14:31 +08:00

24 KiB
Raw Blame History

Linear Algebra Knowledge Graph - Complete Dependency Map

🌳 Knowledge Tree Structure

This document maps all Linear Algebra concepts with their dependencies and optimal learning order.


Level 1: Foundation Concepts (No Prerequisites)

1.1 Basic Algebra Review

┌──────────────────────────────┐
│ Arithmetic Operations        │
│ - Addition, Subtraction      │
│ - Multiplication, Division   │
│ - Order of operations        │
└──────────────────────────────┘
        │
        ├─> Algebraic Expressions
        ├─> Solving Linear Equations
        ├─> Polynomials
        └─> Summation Notation (Σ)

1.2 Coordinate Systems

┌──────────────────────────────┐
│ Cartesian Coordinates        │
│ - 2D plane (x, y)           │
│ - 3D space (x, y, z)        │
│ - Points and plotting       │
└──────────────────────────────┘
        │
        ├─> Distance Formula
        ├─> Midpoint Formula
        └─> Graphing Functions

Level 2: Vectors - Geometric (Requires Level 1)

2.1 Vector Basics

┌──────────────────────────────┐
│ Vector Definition            │
│ - Magnitude and Direction   │
│ - Component Form            │
│ - Position Vectors          │
└──────────────────────────────┘
        │
        ├─> Vector Notation (bold, arrow)
        ├─> ℝ² and ℝ³ vectors
        ├─> n-dimensional vectors (ℝⁿ)
        └─> Column vs Row Vectors

┌──────────────────────────────┐
│ Vector Visualization         │
│ - Geometric arrows          │
│ - Head and tail             │
│ - Parallel vectors          │
└──────────────────────────────┘

2.2 Vector Operations (Geometric)

┌──────────────────────────────┐
│ Vector Addition              │
│ - Parallelogram Rule        │
│ - Tip-to-tail Method        │
│ - Component-wise Addition   │
└──────────────────────────────┘
        │
        ├─> Vector Subtraction
        ├─> Scalar Multiplication (Scaling)
        ├─> Linear Combinations
        └─> Zero Vector

┌──────────────────────────────┐
│ Special Vectors              │
│ - Unit Vectors              │
│ - Standard Basis (i, j, k)  │
│ - Normalization             │
└──────────────────────────────┘

Level 3: Vector Products (Requires Level 2)

3.1 Dot Product (Inner Product)

┌──────────────────────────────┐
│ Dot Product                  │
│ - a · b = Σ aᵢbᵢ           │
│ - a · b = ||a|| ||b|| cos θ│
│ - Scalar result             │
└──────────────────────────────┘
        │
        ├─> Vector Length (Norm): ||v|| = √(v · v)
        ├─> Distance: ||a - b||
        ├─> Angle Between Vectors
        ├─> Orthogonality (a · b = 0)
        ├─> Vector Projection
        └─> Cauchy-Schwarz Inequality

┌──────────────────────────────┐
│ Properties of Dot Product    │
│ - Commutative: a · b = b · a│
│ - Distributive              │
│ - Linearity                 │
└──────────────────────────────┘

3.2 Cross Product (3D Only)

┌──────────────────────────────┐
│ Cross Product (a × b)        │
│ - Vector result             │
│ - Perpendicular to both     │
│ - Right-hand rule           │
└──────────────────────────────┘
        │
        ├─> Magnitude: ||a × b|| = ||a|| ||b|| sin θ
        ├─> Area of Parallelogram
        ├─> Determinant Form
        ├─> Anti-commutative: a × b = -(b × a)
        └─> Triple Scalar Product

┌──────────────────────────────┐
│ Applications                 │
│ - Normal vectors            │
│ - Torque calculations       │
│ - Area and volume           │
└──────────────────────────────┘

Level 4: Matrices - Basics (Requires Level 2-3)

4.1 Matrix Fundamentals

┌──────────────────────────────┐
│ Matrix Definition            │
│ - m × n array of numbers    │
│ - Rows and columns          │
│ - Matrix indexing Aᵢⱼ       │
└──────────────────────────────┘
        │
        ├─> Matrix Addition/Subtraction
        ├─> Scalar Multiplication
        ├─> Transpose (Aᵀ)
        ├─> Special Matrices (I, O, Diagonal)
        └─> Matrix Equality

4.2 Matrix Multiplication

┌──────────────────────────────┐
│ Matrix Product              │
│ - (AB)ᵢⱼ = Σ AᵢₖBₖⱼ        │
│ - Dimension compatibility   │
│ - Non-commutative           │
└──────────────────────────────┘
        │
        ├─> Properties (Associative, Distributive)
        ├─> Identity: AI = IA = A
        ├─> Matrix Powers: A², A³, ...
        ├─> Matrix as Linear Transformation
        └─> Block Matrix Multiplication

Level 5: Linear Systems (Requires Level 4)

5.1 Systems of Linear Equations

┌──────────────────────────────┐
│ System Representation        │
│ - Ax = b                    │
│ - Augmented Matrix [A|b]    │
│ - Coefficient Matrix        │
└──────────────────────────────┘
        │
        ├─> Gaussian Elimination
        ├─> Row Operations
        ├─> Row Echelon Form (REF)
        ├─> Reduced Row Echelon Form (RREF)
        └─> Back Substitution

┌──────────────────────────────┐
│ Solution Types               │
│ - Unique Solution           │
│ - Infinite Solutions        │
│ - No Solution               │
└──────────────────────────────┘
        │
        ├─> Consistency
        ├─> Homogeneous Systems
        ├─> Parametric Solutions
        └─> Geometric Interpretation

Level 6: Matrix Inverses & Determinants (Requires Level 5)

6.1 Matrix Inverse

┌──────────────────────────────┐
│ Inverse Definition           │
│ - AA⁻¹ = A⁻¹A = I          │
│ - Exists iff det(A) ≠ 0     │
│ - Unique if exists          │
└──────────────────────────────┘
        │
        ├─> Computing Inverses (Gauss-Jordan)
        ├─> Inverse Properties: (AB)⁻¹ = B⁻¹A⁻¹
        ├─> Inverse and Transpose: (Aᵀ)⁻¹ = (A⁻¹)ᵀ
        ├─> Solving Systems: x = A⁻¹b
        └─> Invertible Matrix Theorem

6.2 Determinants

┌──────────────────────────────┐
│ Determinant                  │
│ - det(A) or |A|             │
│ - Scalar value              │
│ - Invertibility test        │
└──────────────────────────────┘
        │
        ├─> 2×2: ad - bc
        ├─> 3×3: Rule of Sarrus or Cofactor
        ├─> n×n: Cofactor Expansion
        ├─> Properties: det(AB) = det(A)det(B)
        ├─> det(Aᵀ) = det(A)
        ├─> Row Operations Effect
        ├─> Cramer's Rule
        └─> Geometric Meaning (Area/Volume)

Level 7: Vector Spaces (Requires Level 2-6)

7.1 Abstract Vector Spaces

┌──────────────────────────────┐
│ Vector Space Definition      │
│ - 10 Axioms                 │
│ - Closure under + and ·     │
│ - Examples: ℝⁿ, Polynomials │
└──────────────────────────────┘
        │
        ├─> Subspaces
        ├─> Span of Vectors
        ├─> Linear Independence
        ├─> Linear Dependence
        ├─> Basis
        └─> Dimension

┌──────────────────────────────┐
│ Important Subspaces          │
│ - Null Space (Kernel)       │
│ - Column Space (Range)      │
│ - Row Space                 │
│ - Left Null Space           │
└──────────────────────────────┘
        │
        ├─> Rank of Matrix
        ├─> Nullity of Matrix
        ├─> Rank-Nullity Theorem
        └─> Fundamental Theorem of Linear Algebra

7.2 Basis & Dimension

┌──────────────────────────────┐
│ Basis                        │
│ - Linearly independent      │
│ - Spans the space           │
│ - Minimum spanning set      │
└──────────────────────────────┘
        │
        ├─> Standard Basis
        ├─> Dimension = # basis vectors
        ├─> Change of Basis
        ├─> Coordinates Relative to Basis
        └─> Uniqueness of Dimension

Level 8: Linear Transformations (Requires Level 7)

8.1 Linear Transformations

┌──────────────────────────────┐
│ Transformation T: V → W      │
│ - T(u + v) = T(u) + T(v)    │
│ - T(cv) = cT(v)             │
│ - Matrix representation     │
└──────────────────────────────┘
        │
        ├─> Kernel (Null Space): ker(T) = {v : T(v) = 0}
        ├─> Range (Image): range(T) = {T(v) : v ∈ V}
        ├─> Rank-Nullity Theorem
        ├─> One-to-one Transformations
        ├─> Onto Transformations
        └─> Isomorphisms

┌──────────────────────────────┐
│ Standard Transformations     │
│ - Rotation                  │
│ - Reflection                │
│ - Projection                │
│ - Scaling                   │
└──────────────────────────────┘
        │
        └─> Composition of Transformations

Level 9: Eigenvalues & Eigenvectors (Requires Level 6-8)

9.1 Eigen-Theory

┌──────────────────────────────┐
│ Eigenvalue Problem           │
│ - Av = λv                   │
│ - Characteristic Polynomial │
│ - det(A - λI) = 0           │
└──────────────────────────────┘
        │
        ├─> Computing Eigenvalues
        ├─> Computing Eigenvectors
        ├─> Eigenspace
        ├─> Algebraic Multiplicity
        ├─> Geometric Multiplicity
        └─> Diagonalization

┌──────────────────────────────┐
│ Diagonalization              │
│ - A = PDP⁻¹                 │
│ - D diagonal (eigenvalues)  │
│ - P columns (eigenvectors)  │
└──────────────────────────────┘
        │
        ├─> Diagonalizable Matrices
        ├─> Similar Matrices
        ├─> Powers: Aⁿ = PDⁿP⁻¹
        └─> Applications: Differential Equations, Markov Chains

Level 10: Orthogonality (Requires Level 3, 7)

10.1 Orthogonal Sets

┌──────────────────────────────┐
│ Orthogonality                │
│ - v · w = 0                 │
│ - Perpendicular vectors     │
│ - Orthogonal sets           │
└──────────────────────────────┘
        │
        ├─> Orthogonal Basis
        ├─> Orthonormal Basis
        ├─> Orthogonal Complement
        └─> Orthogonal Decomposition

┌──────────────────────────────┐
│ Gram-Schmidt Process         │
│ - Orthogonalization         │
│ - Creates orthonormal basis │
│ - QR Decomposition          │
└──────────────────────────────┘
        │
        └─> Applications: Least Squares, QR Algorithm

10.2 Orthogonal Matrices

┌──────────────────────────────┐
│ Orthogonal Matrix Q          │
│ - QᵀQ = QQᵀ = I            │
│ - Columns orthonormal       │
│ - Preserves lengths         │
└──────────────────────────────┘
        │
        ├─> Rotation Matrices
        ├─> Reflection Matrices
        ├─> det(Q) = ±1
        └─> Q⁻¹ = Qᵀ

Level 11: Inner Product Spaces (Requires Level 3, 7, 10)

11.1 Inner Products

┌──────────────────────────────┐
│ Inner Product ⟨u, v⟩         │
│ - Generalizes dot product   │
│ - 4 Axioms                  │
│ - Induces norm & metric     │
└──────────────────────────────┘
        │
        ├─> Cauchy-Schwarz Inequality
        ├─> Triangle Inequality
        ├─> Parallelogram Law
        ├─> Pythagorean Theorem
        └─> Norm: ||v|| = √⟨v, v⟩

┌──────────────────────────────┐
│ Applications                 │
│ - Function spaces           │
│ - Polynomial inner products │
│ - Weighted inner products   │
└──────────────────────────────┘

Level 12: Matrix Decompositions (Requires Level 6, 9, 10)

12.1 LU Decomposition

┌──────────────────────────────┐
│ LU Factorization             │
│ - A = LU                    │
│ - L: Lower triangular       │
│ - U: Upper triangular       │
└──────────────────────────────┘
        │
        ├─> Existence Conditions
        ├─> Computing LU
        ├─> Solving Systems with LU
        ├─> Computational Efficiency
        └─> PLU (with Pivoting)

12.2 QR Decomposition

┌──────────────────────────────┐
│ QR Factorization             │
│ - A = QR                    │
│ - Q: Orthogonal             │
│ - R: Upper triangular       │
└──────────────────────────────┘
        │
        ├─> Gram-Schmidt Method
        ├─> Householder Reflections
        ├─> Givens Rotations
        ├─> Least Squares Solutions
        └─> QR Algorithm for Eigenvalues

12.3 Eigenvalue Decomposition (Spectral)

┌──────────────────────────────┐
│ Spectral Decomposition       │
│ - A = QΛQᵀ                  │
│ - Symmetric matrices        │
│ - Real eigenvalues          │
└──────────────────────────────┘
        │
        ├─> Orthogonal Eigenvectors
        ├─> Spectral Theorem
        ├─> Applications
        └─> Positive Definite Matrices

12.4 Singular Value Decomposition (SVD)

┌──────────────────────────────┐
│ SVD: A = UΣVᵀ                │
│ - U: Left singular vectors  │
│ - Σ: Singular values        │
│ - V: Right singular vectors │
└──────────────────────────────┘
        │
        ├─> Always Exists (any matrix)
        ├─> Singular Values
        ├─> Relationship to Eigenvalues
        ├─> Pseudoinverse (A⁺)
        ├─> Low-rank Approximation
        ├─> Image Compression
        ├─> Data Analysis (PCA)
        └─> Recommender Systems

Level 13: Advanced Theory (Requires Level 7-12)

13.1 Abstract Algebra Connections

┌──────────────────────────────┐
│ Algebraic Structures         │
│ - Groups                    │
│ - Rings                     │
│ - Fields                    │
└──────────────────────────────┘
        │
        ├─> Vector Space as Module
        ├─> Linear Algebra over Fields
        └─> Quotient Spaces

13.2 Norms & Metrics

┌──────────────────────────────┐
│ Vector Norms                 │
│ - L¹ norm: Σ|vᵢ|           │
│ - L² norm (Euclidean)       │
│ - L∞ norm: max|vᵢ|         │
│ - p-norms                   │
└──────────────────────────────┘
        │
        ├─> Matrix Norms
        ├─> Frobenius Norm
        ├─> Operator Norm
        ├─> Condition Number
        └─> Error Analysis

┌──────────────────────────────┐
│ Metric Spaces                │
│ - Distance Function         │
│ - Metric Properties         │
│ - Induced by Norms          │
└──────────────────────────────┘

Level 14: Applications - Machine Learning (Requires All Previous)

14.1 ML Fundamentals

┌──────────────────────────────┐
│ Linear Regression            │
│ - Normal Equations          │
│ - θ = (XᵀX)⁻¹Xᵀy           │
│ - Least Squares             │
└──────────────────────────────┘
        │
        ├─> Gradient Descent
        ├─> Ridge Regression (L2)
        ├─> Lasso Regression (L1)
        └─> Regularization

┌──────────────────────────────┐
│ Dimensionality Reduction     │
│ - PCA (Principal Components)│
│ - SVD for PCA               │
│ - Explained Variance        │
└──────────────────────────────┘
        │
        ├─> Eigenfaces
        ├─> Feature Extraction
        ├─> Data Visualization
        └─> Compression

14.2 Neural Networks

┌──────────────────────────────┐
│ Neural Network Math          │
│ - Forward Pass: y = Wx + b  │
│ - Backpropagation           │
│ - Gradient Computation      │
└──────────────────────────────┘
        │
        ├─> Weight Matrices
        ├─> Activation Functions
        ├─> Loss Gradients
        └─> Optimization (SGD, Adam)

Level 15: Applications - Graphics (Requires Level 4, 8)

15.1 Geometric Transformations

┌──────────────────────────────┐
│ 2D Transformations           │
│ - Translation               │
│ - Rotation                  │
│ - Scaling                   │
│ - Shearing                  │
└──────────────────────────────┘
        │
        ├─> Homogeneous Coordinates
        ├─> Transformation Matrices
        ├─> Composition
        └─> Inverse Transformations

┌──────────────────────────────┐
│ 3D Graphics                  │
│ - 3D Rotations              │
│ - View Transformations      │
│ - Projection (Orthographic) │
│ - Projection (Perspective)  │
└──────────────────────────────┘
        │
        ├─> Camera Matrices
        ├─> Model-View-Projection
        ├─> Quaternions
        └─> Euler Angles

🔗 Dependency Map Summary

Critical Learning Path

Level 1 (Algebra Review)
    ↓
Level 2 (Vectors - Geometric)
    ↓
Level 3 (Vector Products)
    ↓
Level 4 (Matrices - Basics)
    ↓
Level 5 (Linear Systems)
    ↓
Level 6 (Inverses & Determinants)
    ↓
Level 7 (Vector Spaces) [Theoretical Branch]
    ↓
Level 8 (Linear Transformations)
    ↓
Level 9 (Eigenvalues) ←─────────┐
    ↓                           │
Level 10 (Orthogonality) ───────┤ Can parallelize
    ↓                           │
Level 11 (Inner Products) ──────┘
    ↓
Level 12 (Decompositions)
    ↓
Level 13 (Advanced Theory) ←────┐
    ↓                           │ Can parallelize
Level 14 (ML Applications) ─────┤
    ↓                           │
Level 15 (Graphics Applications)┘

Parallel Learning Opportunities

  • Levels 9, 10, 11 can be learned in parallel after Level 8
  • Level 13 (theory) can parallel with Levels 14-15 (applications)
  • Applications (14-15) depend on decompositions but can be learned in any order

📊 Prerequisite Matrix

Topic Must Know First Can Learn In Parallel
Dot Product Vector basics Cross product
Matrices Vectors -
Matrix Mult Matrix basics Transpose
Linear Systems Matrix multiplication -
Determinants Matrix multiplication Inverses
Inverses Determinants, Systems -
Vector Spaces Linear systems, Span -
Eigenvalues Determinants, Vector spaces -
Orthogonality Dot product, Basis Inner products
SVD Eigenvalues, Orthogonality -
PCA SVD, Statistics basics -

🎯 Learning Strategies

Geometric First, Then Abstract

  1. Start with 2D/3D vectors (can visualize)
  2. Build geometric intuition
  3. Generalize to n dimensions
  4. Then study abstract theory
  5. Apply to real problems

Computation Supports Theory

  1. Solve many numerical examples
  2. Use Python/MATLAB to verify
  3. See patterns emerge
  4. Then learn why (proofs)
  5. Deepen understanding

Applications Motivate Learning

  1. See where linear algebra is used
  2. Understand why we need it
  3. Learn concepts to solve problems
  4. Apply immediately
  5. Build projects

This knowledge graph ensures you build strong foundations before tackling abstract concepts and applications!