λvv
Algebra · Concept hub

Eigenvalues and Eigenvectors

1822 CE19th-century France (Cauchy)

Concept

Vectors a matrix only stretches without rotating. The "principal axes" of high-dimensional data — PageRank, PCA, quantum mechanics.

Understand it in one breath

A nonzero vector v that remains on the same line after multiplication by A, scaled by λ. A negative λ reverses orientation, and complex eigenvalues may be needed. PageRank models links as a stochastic transition matrix and approximates an eigenvector corresponding to its stationary distribution.

At a glance

Matrix A

Eigenvalue λ

Eigenvector v

Intuition

[[2,0],[0,3]]

2, 3

(1,0), (0,1)

Only the x- and y-axis directions stay fixed

[[3,1],[0,2]]

3, 2

(1,0), (1,−1)

Two tilted invariant axes

90° rotation [[0,-1],[1,0]]

Complex ±i

(1, ±i)

No real eigenvectors

PCA data covariance

Principal-component variance

Principal-axis direction

The data’s dominant directions

Google link matrix

1 (scalar)

PageRank vector

Stationary distribution of the web

Av = λv. Over the complex numbers, the characteristic polynomial of an n×n matrix has n roots counting multiplicity, but the matrix need not have n linearly independent eigenvectors.

Key formula

Av=λvA\mathbf{v} = \lambda \mathbf{v}

Av = λv

Modern applications

PCA dimensionality reduction, PageRank, energy states in quantum mechanics, and vibration-mode analysis.

Beyond MathVoyage

Loading…