Understanding Polar Decomposition
The polar decomposition is a fundamental factorization in linear algebra. Any square matrix A can be written as A = UP, where U is a unitary (orthogonal for real matrices) matrix and P is a positive semi-definite Hermitian matrix. This decomposition is analogous to writing a complex number in polar form z = r*e^(i*theta).
Key Components
Matrix A
The input square matrix to be decomposed.
Unitary Matrix U
Represents a rotation (and possibly reflection). U^T * U = I.
Positive Semi-Definite P
Represents a scaling. P is symmetric with non-negative eigenvalues.
A^T * A
Always symmetric positive semi-definite. Its eigenvalues are the squared singular values.
Algorithm
- Compute A^T * A (the Gram matrix).
- Find the eigenvalues and eigenvectors of A^T * A.
- Compute P = sqrt(A^T * A) using the eigendecomposition.
- If A is invertible, compute U = A * P^(-1).
Applications
- Computer Graphics: Separating rotation from scaling in transformations.
- Continuum Mechanics: Decomposing deformation into rotation and stretch.
- Signal Processing: Matrix analysis and filtering.
- Robotics: Analyzing rigid body motions and deformations.
Relationship to SVD
The polar decomposition is closely related to the Singular Value Decomposition (SVD). If A = U_svd * Sigma * V^T is the SVD, then P = V * Sigma * V^T and U = U_svd * V^T. This connection makes polar decomposition computable via SVD.