Visual Tools
Calculators
Tables
Mathematical Keyboard
Converters
Other Tools

Matrix Decomposition Methods: LU, QR, Cholesky, Spectral, SVD

Every matrix decomposition on one page, each computed step by step: LU, QR, Cholesky, spectral and SVD, with a guide to choosing. Each links to the decompositions lesson.

Level
I

The five decompositions

Composite · each opens its own page
How to find the LU decompositionRun Gaussian elimination, keep the multipliers in L and the result in U; pivot when a zero appears.uses: Gaussian elimination · multipliers · permutation matrixComposite5 steps · 3 examplesHow to find the QR decompositionRun Gram-Schmidt on the columns: the orthonormal vectors form Q, the coefficients form R.uses: Gram-Schmidt · dot product · normalizationComposite5 steps · 2 examplesHow to find the Cholesky decompositionSymmetric positive definite matrix: build L column by column with square roots so that A=LLTA = LL^T.uses: square roots · symmetric matrix · positive definitenessComposite5 steps · 3 examplesHow to find the spectral decompositionSymmetric matrix: orthonormal eigenvectors and eigenvalues give A=QDQTA = QDQ^T, a sum of projections.uses: eigenvalues · eigenvectors · Gram-SchmidtComposite5 steps · 3 examplesHow to find the singular value decompositionEigenvalues of ATAA^TA give the singular values and VV; then ui=Avi/σi\mathbf{u}_i = A\mathbf{v}_i / \sigma_i gives UU.uses: eigenvalues · spectral decomposition · normalizationComposite5 steps · 3 examples
II

Choosing a decomposition

Strategy