Cholesky Decomposition
Visualize the factorization of a symmetric positive-definite matrix into a lower triangular matrix and its transpose.
Free theory articleVisualize the factorization of a symmetric positive-definite matrix into a lower triangular matrix and its transpose.
Concept & interpretation
A real symmetric positive-definite matrix A has a unique factorization A = LLT with L lower triangular and strictly positive diagonal entries. Transposing L exchanges its rows and columns. Multiplying a row of L by a row of L gives an entry of LLT, so the product is symmetric.
Predict a sign change, then check the matrices
Open Cholesky Decomposition. Each slider edits B; the displayed matrices are A, L and LT. On a small screen, scroll the control panel for its remaining sliders and the matrix panel for later matrices and formulas.
- Set B11 to 2, B21 to 2, B22 to 2 and B31 to −1. Check A22 = 8.00 and A23 = 0.0 against the calculation above.
- Change only B31 to 1. Now A13 = 2.00, A23 = 4.00 and A33 = 3.00. Squaring the changed entry preserves A33, while the cross products change sign or magnitude.
- Keep those values and reduce B11 to its minimum 0.1. Predict the first row (0.01, 0.20, 0.10); check L11 = 0.10. The determinant of A is (0.1·2·1)2 = 0.04, so this boundary remains positive definite.
- Restore B11 to 2, then reduce B22 to 0.1. The second diagonal becomes 4.01 and A23 becomes 2.10. Check that L22 is still positive.