The Cholesky Decomposition
What You'll Learn
You'll learn how to factor a positive definite matrix into a product of a lower triangular matrix and its transpose—a powerful tool for efficient computation.
The Core Idea
The Cholesky decomposition is a specialized factorization available only for positive definite matrices. If A is a positive definite matrix (meaning all its eigenvalues are positive and it's symmetric), you can write:
A = LL^T
where L is a lower triangular matrix with positive diagonal entries.
Think of it as finding the "square root" of a matrix. Just as a positive number a can be written as a = b², a positive definite matrix can be "square-rooted" into L times its transpose.
Why It Matters
The Cholesky decomposition is:
- More efficient than general LU decomposition (about half the computational work)
- Numerically stable for positive definite systems
- Unique when we require positive diagonal entries in L
It's particularly useful for solving systems Ax = b because once you have A = LL^T, you solve two simple triangular systems instead of working with A directly.
Connection to What You Know
You've seen that positive definite matrices have all positive eigenvalues (Eigenvalue Criterion for Definiteness) and can be confirmed using Sylvester's Criterion. These matrices have special structure that Cholesky exploits—unlike general matrices that need full LU or QR decomposition, positive definite matrices allow this simpler, elegant factorization.
Key Takeaway: The Cholesky decomposition factors any positive definite matrix A as A = LL^T where L is lower triangular with positive diagonal—providing an efficient, numerically stable alternative to general matrix factorizations for this special class of matrices.