Module lu 

LU Factorization with partial pivoting.

Computes P * A = L * U where:

• P is a permutation matrix (represented by pivot indices)
• L is unit lower triangular
• U is upper triangular

Uses a blocked right-looking algorithm:

1. Panel factorization: factor a narrow column panel using a dedicated GPU kernel
2. Apply pivots: swap rows in the trailing portion
3. TRSM: solve for the upper triangle block
4. GEMM: update the trailing submatrix

The L and U factors overwrite the input matrix A in-place (LAPACK-style packed storage with unit diagonal for L implicitly assumed).

Structs

LuResult

Result of an LU factorization.

Functions

lu_factorize

Performs LU factorization with partial pivoting in-place.

lu_solve

Solves A * X = B given an LU-factored matrix.