Crate quadprec

Modules§

backward_solve: Backward solve: L^T * x = y (L is lower-triangular, row-major n x n).
cholesky: In-place Cholesky: A = L * L^T. Row-major, n x n. Returns the lower-triangular factor in the lower triangle of a.
cholesky_blocked: Blocked Cholesky factorization: A = L * L^T. Row-major, n x n.
cond_estimate: Condition number estimate from Cholesky diagonal: kappa ~ (max/min)^2.
dot: Dot product of two f128 slices, parallelized via Rayon for large inputs.
forward_solve: Forward solve: L * y = b (L is lower-triangular, row-major n x n).
gemm: C = A * B, row-major. A is m×k, B is k×n, C is m×n.
gemm_atb: C = A^T * B, row-major. A is k×m (so A^T is m×k), B is k×n, C is m×n.
gemv: y = A * x, A is row-major [rows x cols].
jacobi_eigen: Classic cyclic Jacobi eigenvalue solver for symmetric matrices. Returns (eigenvalues, eigenvectors) where eigenvectors are row-major n x n.
matvec: Matrix-vector product y = A * x (row-major, n x n).
solve_cholesky: Solve L L^T x = b. L must already be factored.