Module impl_generic 

Generic implementations of iterative solvers

These implementations use primitive operations (BinaryOps, UnaryOps, ReduceOps) and can run on any backend that implements those traits.

Functions

adaptive_gmres_impl

Adaptive GMRES implementation

amg_preconditioned_cg

Apply AMG as a preconditioner within CG

amg_setup

Build the AMG multigrid hierarchy (setup phase)

amg_vcycle

Apply one AMG V-cycle: smooth → restrict → coarse-solve → prolongate → smooth

arnoldi_eig_impl

Generic Arnoldi eigensolver implementation

bicgstab_impl

Generic BiCGSTAB implementation

cg_impl

Generic preconditioned CG implementation

cgs_impl

Generic preconditioned CGS implementation

gmres_impl

Generic GMRES implementation

jacobi_impl

Generic weighted Jacobi implementation

lanczos_eig_impl

Generic Lanczos eigensolver implementation

lgmres_impl

Generic LGMRES implementation

minres_impl

Generic MINRES implementation following Saad, “Iterative Methods for Sparse Linear Systems”, 2nd ed., Algorithm 6.12.

qmr_impl

Generic QMR implementation

sor_impl

Generic SOR implementation via sparse triangular solve

svds_impl

Generic sparse SVD implementation via Lanczos bidiagonalization