|
core
Shared MPI block decomposition + asynchronous ghost-layer exchange (header-only C++20)
|
Smoothed-aggregation AMG hierarchy usable as z = M⁻¹ r (one symmetric V-cycle from a zero initial guess) — the CG preconditioner that makes the iteration count mesh-independent.
More...
#include <graph_amg.hpp>
Public Member Functions | |
| void | build (const HostCsrOp &A, const AmgParams &prm={}) |
| void | apply (const std::vector< double > &r, std::vector< double > &z) const |
| z = M⁻¹ r: one V-cycle (correction scheme) from a zero initial guess. | |
| int | numLevels () const |
| Index | size (int L=0) const |
| double | operatorComplexity () const |
| Total nonzeros / finest nonzeros — the grid+operator complexity (should stay ~<2 for a healthy hierarchy; a useful health check when tuning theta). | |
Smoothed-aggregation AMG hierarchy usable as z = M⁻¹ r (one symmetric V-cycle from a zero initial guess) — the CG preconditioner that makes the iteration count mesh-independent.
Definition at line 89 of file graph_amg.hpp.
|
inline |
Definition at line 91 of file graph_amg.hpp.
|
inline |
z = M⁻¹ r: one V-cycle (correction scheme) from a zero initial guess.
Definition at line 111 of file graph_amg.hpp.
Referenced by peclet::core::solver::amgPcg().
|
inline |
Definition at line 119 of file graph_amg.hpp.
|
inline |
Definition at line 120 of file graph_amg.hpp.
References peclet::core::solver::HostCsrOp::n.
|
inline |
Total nonzeros / finest nonzeros — the grid+operator complexity (should stay ~<2 for a healthy hierarchy; a useful health check when tuning theta).
Definition at line 124 of file graph_amg.hpp.