solver.hpp File Reference

Solver utility functions. More...

#include <algorithm>
#include <cmath>
#include <concepts>
#include <vector>
#include <iganet/core/core.hpp>

Go to the source code of this file.

Namespaces
namespace	iganet
namespace	iganet::utils
namespace	iganet::utils::detail

Concepts
concept	iganet::utils::IterativeSolverPreconditioner
	Specifies the callable interface required by the preconditioned iterative solvers.

Functions
auto	iganet::utils::bicgstab (const torch::Tensor &A, const torch::Tensor b, int max_iter=1000, double tol=1e-10)
	Solves the linear system A * x = b using the Bi-Conjugate Gradient Stabilized (BiCGStab) method.
auto	iganet::utils::cg (const torch::Tensor &A, const torch::Tensor b, int max_iter=1000, double tol=1e-10)
	Solves the linear system A * x = b using the Conjugate Gradient (CG) method.
void	iganet::utils::detail::check_finite_scalar (const torch::Tensor &value, const char *message)
	Provides the check_finite_scalar operation.
void	iganet::utils::detail::check_finite_tensor (const torch::Tensor &value, const char *message)
	Provides the check_finite_tensor operation.
void	iganet::utils::detail::check_nonnegative_finite_scalar (const torch::Tensor &value, const char *message)
	Provides the check_nonnegative_finite_scalar operation.
void	iganet::utils::detail::check_nonzero_finite_scalar (const torch::Tensor &value, const char *message)
	Provides the check_nonzero_finite_scalar operation.
void	iganet::utils::detail::check_positive_finite_scalar (const torch::Tensor &value, const char *message)
	Provides the check_positive_finite_scalar operation.
auto	iganet::utils::fgmres (const torch::Tensor &A, const torch::Tensor b, const torch::Tensor &inverse_preconditioner, int max_iter=1000, double tol=1e-10, int restart=30)
	FGMRES overload taking the inverse preconditioner as a tensor.
template<IterativeSolverPreconditioner Preconditioner>
auto	iganet::utils::fgmres (const torch::Tensor &A, const torch::Tensor b, Preconditioner &&preconditioner, int max_iter=1000, double tol=1e-10, int restart=30)
	Solves A * x = b using restarted, right-preconditioned GMRES.
auto	iganet::utils::gmres (const torch::Tensor &A, const torch::Tensor b, int max_iter=1000, double tol=1e-10, int restart=30)
	Solves A * x = b using restarted GMRES.
auto	iganet::utils::minres (const torch::Tensor &A, const torch::Tensor b, int max_iter=1000, double tol=1e-10)
	Solves A * x = b using MINRES.
auto	iganet::utils::pbicgstab (const torch::Tensor &A, const torch::Tensor b, const torch::Tensor &inverse_preconditioner, int max_iter=1000, double tol=1e-10)
	PBiCGStab overload taking the inverse preconditioner as a tensor.
template<IterativeSolverPreconditioner Preconditioner>
auto	iganet::utils::pbicgstab (const torch::Tensor &A, const torch::Tensor b, Preconditioner &&preconditioner, int max_iter=1000, double tol=1e-10)
	Solves the linear system A * x = b using the preconditioned Bi-Conjugate Gradient Stabilized (PBiCGStab) method.
auto	iganet::utils::pcg (const torch::Tensor &A, const torch::Tensor b, const torch::Tensor &inverse_preconditioner, int max_iter=1000, double tol=1e-10)
	PCG overload taking the inverse preconditioner as a tensor.
template<IterativeSolverPreconditioner Preconditioner>
auto	iganet::utils::pcg (const torch::Tensor &A, const torch::Tensor b, Preconditioner &&preconditioner, int max_iter=1000, double tol=1e-10)
	Solves the linear system A * x = b using the preconditioned Conjugate Gradient (PCG) method.
auto	iganet::utils::pminres (const torch::Tensor &A, const torch::Tensor b, const torch::Tensor &inverse_preconditioner, int max_iter=1000, double tol=1e-10)
	PMINRES overload taking the inverse preconditioner as a tensor.
template<IterativeSolverPreconditioner Preconditioner>
auto	iganet::utils::pminres (const torch::Tensor &A, const torch::Tensor b, Preconditioner &&preconditioner, int max_iter=1000, double tol=1e-10)
	Solves A * x = b using preconditioned MINRES.
bool	iganet::utils::detail::tensor_values_are_finite (const torch::Tensor &tensor)
	Provides the tensor_values_are_finite operation.
void	iganet::utils::detail::validate_gmres_parameters (int restart)
	Provides the validate_gmres_parameters operation.
void	iganet::utils::detail::validate_inverse_preconditioner (const torch::Tensor &inverse_preconditioner, const torch::Tensor &A)
	Provides the validate_inverse_preconditioner operation.
void	iganet::utils::detail::validate_iterative_solver_inputs (const torch::Tensor &A, const torch::Tensor &b, int max_iter, double tol)
	Provides the validate_iterative_solver_inputs operation.
void	iganet::utils::detail::validate_preconditioner_output (const torch::Tensor &value, const torch::Tensor &residual)
	Provides the validate_preconditioner_output operation.

Detailed Description

Solver utility functions.

Author

Matthias Moller.

Copyright

This file is part of the IgANet project.

This Source Code Form is subject to the terms of the Mozilla Public License, v. 2.0. If a copy of the MPL was not distributed with this file, You can obtain one at http://mozilla.org/MPL/2.0/.