klu2_simple.cpp File Reference

#include <complex>
#include <iostream>
#include <stdexcept>
#include <vector>
#include "klu2_defaults.hpp"
#include "klu2_analyze.hpp"
#include "klu2_factor.hpp"
#include "klu2_solve.hpp"
#include "klu2_tsolve.hpp"
#include "klu2_free_symbolic.hpp"
#include "klu2_free_numeric.hpp"

Include dependency graph for klu2_simple.cpp:

Macros
#define	ASSERT_EQ(a, b)   if(a != b) throw std::runtime_error(#a " is different from " #b);
	Check that a and b are equal.
#define	ASSERT_TOL_EQ(a, b, tol)   if( std::abs(a-b) > tol ) throw std::runtime_error(#a " is different from " #b);
	Check that a and b are equal to the given asolute tolerance.

Functions
template<typename T , typename tol_t >
void	print_and_compare (const std::vector< T > &values_a, const std::vector< T > &values_b, const tol_t tol)
	Print vectors and compare them.
template<typename T , typename D = int, typename tol_t >
void	solve_and_check (const tol_t tol)
int	main (void)

Data.
This simple example will solve the following linear system: * * Matrix Solution Right-hand side * * | 2. | 3. | | | | [ 1 ] [ 8 ] * | 3. | | 4. | | 6. | [ 2 ] [ 45 ] * | | -1. | -3. | 2. | | [ 3 ] = [ -3 ] * | | | 1. | | | [ 4 ] [ 3 ] * | | 4. | 2. | | 1. | [ 5 ] [ 19 ] * * Warning The algorithm works with the compressed column storage format (CCS).
const int	size = 5
int	Ap [] = {0, 2, 5, 9, 10, 12}
	Column offsets.
int	Ai [] = { 0 , 1 , 0 , 2 , 4 , 1 , 2 , 3 , 4 , 2 , 1 , 4 }
	Row values.
template<typename T >
auto	get_Ax ()
	Get entries values.
template<typename T >
auto	get_b ()
	Get right-hand side values.
template<typename T >
auto	get_x ()
	Get solution.
template<typename T >
auto	get_x_transpose ()
	Get solution for transposed matrix.

Detailed Description

This file contains a standalone test for the KLU2 solver.

Function Documentation

template<typename T , typename D = int, typename tol_t >

void solve_and_check

(

const tol_t

tol

)

Solve and check the solution.

Query data for

Template Parameters

T.

Symbolic step.

Numeric step.

Solve and check against expected solution. Note that klu_solve writes the solution into b.

Transpose solve. Note that klu_solve writes the solution into b.

Free memory.

References Ai, Ap, and size.

int main

(

void

)

Solve for float.

Solve for double.

Solve for std::complex<float>.

Todo:

This does not compile for now!

Examples:

MultipleSolves_File.cpp, SimpleSolve.cpp, SimpleSolve_File.cpp, SimpleSolve_WithParameters.cpp, and TwoPartSolve.cpp.

Variable Documentation

const int size = 5

Size of the system.

Referenced by Amesos2::MultiVecAdapter< Epetra_MultiVector >::getVector(), Amesos2::MultiVecAdapter< Epetra_MultiVector >::getVectorNonConst(), Amesos2::cuSOLVER< Matrix, Vector >::numericFactorization_impl(), solve_and_check(), Amesos2::cuSOLVER< Matrix, Vector >::solve_impl(), and Amesos2::cuSOLVER< Matrix, Vector >::symbolicFactorization_impl().