From fddf37b8ef49cdaea1b1500a44c20e16981f29a9 Mon Sep 17 00:00:00 2001
From: Norihiro Watanabe <norihiro.watanabe@ufz.de>
Date: Fri, 10 Jul 2015 09:37:20 +0200
Subject: [PATCH] improved error messages
---
MathLib/LinAlg/Eigen/EigenLinearSolver.cpp | 42 +++++++++++-----------
1 file changed, 22 insertions(+), 20 deletions(-)
diff --git a/MathLib/LinAlg/Eigen/EigenLinearSolver.cpp b/MathLib/LinAlg/Eigen/EigenLinearSolver.cpp
index 004569349b5..4a38c4c77ee 100644
--- a/MathLib/LinAlg/Eigen/EigenLinearSolver.cpp
+++ b/MathLib/LinAlg/Eigen/EigenLinearSolver.cpp
@@ -24,30 +24,31 @@ using boost::property_tree::ptree;
namespace details
{
+/// Template class for Eigen direct linear solvers
template <class T_SOLVER>
-class EigenDirectSolver : public EigenLinearSolver::IEigenSolver
+class EigenDirectLinearSolver : public EigenLinearSolver::IEigenSolver
{
public:
- EigenDirectSolver(EigenMatrix::RawMatrixType &A) : _A(A)
+ explicit EigenDirectLinearSolver(EigenMatrix::RawMatrixType &A) : _A(A)
{
- // fill A and b;
- // Compute the ordering permutation vector from the structural pattern of A
- _solver.analyzePattern(A);
- // Compute the numerical factorization
- _solver.factorize(A);
+ INFO("-> initialize with the coefficient matrix");
+ _solver.compute(A);
if(_solver.info()!=Eigen::Success) {
- ERR("The numerical factorization failed in Eigen");
+ ERR("Failed during Eigen linear solver initialization");
return;
}
}
- virtual ~EigenDirectSolver() {}
+ virtual ~EigenDirectLinearSolver() {}
- void solve(EigenVector::RawVectorType &b, EigenVector::RawVectorType &x, EigenOption &) override
+ void solve(EigenVector::RawVectorType &b, EigenVector::RawVectorType &x, EigenOption &/*opt*/) override
{
- //Use the factors to solve the linear system
INFO("-> solve");
x = _solver.solve(b);
+ if(_solver.info()!=Eigen::Success) {
+ ERR("Failed during Eigen linear solve");
+ return;
+ }
}
private:
@@ -55,21 +56,22 @@ private:
EigenMatrix::RawMatrixType& _A;
};
+/// Template class for Eigen iterative linear solvers
template <class T_SOLVER>
-class EigenIterativeSolver : public EigenLinearSolver::IEigenSolver
+class EigenIterativeLinearSolver : public EigenLinearSolver::IEigenSolver
{
public:
- EigenIterativeSolver(EigenMatrix::RawMatrixType &A) : _A(A)
+ explicit EigenIterativeLinearSolver(EigenMatrix::RawMatrixType &A) : _A(A)
{
INFO("-> initialize with the coefficient matrix");
_solver.compute(A);
if(_solver.info()!=Eigen::Success) {
- INFO("\t failed");
+ ERR("Failed during Eigen linear solver initialization");
return;
}
}
- virtual ~EigenIterativeSolver() {}
+ virtual ~EigenIterativeLinearSolver() {}
void solve(EigenVector::RawVectorType &b, EigenVector::RawVectorType &x, EigenOption &opt) override
{
@@ -78,8 +80,8 @@ public:
_solver.setMaxIterations(opt.max_iterations);
x = _solver.solveWithGuess(b, x);
if(_solver.info()!=Eigen::Success) {
- INFO("\t solving failed");
- return;
+ ERR("Failed during Eigen linear solve");
+ return;
}
INFO("\t iteration: %d/%ld", _solver.iterations(), opt.max_iterations);
INFO("\t residual: %e\n", _solver.error());
@@ -100,13 +102,13 @@ EigenLinearSolver::EigenLinearSolver(EigenMatrix &A, ptree const*const option)
A.getRawMatrix().makeCompressed();
if (_option.solver_type==EigenOption::SolverType::SparseLU) {
typedef Eigen::SparseLU<EigenMatrix::RawMatrixType, Eigen::COLAMDOrdering<int> > SolverType;
- _solver = new details::EigenDirectSolver<SolverType>(A.getRawMatrix());
+ _solver = new details::EigenDirectLinearSolver<SolverType>(A.getRawMatrix());
} else if (_option.solver_type==EigenOption::SolverType::BiCGSTAB) {
typedef Eigen::BiCGSTAB<EigenMatrix::RawMatrixType, Eigen::DiagonalPreconditioner<double>> SolverType;
- _solver = new details::EigenIterativeSolver<SolverType>(A.getRawMatrix());
+ _solver = new details::EigenIterativeLinearSolver<SolverType>(A.getRawMatrix());
} else if (_option.solver_type==EigenOption::SolverType::CG) {
typedef Eigen::ConjugateGradient<EigenMatrix::RawMatrixType, Eigen::Lower, Eigen::DiagonalPreconditioner<double>> SolverType;
- _solver = new details::EigenIterativeSolver<SolverType>(A.getRawMatrix());
+ _solver = new details::EigenIterativeLinearSolver<SolverType>(A.getRawMatrix());
}
}
--
GitLab