I'm trying to solve something using the Conjugate Gradient method:

Code: Select all

Eigen::ConjugateGradient< Eigen::SparseMatrix<double> > cg;
cg.compute(iA);
iX = cg.solve(iB);


both iX and iB are of type VectorXd

I'm quite confident the sizes of the matrices and vectors match.

The problem I'm facing is I crash on this assert somewhere in the solve method.

assert:

Code: Select all

assertion failed: (i>=0) && ( ((BlockRows==1) && (BlockCols==XprType::ColsAtCompileTime) && i<xpr.rows()) ||((BlockRows==XprType::RowsAtCompileTime) && (BlockCols==1) && i<xpr.cols())) in function Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Block(XprType&, typename Eigen::internal::traits<Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true> >::Index) [with XprType = Eigen::Matrix<double, -0x00000000000000001, 1, 0, -0x00000000000000001, 1>, int BlockRows = 1, int BlockCols = 1, bool InnerPanel = false] at Eigen/Eigen/src/Core/Block.h:293

stacktrace:

Code: Select all

#4   0x00095a3a in Eigen::internal::assert_fail(char const*, char const*, char const*, int) at .../Eigen/Eigen/src/Core/util/Macros.h:208
#5   0x002a0b83 in Eigen::Block<Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1, 1, false, true>::Block(Eigen::Matrix<double, -1, 1, 0, -1, 1>&, int) ()
#6   0x002a0bd3 in Eigen::DenseBase<Eigen::Matrix<double, -1, 1, 0, -1, 1> >::row(int) ()
#7   0x002a2b95 in void Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1>::scaleAndAddTo<Eigen::Matrix<double, -1, 1, 0, -1, 1> >(Eigen::Matrix<double, -1, 1, 0, -1, 1>&, double) const ()
#8   0x002a2d2f in void Eigen::ProductBase<Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1>, Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1> >::scaleAndAddTo<Eigen::Matrix<double, -1, 1, 0, -1, 1> >(Eigen::Matrix<double, -1, 1, 0, -1, 1>&, double) const ()
#9   0x002a2d72 in void Eigen::ProductBase<Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1>, Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1> >::evalTo<Eigen::Matrix<double, -1, 1, 0, -1, 1> >(Eigen::Matrix<double, -1, 1, 0, -1, 1>&) const ()
#10   0x002a2db2 in Eigen::Matrix<double, -1, 1, 0, -1, 1>& Eigen::MatrixBase<Eigen::Matrix<double, -1, 1, 0, -1, 1> >::lazyAssign<Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1>, Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1> >(Eigen::ProductBase<Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1>, Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1> > const&) ()
#11   0x002a2e0c in Eigen::Matrix<double, -1, 1, 0, -1, 1>& Eigen::PlainObjectBase<Eigen::Matrix<double, -1, 1, 0, -1, 1> >::lazyAssign<Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1> >(Eigen::DenseBase<Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1> > const&) ()
#12   0x002a2e4c in Eigen::internal::assign_selector<Eigen::Matrix<double, -1, 1, 0, -1, 1>, Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1>, false, false>::run(Eigen::Matrix<double, -1, 1, 0, -1, 1>&, Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1> const&) ()
#13   0x002a2e8c in Eigen::Matrix<double, -1, 1, 0, -1, 1>& Eigen::NoAlias<Eigen::Matrix<double, -1, 1, 0, -1, 1>, Eigen::MatrixBase>::operator=<Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1> >(Eigen::MatrixBase<Eigen::SparseSelfAdjointTimeDenseProduct<Eigen::SparseMatrix<double, 0, int>, Eigen::Matrix<double, -1, 1, 0, -1, 1>, 1> > const&) ()
#14   0x002a4b2d in void Eigen::internal::conjugate_gradient<Eigen::SparseSelfAdjointView<Eigen::SparseMatrix<double, 0, int>, 1u>, Eigen::Block<Eigen::Matrix<double, -1, 1, 0, -1, 1> const, -1, 1, true, true>, Eigen::Block<Eigen::Matrix<double, -1, 1, 0, -1, 1>, -1, 1, true, true>, Eigen::DiagonalPreconditioner<double> >(Eigen::SparseSelfAdjointView<Eigen::SparseMatrix<double, 0, int>, 1u> const&, Eigen::Block<Eigen::Matrix<double, -1, 1, 0, -1, 1> const, -1, 1, true, true> const&, Eigen::Block<Eigen::Matrix<double, -1, 1, 0, -1, 1>, -1, 1, true, true>&, Eigen::DiagonalPreconditioner<double> const&, int&, Eigen::Block<Eigen::Matrix<double, -1, 1, 0, -1, 1>, -1, 1, true, true>::RealScalar&) ()
#15   0x002a5236 in void Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >::_solveWithGuess<Eigen::Matrix<double, -1, 1, 0, -1, 1>, Eigen::Matrix<double, -1, 1, 0, -1, 1> >(Eigen::Matrix<double, -1, 1, 0, -1, 1> const&, Eigen::Matrix<double, -1, 1, 0, -1, 1>&) const ()
#16   0x002a5350 in void Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >::_solve<Eigen::Matrix<double, -1, 1, 0, -1, 1>, Eigen::Matrix<double, -1, 1, 0, -1, 1> >(Eigen::Matrix<double, -1, 1, 0, -1, 1> const&, Eigen::Matrix<double, -1, 1, 0, -1, 1>&) const ()
#17   0x002a5397 in void Eigen::internal::solve_retval<Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >, Eigen::Matrix<double, -1, 1, 0, -1, 1> >::evalTo<Eigen::Matrix<double, -1, 1, 0, -1, 1> >(Eigen::Matrix<double, -1, 1, 0, -1, 1>&) const ()
#18   0x002a53c0 in void Eigen::internal::solve_retval_base<Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >, Eigen::Matrix<double, -1, 1, 0, -1, 1> >::evalTo<Eigen::Matrix<double, -1, 1, 0, -1, 1> >(Eigen::Matrix<double, -1, 1, 0, -1, 1>&) const ()
#19   0x002a53ea in void Eigen::ReturnByValue<Eigen::internal::solve_retval_base<Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >, Eigen::Matrix<double, -1, 1, 0, -1, 1> > >::evalTo<Eigen::Matrix<double, -1, 1, 0, -1, 1> >(Eigen::Matrix<double, -1, 1, 0, -1, 1>&) const ()
#20   0x002a541c in Eigen::Matrix<double, -1, 1, 0, -1, 1>& Eigen::MatrixBase<Eigen::Matrix<double, -1, 1, 0, -1, 1> >::operator=<Eigen::internal::solve_retval_base<Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >, Eigen::Matrix<double, -1, 1, 0, -1, 1> > >(Eigen::ReturnByValue<Eigen::internal::solve_retval_base<Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >, Eigen::Matrix<double, -1, 1, 0, -1, 1> > > const&) ()
#21   0x002a548b in Eigen::Matrix<double, -1, 1, 0, -1, 1>& Eigen::PlainObjectBase<Eigen::Matrix<double, -1, 1, 0, -1, 1> >::operator=<Eigen::internal::solve_retval_base<Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >, Eigen::Matrix<double, -1, 1, 0, -1, 1> > >(Eigen::ReturnByValue<Eigen::internal::solve_retval_base<Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >, Eigen::Matrix<double, -1, 1, 0, -1, 1> > > const&) at .../Eigen/Eigen/src/Core/PlainObjectBase.h:392
#22   0x002a54c0 in Eigen::Matrix<double, -1, 1, 0, -1, 1>& Eigen::Matrix<double, -1, 1, 0, -1, 1>::operator=<Eigen::internal::solve_retval_base<Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >, Eigen::Matrix<double, -1, 1, 0, -1, 1> > >(Eigen::ReturnByValue<Eigen::internal::solve_retval_base<Eigen::ConjugateGradient<Eigen::SparseMatrix<double, 0, int>, 1, Eigen::DiagonalPreconditioner<double> >, Eigen::Matrix<double, -1, 1, 0, -1, 1> > > const&) at .../Eigen/Eigen/src/Core/Matrix.h:203


Does anybody have an idea what is going wrong?

Thanks a lot

It only seems to happen when the lower triangle of the A matrix is used. If I explicitly specify to use the upper triangle:

Code: Select all

Eigen::ConjugateGradient< Eigen::SparseMatrix<double>, Eigen::Upper> cg;

the assert isn't fired.

Kind regards

How is your matrix? Is it complete, i.e., both the upper and lower parts are stored? only the lower? only the upper?

either way, I cannot reproduce. Make sure you have the latest development version.

If you still get the issue, could you send us your matrix:

#include <unsupported/Eigen/SparseExtra>

...

saveMarket(iA, "iA.mtx");

and send a gzipped version of the iA.mtx file to me.

thanks

I'm working with Eigen 3.1.0-alpha1 from the mainpage. Is this the latest version or should I get some version somewhere else. If so, where do I find this version?

The matrix I'm using is completely filled. It is constructed by multiplying the transpose of a SparseMatrix with the SparseMatrix itself.

Thanks

Solving using SimplicialLDLt does work. But the matrix could grow quite large, so I expect performance issues with this method, which is the reason I would prefer to use the conjugate gradient method.

Regarding a matrix with which I have issues, as I couldn't find your e-mail address immediately and this forums do not allow attachments as far as I know I'v copy-pasted the contents of the mtx file here:

Code: Select all

22 22 148
1 1 4.0467060940744698882554075680673122406005859375000000000000000000e+02
1 2 0.0000000000000000000000000000000000000000000000000000000000000000e+00
1 3 -7.6741879471417632885277271270751953125000000000000000000000000000e+01
1 4 2.4666340150362984218190831597894430160522460937500000000000000000e+00
1 19 -2.7636018149649316910654306411743164062500000000000000000000000000e+01
1 20 -3.1328868745033830123247753363102674484252929687500000000000000000e+01
1 21 -6.4424171952960008979971462395042181015014648437500000000000000000e+01
1 22 1.3795308458256652528461927431635558605194091796875000000000000000e+01
2 1 0.0000000000000000000000000000000000000000000000000000000000000000e+00
2 2 4.0467060940744698882554075680673122406005859375000000000000000000e+02
2 3 -2.4666340150362984218190831597894430160522460937500000000000000000e+00
2 4 -7.6741879471417632885277271270751953125000000000000000000000000000e+01
2 19 3.1328868745033830123247753363102674484252929687500000000000000000e+01
2 20 -2.7636018149649316910654306411743164062500000000000000000000000000e+01
2 21 -1.3795308458256652528461927431635558605194091796875000000000000000e+01
2 22 -6.4424171952960008979971462395042181015014648437500000000000000000e+01
3 1 -7.6741879471417632885277271270751953125000000000000000000000000000e+01
3 2 -2.4666340150362984218190831597894430160522460937500000000000000000e+00
3 3 1.4172374091814344865269958972930908203125000000000000000000000000e+02
3 4 0.0000000000000000000000000000000000000000000000000000000000000000e+00
3 5 -6.9238506120225792983546853065490722656250000000000000000000000000e+01
3 6 4.2449276607203678679525182815268635749816894531250000000000000000e+01
3 21 -2.2774146490635754247477962053380906581878662109375000000000000000e+01
3 22 -4.6314247061511501613040309166535735130310058593750000000000000000e+01
4 1 2.4666340150362984218190831597894430160522460937500000000000000000e+00
4 2 -7.6741879471417632885277271270751953125000000000000000000000000000e+01
4 3 0.0000000000000000000000000000000000000000000000000000000000000000e+00
4 4 1.4172374091814344865269958972930908203125000000000000000000000000e+02
4 5 -4.2449276607203678679525182815268635749816894531250000000000000000e+01
4 6 -6.9238506120225792983546853065490722656250000000000000000000000000e+01
4 21 4.6314247061511501613040309166535735130310058593750000000000000000e+01
4 22 -2.2774146490635754247477962053380906581878662109375000000000000000e+01
5 3 -6.9238506120225792983546853065490722656250000000000000000000000000e+01
5 4 -4.2449276607203678679525182815268635749816894531250000000000000000e+01
5 5 1.7166830055246467168217350263148546218872070312500000000000000000e+02
5 6 0.0000000000000000000000000000000000000000000000000000000000000000e+00
5 7 -4.0310680181028146762400865554809570312500000000000000000000000000e+01
5 8 3.0654314605490046830027495161630213260650634765625000000000000000e+01
6 3 4.2449276607203678679525182815268635749816894531250000000000000000e+01
6 4 -6.9238506120225792983546853065490722656250000000000000000000000000e+01
6 5 0.0000000000000000000000000000000000000000000000000000000000000000e+00
6 6 1.7166830055246467168217350263148546218872070312500000000000000000e+02
6 7 -3.0654314605490046830027495161630213260650634765625000000000000000e+01
6 8 -4.0310680181028146762400865554809570312500000000000000000000000000e+01
7 5 -4.0310680181028146762400865554809570312500000000000000000000000000e+01
7 6 -3.0654314605490046830027495161630213260650634765625000000000000000e+01
7 7 9.0122734518631659739185124635696411132812500000000000000000000000e+01
7 8 0.0000000000000000000000000000000000000000000000000000000000000000e+00
7 9 -2.6587549050434976294354783021844923496246337890625000000000000000e+01
7 10 3.2555664462677981418892159126698970794677734375000000000000000000e+01
8 5 3.0654314605490046830027495161630213260650634765625000000000000000e+01
8 6 -4.0310680181028146762400865554809570312500000000000000000000000000e+01
8 7 0.0000000000000000000000000000000000000000000000000000000000000000e+00
8 8 9.0122734518631659739185124635696411132812500000000000000000000000e+01
8 9 -3.2555664462677981418892159126698970794677734375000000000000000000e+01
8 10 -2.6587549050434976294354783021844923496246337890625000000000000000e+01
9 7 -2.6587549050434976294354783021844923496246337890625000000000000000e+01
9 8 -3.2555664462677981418892159126698970794677734375000000000000000000e+01
9 9 1.3994290817955334205180406570434570312500000000000000000000000000e+02
9 10 0.0000000000000000000000000000000000000000000000000000000000000000e+00
9 11 -3.8510392573554767636778706219047307968139648437500000000000000000e+01
9 12 4.3526959458391118573672429192811250686645507812500000000000000000e+01
10 7 3.2555664462677981418892159126698970794677734375000000000000000000e+01
10 8 -2.6587549050434976294354783021844923496246337890625000000000000000e+01
10 9 0.0000000000000000000000000000000000000000000000000000000000000000e+00
10 10 1.3994290817955334205180406570434570312500000000000000000000000000e+02
10 11 -4.3526959458391118573672429192811250686645507812500000000000000000e+01
10 12 -3.8510392573554767636778706219047307968139648437500000000000000000e+01
11 9 -3.8510392573554767636778706219047307968139648437500000000000000000e+01
11 10 -4.3526959458391118573672429192811250686645507812500000000000000000e+01
11 11 1.5436567823478708305628970265388488769531250000000000000000000000e+02
11 12 0.0000000000000000000000000000000000000000000000000000000000000000e+00
11 13 2.3785945574469224084168672561645507812500000000000000000000000000e+01
11 14 5.4928401268043032246168877463787794113159179687500000000000000000e+01
12 9 4.3526959458391118573672429192811250686645507812500000000000000000e+01
12 10 -3.8510392573554767636778706219047307968139648437500000000000000000e+01
12 11 0.0000000000000000000000000000000000000000000000000000000000000000e+00
12 12 1.5436567823478705463458027224987745285034179687500000000000000000e+02
12 13 -5.4928401268043032246168877463787794113159179687500000000000000000e+01
12 14 2.3785945574469224084168672561645507812500000000000000000000000000e+01
13 11 2.3785945574469224084168672561645507812500000000000000000000000000e+01
13 12 -5.4928401268043032246168877463787794113159179687500000000000000000e+01
13 13 3.8771747634530044024359085597097873687744140625000000000000000000e+02
13 14 0.0000000000000000000000000000000000000000000000000000000000000000e+00
13 15 -9.1945043368490601665143913123756647109985351562500000000000000000e+01
13 16 1.4743732717316754587955074384808540344238281250000000000000000000e+02
14 11 5.4928401268043032246168877463787794113159179687500000000000000000e+01
14 12 2.3785945574469224084168672561645507812500000000000000000000000000e+01
14 13 0.0000000000000000000000000000000000000000000000000000000000000000e+00
14 14 3.8771747634530044024359085597097873687744140625000000000000000000e+02
14 15 -1.4743732717316754587955074384808540344238281250000000000000000000e+02
14 16 -9.1945043368490601665143913123756647109985351562500000000000000000e+01
15 13 -9.1945043368490601665143913123756647109985351562500000000000000000e+01
15 14 -1.4743732717316754587955074384808540344238281250000000000000000000e+02
15 15 2.0126110204916042789591301698237657546997070312500000000000000000e+02
15 16 0.0000000000000000000000000000000000000000000000000000000000000000e+00
15 17 -5.2165597644569061230868101119995117187500000000000000000000000000e+01
15 18 1.3673105006512798809126252308487892150878906250000000000000000000e+02
16 13 1.4743732717316754587955074384808540344238281250000000000000000000e+02
16 14 -9.1945043368490601665143913123756647109985351562500000000000000000e+01
16 15 0.0000000000000000000000000000000000000000000000000000000000000000e+00
16 16 2.0126110204916042789591301698237657546997070312500000000000000000e+02
16 17 -1.3673105006512798809126252308487892150878906250000000000000000000e+02
16 18 -5.2165597644569061230868101119995117187500000000000000000000000000e+01
17 15 -5.2165597644569061230868101119995117187500000000000000000000000000e+01
17 16 -1.3673105006512798809126252308487892150878906250000000000000000000e+02
17 17 3.3725618489274347666651010513305664062500000000000000000000000000e+02
17 18 0.0000000000000000000000000000000000000000000000000000000000000000e+00
17 19 -5.9122571500287435242171341087669134140014648437500000000000000000e+01
17 20 5.7057835760936420399502821965143084526062011718750000000000000000e+01
18 15 1.3673105006512798809126252308487892150878906250000000000000000000e+02
18 16 -5.2165597644569061230868101119995117187500000000000000000000000000e+01
18 17 0.0000000000000000000000000000000000000000000000000000000000000000e+00
18 18 3.3725618489274347666651010513305664062500000000000000000000000000e+02
18 19 -5.7057835760936420399502821965143084526062011718750000000000000000e+01
18 20 -5.9122571500287435242171341087669134140014648437500000000000000000e+01
19 1 -2.7636018149649316910654306411743164062500000000000000000000000000e+01
19 2 3.1328868745033830123247753363102674484252929687500000000000000000e+01
19 17 -5.9122571500287435242171341087669134140014648437500000000000000000e+01
19 18 -5.7057835760936420399502821965143084526062011718750000000000000000e+01
19 19 1.9264168421482827397994697093963623046875000000000000000000000000e+02
19 20 0.0000000000000000000000000000000000000000000000000000000000000000e+00
19 21 -5.0503946656313928542658686637878417968750000000000000000000000000e+01
19 22 3.2518938603254845531864702934399247169494628906250000000000000000e+01
20 1 -3.1328868745033830123247753363102674484252929687500000000000000000e+01
20 2 -2.7636018149649316910654306411743164062500000000000000000000000000e+01
20 17 5.7057835760936420399502821965143084526062011718750000000000000000e+01
20 18 -5.9122571500287435242171341087669134140014648437500000000000000000e+01
20 19 0.0000000000000000000000000000000000000000000000000000000000000000e+00
20 20 1.9264168421482827397994697093963623046875000000000000000000000000e+02
20 21 -3.2518938603254845531864702934399247169494628906250000000000000000e+01
20 22 -5.0503946656313928542658686637878417968750000000000000000000000000e+01
21 1 -6.4424171952960008979971462395042181015014648437500000000000000000e+01
21 2 -1.3795308458256652528461927431635558605194091796875000000000000000e+01
21 3 -2.2774146490635754247477962053380906581878662109375000000000000000e+01
21 4 4.6314247061511501613040309166535735130310058593750000000000000000e+01
21 19 -5.0503946656313928542658686637878417968750000000000000000000000000e+01
21 20 -3.2518938603254845531864702934399247169494628906250000000000000000e+01
21 21 1.3770226509990968111196707468479871749877929687500000000000000000e+02
21 22 0.0000000000000000000000000000000000000000000000000000000000000000e+00
22 1 1.3795308458256652528461927431635558605194091796875000000000000000e+01
22 2 -6.4424171952960008979971462395042181015014648437500000000000000000e+01
22 3 -4.6314247061511501613040309166535735130310058593750000000000000000e+01
22 4 -2.2774146490635754247477962053380906581878662109375000000000000000e+01
22 19 3.2518938603254845531864702934399247169494628906250000000000000000e+01
22 20 -5.0503946656313928542658686637878417968750000000000000000000000000e+01
22 21 0.0000000000000000000000000000000000000000000000000000000000000000e+00
22 22 1.3770226509990968111196707468479871749877929687500000000000000000e+02


I cannot reproduce, here is the code I used:

Code: Select all


#include <Eigen/Sparse>
#include <unsupported/Eigen/SparseExtra>
using namespace Eigen;

int main()
{
  SparseMatrix<double> M;
  loadMarket(M, "cgsegfaul.mtx");
  VectorXd x(M.cols()), b(M.cols());
  b.setRandom();

  Eigen::ConjugateGradient< Eigen::SparseMatrix<double> > cg;
  cg.compute(M);
  x = cg.solve(b);

  return 0;
}


mmm, weird.. I do basically the same thing, except my b-vector entries are mostly zeros.

One thing I could think of is that I do something wrong here, which results in an invalid matrix or so:

Code: Select all

// Create iA and iB
// Fill iA and iB
// Need to solve A'Ax = A'b
iB = Eigen::SparseMatrix<double, Eigen::RowMajor>(iA.transpose()) * iB;
iA = Eigen::SparseMatrix<double, Eigen::RowMajor>(iA.transpose()) * iA;
iA.finalize();


Maybe if it would help somehow, I'm on a mac compiling using llvm.

Are there still any changes in the cg code since the 3.1.0-alpha1 release that could maybe explain the difference?

Thanks for your help

The default compiler on mac is known to sometime produce invalid piece of codes, you should give a try to gcc 4.6 of macport. Note that you can simply write:

iB = iA.transpose() * iB;
iA = iA.transpose() * iA;

Thanksm I'll give gcc a try. Do you plan to fix issues with llvm in the future?

the issues I'm talking about are on apple's side