svd help

svd help

Thu Nov 05, 2009 12:29 pm

I get the following error if i use a large matrix

#include </home/syed/eigen2/Eigen/Core>
#include </home/syed/eigen2/Eigen/Array>
#include </home/syed/eigen2/Eigen/LU>
using namespace Eigen;
int main () {

MatrixXf A = MatrixXf::Random(150000,4000);
VectorXf b = VectorXf::Random(150000);
VectorXf x;
A.lu().solve(b, &x);

}


terminate called after throwing an instance of 'std::bad_alloc'
what(): std::bad_alloc
Aborted (core dumped)

Beside, for a matrix of size 15000 x 4000, it kept running for more than 1 hr and i killed it
How long does it take to run SVD for larger matrices. Well, i am using a regular old PC P4 2.8 GhZ

I get the following error if i use a large matrix

#include </home/syed/eigen2/Eigen/Core>
#include </home/syed/eigen2/Eigen/Array>
#include </home/syed/eigen2/Eigen/LU>
using namespace Eigen;
int main () {

MatrixXf A = MatrixXf::Random(150000,4000);
VectorXf b = VectorXf::Random(150000);
VectorXf x;
A.lu().solve(b, &x);

}


terminate called after throwing an instance of 'std::bad_alloc'
what(): std::bad_alloc
Aborted (core dumped)

Beside, for a matrix of size 15000 x 4000, it kept running for more than 1 hr and i killed it
How long does it take to run SVD for larger matrices. Well, i am using a regular old PC P4 2.8 GhZ

Re: svd help

Thu Nov 05, 2009 5:46 pm

Ummh...

150000 x 4000 = 6x10^8 Doubles = 4.47 GByte.
This will take a while....

Ummh...

150000 x 4000 = 6x10^8 Doubles = 4.47 GByte.
This will take a while.... 8)

[bjacob]

Re: svd help

Thu Nov 05, 2009 6:49 pm

yep, for matrices of this size, the only way it can terminate in reasonable time is if that's sparse matrices, and we don't do sparse svd, so, you need to look for another library! Sorry.

yep, for matrices of this size, the only way it can terminate in reasonable time is if that's sparse matrices, and we don't do sparse svd, so, you need to look for another library! Sorry.