Intel® oneAPI Math Kernel Library
Ask questions and share information with other developers who use Intel® Math Kernel Library.
6981 Discussions

Re: dsytrd & dsteqr - How to properly computing all eigenvalues and eigenvectors?

Sergey_K_Intel1
Employee
189 Views

I'd recommend to use dsyev routine for computing all eigenvalues and eigenvectors of symmetric matrices. Here is an C example

#include

"mkl_lapack.h"

int

main(){

intm=3;

doublea[9] = {294.0, -180.0, 108.0, -180.0, -25.0, 240.0, 108.0, 240.0, 231.0};

charjobz='v', uplo='l';

intinfo, n=3, lwork=18, i, j;

doublew[3], work[18];

dsyev(&jobz, &uplo, &n, a, &n, w, work, &lwork, &info);

printf(" Eigenvalues computed ");

for(i=0; i

printf("%.5f ", w);

};

}

The input/output arraya if jobz= 'V' andinfo = 0, contains the orthonormal eigenvectors of the matrix A. You are not able to use dsteqr for the matrix given by yousince dsteqr is only for tridiagonal matrices and accepts onlythe main and subdiagonals as input parameters. So you have to use the following lapack routines dsytrd, dorgtr and dsteqr for solving the full eigenvalue problemfor dense symmetric matrix as it isimplemented in dsyev routine. Besides dsyev routine is doing a scaling of input matrix if it is necessary to get thehigh accuracy.

0 Kudos
0 Replies
Reply