Page 1 of 1

DGEEV

Posted: 12 Nov 2009, 11:24
by kimsongoik
Hello Elmer
What is “DGEEV” that is called in ‘Integration module’. I searched all Elmer’s source codes but I didn’t find any match for it. How does “DGEEV” work since it has not been defined anywhere in Elmer and more importantly how can I find a source code of it?
Thanks,
Kimsongoik

Re: DGEEV

Posted: 12 Nov 2009, 11:47
by Juha
Hi,

DGEEV is a LAPACK routine for computing eigenvalues & -vectors.
Source code is to be found from Netlib for example, but is also
distributed with Elmer source package in mathlibs/src/lapack/dgeev.f.

If this version is actually used by the ElmerSolver binary depends on
your configuration options for LAPACK libraries (it is the default,
however).

Juha

Integration Module

Posted: 12 Nov 2009, 16:14
by kimsongoik
Would anybody mind to tell me which of the followings is the result of subroutine ‘ComputeGaussPoints1D)’ for “n=2”?
1- Points=(0.577,0.577) and Weights=(1,-1)
Or
2- Points=(0.577,-0.577) and Weights=(1,1)

and what are integers 'thread' and 'omp_get_thread_num' in the 'Integration' module?
Thanks,

Re: DGEEV

Posted: 12 Nov 2009, 18:33
by Juha
Hi,

the latter is correct, sum of the weights should always
equal two.

The variables thread, etc... are for threading parallel
runs when OpenMP is used. The OpenMP support in
Elmer should be considered 'experimental' at best,
however.

Normally you should not worry 'bout those.

Juha