Hi Matt,

I put a write statement before and after the call to dsyev in src/util/util.f and lo and behold, info is equal to..

before dsyev 6917529027701845120
after dsyev  6917529027641081856

looks like its pointing to a random bit of memory, or its begin trodden on somewhere by something else.

James.

--On 05 August 2003 16:43 +0100 Matt Hodges <[EMAIL PROTECTED]> wrote:

James A J Fitzpatrick writes:

> I do however have a problem with linking in the intel MKL libs for > the blas and lapack routines. MOLPRO links fine but the calculation > fails when trying to calculate the 2-electron integrals with a > "failure in dsyev" error message. I have looked in the source code > and there is a note mentioning that some vendor lapack libs fail in > dsyev, and I assume that is what is happening here.

This fix was specifically for the NEC version of DSYEV, but should not
affect other implementations adversely. Can you find out what value of
INFO is returned when it fails?

Matt




--- James A.J. Fitzpatrick Laser Chemistry and Dynamics Group School of Chemistry, Cantock's Close, Bristol BS8 1TS, UK. Telephone (Office) : +44 117 928 9938, Telephone (Lab) : +44 117 928 8187 Fax : +44 117 925 0612

E-mail : [EMAIL PROTECTED]
Web    : http://www.chm.bris.ac.uk/~chjajf

Reply via email to