[molpro-user] compile molpro for more than 20 state MRCI calculation

[Andy May]

Can you please try with the latest tarball (currently 2012.1.12) and the 
first set of tools (ifort,icc,mkl). If this still fails, please send the 
CONFIG file and full output file and I'll take a look.

Best wishes,

Andy

On 26/03/14 19:18, 김영록 wrote:
> Dear MOLPRO users.
>
>
> I'm trying to calculate over 20 state MRCI calculation.
>
> Official distribution of pre-built binary support states up to 20, so I
> tried compile myself with following condition
>
> Ubuntu 12.04
> Intel Composer XE 2013 sp1 ( ifort / icc / mkl )
> ./configure -auto-ga-mpich -mpp -i8 -ifort -icc
> molpro.2012.1.8.tar.gz
>
> After complete compile without any warning/error, It seems all right.
>
> But when I do some MRCI calculation, which was successfully done by
> pre-built binary version of MOLPRO, after several CI iteration, this
> kind of error pops out
>
>   Internal expansion vectors linearly dependent.
>   Smallest eigenvalue of S: -0.4910381E-15. Vector removed
>   Catastrophic failure in diagonalization(hsdel)
>   The expansion set has become singular
>
> So I tried compile with nearly all kind of combination of these
>
> ( gfortran / gcc
>   ifort / icc
>   intel-mkl / blas library from apt-get / ./configure -noblas
> on ./configure set maximum state per state 20/25 )
>
> but error continued.
>
>
> Could somebody post CONFIG file generated by ./configure for single
> multi-core workstation?
> Maybe pre-built binary's CONFIG file will be very helpful, I guess.
>
>
>
> Sincerely, Yeongrok Gim
>
>
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user
>