Integer too large to be represented in 4 bytes

[Yuan Li]

Dear Molpro users,

I compiled the molpro version 2021.1 from source code. I have one calculation and it was crashed in computing the spin-orbit coupling matrix. The error message is as follows: 

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 Symmetry of spin-orbit operator:         7

 Symmetry of bra wavefunction:            4   S= 1.0  MS= 0.0

 Symmetry of ket wavefunction:            6   S= 1.0  MS= 1.0

 ? Error

 ? integer (dscal:n) too large to be represented in 4 bytes

 ? The problem occurs in lapack_int

 GLOBAL ERROR fehler on processor   1

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Then I was reconfiguring the molpro with option -with-lapack-int64 , but I got configure error:

LAPACK with 8-byte integers no longer supported

But I did try with molpro 2020, the configure option -with-lapack-int64 works. I would like to ask, is there anyways to configure molpro 2021.1 with LAPACK 8-byte integers? 

Thanks for any suggestions,

Yuan

[Andy May]

Dear Yuan,

This problem has been fixed already, please use the latest source from the repository (2021.2.1) which contains the fix that batches large daxpy calls.

Best wishes,

Andy

--
You received this message because you are subscribed to the Google Groups "molpro-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to molpro-user...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/molpro-user/b34835cc-2c5a-4892-a13e-dfc47dc04475n%40googlegroups.com.

[iri...@verizon.net]

Dear Molpro experts,

I got the same error using version 2022.3 (pre-compiled binary).  Do I need to build from source to avoid the error? 

 Bra wavefunction restored from record  5101.2 to file 7

 Bra wavefunction restored from record  5103.2 to file 7

 ? Error

 ? integer (dscal:n) too large to be represented in 4 bytes

 ? The problem occurs in lapack_int

Thanks for any tips,

Karl

[Karl Irikura]

Hi again,

Same error with Molpro version 2024.1.  This is an atomic (Pb, basis=aug-cc-pwCVTZ-PP) spin-orbit CI calculation that failed after 15 days. 

Regards,

Karl

[Andy May]

Dear Karl,

I've made another fix for this and pushed out some new binaries for Linux (2024.1.1), hopefully this fixes the problem. The only difference compared to 2024.1.0 is the fix to address this specific error, so users do not need to update unless they encounter this issue.

Best wishes,

Andy

To view this discussion on the web, visit https://groups.google.com/d/msgid/molpro-user/f9cee04a-7539-49c4-b6f0-e2a9e4c45ffan%40googlegroups.com.