Molcas Forum

abid

Member

Registered: 2015-11-12

Posts: 41

RASSI calculation problem

Dear all
I have two sets of rasscf calculations:
1. One RASSCF with one electron in RAS3
2. Other RASSCF with two electrons in RAS3
Active space is same for both calculations. Such calculations are easily running on the MOLCAS 7.8 but I was surprised when I run same set of calculations on MOLCAS 8.1
with following error.

   Specific data for JOBIPH file JOB001
   -------------------------------------
  IFEJOB= F

   Header from SEWARD:
                                     water-dimer
        Integrals generated by seward 4.2.0  , Sat Mar  4 11:21:08 2017

   CASSCF title (first line only):
       (No title given)

   STATE IRREP:                             1
   SPIN MULTIPLICITY:                       1
   ACTIVE ELECTRONS:                       12
   MAX RAS1 HOLES:                          0
   MAX RAS3 ELECTRONS:                      1
   NR OF CONFIG:                       165528

   Specific data for JOBIPH file JOB002
   -------------------------------------
  IFEJOB= F
  RAS SPECIFICATIONS DIFFER.
      THIS STATE: MAX NR OF RAS-1 HOLES:                     0
              MAX NR OF RAS-3 ELECTRONS:                     2
  PREVIOUS STATE: MAX NR OF RAS-1 HOLES:                     0
              MAX NR OF RAS-3 ELECTRONS:                     1
  DATA IN JOBIPH FILE NAMED JOB002   WERE
  INCONSISTENT WITH EARLIER DATA. PROGRAM STOPS.

  Errors occured in RASSI/RDJOB.
--- Stop Module:  rassi at Sat Mar  4 12:34:26 2017 /rc= _INTERNAL_ERROR_ ---

         Some internal inconsistency of the code was detected

Any help will be appreciated. Thanks all

Rafael

Member

Registered: 2016-11-16

Posts: 12

Re: RASSI calculation problem

I faced the same problem. Apparently, this is a bug in the 8.1 version of MOLCAS.
I used the same input file in the versions 8.0 and 8.1, and I got this error only in the 8.1.
I have already report this bug in MOLCAS website. 
My advice is, if you want to use a newer version of MOLCAS, try the 8.0, which works perfectly.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: RASSI calculation problem

It's not a bug, in the sense that the code in doing what it's supposed to do: check if the wavefunction definition is the same... and it isn't.

The question is whether the results given by 8.0 are to be trusted or are bogus, and that I don't know. If someone added this check, it must be for a reason.