Page 1 of 1

Too many chains - PT1SOR

Posted: 12 Apr 2016, 17:40
by sonco
I was running a rotational g of SO with MCSCF and a relatively small CAS and the aug-cc-pCV5Z basis set.
The calculation dies with this error:

Too many chains - PT1SOR
forrtl: severe (71): integer divide by zero

Does anyone know what this is about? (I have run all basis aug-cc-pVXZ and aug-cc-pCVXZ, X=T,Q without any problem)
I admit I am not familiar with the part of the code that produces the error message, so I am sure what's going on...


Re: Too many chains - PT1SOR

Posted: 12 Apr 2016, 18:57
by taylor
MAXCHN is being exceeded (the forrtl error just results from the QUIT routine --- it's the MAXCHN error message that's meaningful). This relates to determining the number of chains and the size of the buffers for the "backchained" sort for transforming the 2nd-order reduced density matrix. I think you can try increasing MAXCHN (defined in ptrbuf.h which is not in the include directory but in the abacus directory itself --- I don't remember why but we must have had some reason at some point...) and rebuild the code.

Best regards
P.S. There is a routine that prints a lot of the sort parameters: it's in abaptr.F but presumably needs some fiddling with print levels. But in any event if e.g. cc-pCVQZ works with the default I think if you redefine to 400 (so twice the default) cc-pCV5Z ought to work then.