MP2 energies of large systems

[cediev87]

Hello,

I'm trying to use the Ricc2 code to compute MP2 single point energies of a large system (184 atoms). My problem is that I continue to get an error message saying I ran out of memory:

need least         6133  Mb

 ========================
  internal module stack:
 ------------------------
    ricc2
    riccints
    ccmkbaoi
 ========================

 out of memory in ccmkbaoi. (0)
 ricc2 ended abnormally

I started with $maxcor 2000 and continued to increase it until I set it to a really high value of $maxcor 50000 and continued to get the same error. I'm using the def2-TZVP basis set

Do you have any advice? Part of me thinks these systems are just too big to obtain an mp2 energy.
                                 

[cediev87]

Oh I also ran these calculations with 8 nodes, each node containing 8 cores so there should be sufficient memory allocated for the calculation.

Thanks!
C

[uwe]

Hello,

in the parallel MPI version the $maxcor settings are per process, so perhaps your $maxcor value was not too small but too large. Try a smaller value.

Alternatively try the new 'huge' binary of the 6.4 version which uses 64bit integers, so integer overflows are hopefully avoided.

Regards,

Uwe