PC GAMESS/Firefly-related discussion club

Learn how to ask questions correctly

Alex Granovsky
gran@classic.chem.msu.su

as far as I understand you are limited to SP MP2 runs and
thus you cannot calculate MP2 hessian as it would be too expensive.

At present, MP2 hessians are seminumerical and hence the
documentation for MP2 gradient code is relevant.

>I will be running this MP2 job on a Mac (running through Darwine) with 8 processing cores but without a separate physical hard drive for each processing core.  Reading the documentation online for MP2 jobs:

> http://classic.chem.msu.su/gran/gamess/mp2grd.html

>I believe that the final note applies to my case, namely:

>The following set of parameters of the new MP2 gradient code seems to be optimal in the case you are running large problems in parallel on Linux-based SMP or multicore system which do not have a separate physical disks for each instance of the parallel PC GAMESS processes (as recommended above).

> $system splitf=.f. $end
> $system ioflgs(30)=0 $end
> $mp2grd dblbf=0 fused=0 asysnc=1 $end

Note the page above is for MP2 energy gradient code.
As to SP MP2 energy, MP2 energy code is different and
thus most of these options do not have any effect. More precisely,

 $system splitf=.f. $end
 $system ioflgs(30)=0 $end

is fine but the rest will be ignored. What you actually need is to select:

proper P2P options:

 $p2p p2p=.t. dlb=.t. xdlb=.t. $end

method=1 MP2 code

 $mp2 method=1 $end

Hondo integrals:
 
 $contrl inttyp=hondo $end

and sometimes to tighten cutoff in $mp2 group, e.g.: 

 $mp2 cutoff=1d-12 $end

Depending on the size of your system, it may be more efficient
to use only part of all available cpu cores while providing a
separate HDD to each instance of parallel Firefly process.

Regards,
Alex