[molpro-user] memory/disc space requirement for L-CCSD(T)

[Jörg Saßmannshausen]

Dear all,

having successfully tackled the problem with the not converging SCF, I am 
facing a new challenge.

In order to make sure that my molecule of 53 atoms is running smoothly on the 
computational node (Quad core Opteron, 12 GB ram, 4 cores), I decided to give 
the whole process arouhd 11 GB of RAM, so the OS has got some RAM as well. 
However, it was eating up 470 GB of scratch space which is basically all I 
could (somehow, and please don't ask how I done it) utilise for that 
calculation. Unfortunately, that was not enough. So, the question is, is 
there any way to get these figures (scratch mainly) down to something more 
managable?
From this line:
 Integral-direct triples not yet available
I understood that I cannot use direct SCF in oder to save disc space.

Enclosed are the relevant parts of the input file but the coordinates.

memory,350,m    !11 GB of total memory, 2.75 per core
SET,CHARGE=1    !charge of the molecule is +1
geomtyp=xyz
geometry
   53
CCSD(T)/cc-pVTZ//mp2/6-31g*
[ ... ]
end

hf,maxit=300
hf,direct
basis=cc-pVTZ           !cc-pVTZ basis set for all elements
hf                      !Perform HF calculation
lccsd(t)                !Perform L-CCSD(T) calculation

The organic molecule contains F, N, O and Si as heteroatoms.

Would somebody mind to point me in the right direction?

All the best from Glasgow!

Jörg 

-- 
*************************************************************
Jörg Saßmannshausen
Research Fellow
University of Strathclyde
Department of Pure and Applied Chemistry
295 Cathedral St.
Glasgow
G1 1XL

email: jorg.sassmannshausen at strath.ac.uk
web: http://sassy.formativ.net

Please avoid sending me Word or PowerPoint attachments.
See http://www.gnu.org/philosophy/no-word-attachments.html