[molpro-user] Disk space in EOM-CCSD

[Tatiana Korona]

Dear Seth,

With the "direct" option we do not store AO 2-electron integrals and 
(CCSD+EOM-CCSD) do not calculate and store transformed 3-external 
2-electron integrals.

However, amplitudes and some intermediates are still stored on disk. The 
stuff stored on disk is factor* o^2 v^2, which in your case can be several 
GB already (see below). "factor" for EOM-CCSD is about 10 (if I haven't 
forgotten sth:-)

Let's estimate the number of double amplitudes you have:

No. of occ.orb.=126/2=63
No. of pairs 63*(63+1)/2=2016
No. of virt.orb.=250

No. of doub.ampl.=2016*250^2=126*10^6 => 1GB for one double vector

For the EOM-CCSD case we need to save several such vectors per state for 
the Davidson procedure (default per state is 20). This default can be 
changed by typing a command

eompar,maxdav=5 (e.g.)

The pairs C and H*C are saved, so if you have maxdav=5 and want to find 3 
states, you need 5*3*2*dimension_of_one_double_vector double words, in the 
example above this makes already 30GB (plus space for intermediates). Note 
that too small maxdav can result in a slow convergence, or even in a 
convergence to a wrong state in rare cases.

Best wishes,

Tatiana

Dr. Tatiana Korona http://tiger.chem.uw.edu.pl/staff/tania/index.html
Quantum Chemistry Laboratory
University of Warsaw
Pasteura 1, PL-02-093 Warsaw, POLAND

`The man who makes no mistakes does not usually make anything.'
                                        Edward John Phelps (1822-1900)

On Wed, 19 Apr 2006, Dr. Seth Olsen wrote:

>
> Hi Molpro-Users,
>
> I have run into a problem running a direct EOM-CCSD calculation.  I have set 
> 'gdirect', but the calculation is still exceeding my disk space limit of 
> 40gb.  The system in question is large (>250 contracted functions and 126 
> electrons), but if the calculation is running in direct mode, I would not 
> have thought >40gb of disk were necessary.  Does EOM-CCSD (or even CCSD) run 
> in direct mode in molpro?
>
> Cheers,
>
> Seth
>
> -- 
> ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms
>
> Dr Seth Olsen, PhD
> Postdoctoral Fellow, Biomolecular Modeling Group
> Centre for Computational Molecular Science
> Chemistry Building,
> The University of Queensland
> Qld 4072, Brisbane, Australia
>
> tel (617) 33653732
> fax (617) 33654623
> email: s.olsen1 at uq.edu.au
> Web: www.ccms.uq.edu.au 
> ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms
>