[molpro-user] [Molpro-user] Segmentation fault with FCI
l.lodi at ucl.ac.uk
Thu Oct 2 13:30:57 BST 2008
Peter Knowles wrote:
> This calculation will never run on your machine - it is too big. The
> size of the full CI is more than 4.1 * 10^9 and the program has to
> allocate two vectors of this length plus some other stuff, so it would
> not even run in 64 Gbyte. Even if you had the memory, it would take an
> extremely long time to run.
Thank you to both Peter and Kirk.
I had run a frozen-core FCI with cc-pVDZ on the same machine (in 7.5h)
and I wondered if the augmented basis set was still within reach, but
the increase from 24 to 41 contractions is just to great, as you pointed
out; it's a pity it took molpro more than 6 hours to realise the
calculation could not be done.
To estimate the RAM requirement of a FCI calculation with molpro is it
preferable to use the Weil formula (which refers to the number of
spin-adapted configurations) or just binomial(2k,n), because Slater
determinants are used? And how does symmetry reduce the memory requirements?
Incidentally, a Cs cc-pVDZ frozen-core runs in 2.5h using PSI3, instead
of 7.5h using molpro. As far as you can tell is this difference mainly
due to the different algorithm used by the programs or is it more a
matter of implementation/programming choices?
More information about the Molpro-user