[molpro-user] "Correct" orbital domains for DF-LMP2 torsion scan

Gregersen, Brent Brent.Gregersen at deshaw.com
Thu Jul 26 21:17:39 BST 2007


I have 3 related questions that I hoping someone could help with.

1) Is it possible to display the orbital domains during the course of a
geometry optimization (Or after they are frozen during a geometry
optimization)? They currently seem to be printed during the first
calculation only.

2) The documentation and example shown in section "32.9.4 Orbital
domains" of the user manual would seem to indicate that "correct"
orbital domains should correspond to bonds (one orbital for single
bonds, two orbitals for double bonds, three for triple) and one orbital
for each lone pair within a molecule. Using basis=aug-cc-pVTZ and the
procedure:
procedure runmme={
hf;accu=15,maxit=100
df-lmp2;local,thrbp=0.985
pipek,delete=2
}
This seems to hold for NH3 (4 orbitals total, 1 for each N-H bond and 1
for N lone pair), but not for HCN (5 orbitals total, 1 for C-H bond, but
4 for C-N bond, no N lone pair). Is this to be expected? If the orbital
domains should correspond to bond connectivity/lone pairs, would it be
better for the user to manually define these domains (assuming
connectivity doesn't change)?

3) If I am going to perform a relaxed 2-d torsion pes scan at 15 degree
intervals (576 geometry optimizations total) is there an easy way to
merge the domains from all 576 conformations to obtain a continuous
energy surface? The examples I've seen seem to require the domain
merging to occur all in one file, which would seem to make simultaneous
geometry optimizations of all 576 conformers difficult (especially if
the domains change during the course of the optimization).

Thanks for any insight,
Brent




More information about the Molpro-user mailing list