[molpro-user] Bad Localized Orbitals?

[Matthias Heger]

Hello everyone,

I've been doing a quick LMP2 calculation in Molpro 2015.1 on 
formaldehyde. My input is fairly straightforward, and attached to the 
end of this email. With NPASEL=0.07, the domain definitions in the 
output look like this:

    Orb.   Atom     Charge      Total
    3.1    2 C       1.11
           4 H       0.88        1.99
    4.1    2 C       1.11
           3 H       0.88        1.99
    5.1    1 O       1.31
           2 C       0.68        2.00
    6.1    1 O       1.36
           2 C       0.64        2.00
    7.1    1 O       1.91        1.91
    8.1    1 O       1.97        1.97

When visualizing the localized orbitals, #4.1 looks like I would expect 
a localized C-H bond orbital to turn out. However, something bad seems 
to be going on with orbital 3.1:
http://i.imgur.com/JJn1Kwl.png
I noticed something similar happening in pentane. The C-H bond orbitals 
on one end of the molecule looked fine, while those on the other end 
were quite messy.

Does anyone know what's going on here? How will this impact my 
correlation energies?

Best,
Matthias


***
basis=vtz
angstrom
geomtyp=xyz
orient,mass
symmetry,nosym
geometry={
     4
fa
  O          0.0000000000        0.0000000000        0.6034321216
  C          0.0000000000        0.0000000000       -0.6053608971
  H          0.0000000000        0.9357726712       -1.1823928760
  H          0.0000000000       -0.9357726712       -1.1823928760}
df-hf
{df-lmp2
local,npasel=0.07,savorb=3000.2}
{cube,Formaldehyde_loc
orbital,occ,record=3000.2,type=local}

-- 
Dr. Matthias Heger
Department of Chemistry, University of Alberta
11227 Saskatchewan Drive NW
Edmonton, AB T6G 2G2, Canada

Office: CCIS 4-151
Email: heger at ualberta.ca