Multiwfn forum

Fayzullin

Member

From: Russian Federation

Registered: 2022-07-04

Posts: 3

Coulomb correlation in Multiwfn and Gaussian

Could you please clarify something for me?

When I open an fchk file in Multiwfn, the program indicates that the wavefunction is single-determinant, even though the calculation was performed in Gaussian using the keywords
“#p uccsd(full)/aug-cc-pvqz density=current pop=noab geom=connectivity” or
“#p ccsd(full)/aug-cc-pvqz density=current pop=no geom=connectivity”.
Does this mean that the effects of Coulomb correlation are not taken into account (occupancy of n-orbitals is not integer, N-representability (for 1-RDM) is satisfied)?

Furthermore, does it make any difference for Multiwfn whether “pop=no” or “pop=full” is specified when calculating properties that should reflect Coulomb correlation?

Thank you very much in advance for your response.

sobereva

Tian Lu (Multiwfn developer)

From: Athens, Greece

Registered: 2017-09-11

Posts: 2,115

Website

Re: Coulomb correlation in Multiwfn and Gaussian

Hello,

First, please note that Gaussian is unable to generate density matrix at CCSD(T) level, but ORCA and PSI4 can, please check Section "4.A.8 Analyze wavefunction higher than CCSD level" of Multiwfn manual on how to perform wavefunction analysis based on CCSD(T) density. The most accurate density that Gaussian can generate is CCSD.

Even though you have specified density=current and pop=no or noab, Gaussian doesn't write post-HF natural orbitals into .chk file. I suggest using out=wfn or out=wfx keyword, in this case the generated .wfn/wfx file directly contains post-HF natural orbitals. While if you hope to write natural orbitals into .fch file, please check the text in "Generating .fch file" part at the beginning of Chapter 4 of Multiwfn manual.

Fayzullin

Member

From: Russian Federation

Registered: 2022-07-04

Posts: 3

Re: Coulomb correlation in Multiwfn and Gaussian

Thank you! Now it perfectly works!