Multiwfn forum

ffuser

Member

Registered: 2020-10-20

Posts: 9

Negative occupation number in fuzzy bond order analysis

Hi.

I have a question about fuzzy bond order (FBO) analysis.
The command I used is as follows: run Multiwfn a.wfn (excited state density), then input 9 (BOA), then input 7 (FBO).
Then I received a message “Note: Some occupation numbers are negative. “
In this case, what is the occupation number? Bond order?

This didn’t occur for ground state analysis, but excited one.
Thinking about the possibility of the density being wrong, let me describe how I obtained the excited state density.
I ran Gaussian job with option “td=(nstates=1,root=1) density=current our=wfn”
The wfn file I mentioned above was obtained this way.
Furthermore, by using fchk made from this job’s chk file, I obtained Molden file for relaxed excited state density (I ran Multiwfn a.fchk, input 200, input 16, input CI, and input y) and tried FBO analysis using this Molden file, but the same issue occurred.

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,830

Website

Re: Negative occupation number in fuzzy bond order analysis

For TD type of wavefunction, some natural orbitals may have marginally negative occupation numbers, although it is seemingly unphysical, it is not harmful in practice, you can safely ignore this note.

By the way, setting "nstates" to the same value as "root" is highly deprecated. Because Gaussian uses a method similar to Davidson iteration method to solve TDDFT equation, in this case, the highest several obtained excited states are not quite numerically accurate. Therefore, when you are interested in root=1, you should at least set the "nstates" to 3 (which is the default setting).

ffuser

Member

Registered: 2020-10-20

Posts: 9

Re: Negative occupation number in fuzzy bond order analysis

Thank you for replying so quickly.
Good to know it was not a problem.
Also the latter explanation helps me a lot as well.