Multiwfn forum

lijingbai2009

Member

Registered: 2018-03-07

Posts: 2

Question about MO coefficients for f+3 and f-3 shells

Dear forum,
I've been checking the molecular orbitals loaded from molden file and got confused a little bit.

For example, my test job was a single Pt ion with +2 charges using pbe0/def2-svp in ORCA 4.0.0. The f shell coefficient in MO 29 were "-0.000014, 0.000002,-0.391068, -0.000025,-0.000011, -0.000016, 0.919986 ". When I used Multiwfn to print coefficient matrix, it showed "-0.000014, 0.000002,-0.391068, -0.000025,-0.000011, 0.000016, -0.919986 ". Just look at the last two numbers, they are from f+3 and f-3 shell. The first is the second multiplying -1.
After I saved to another molden, the MO coefficients for f+3 and f-3 in the new molden file turned back as "-0.000016, 0.919986".

Then I checked the transformation from spheric harmonic type GTO to cartesian ones. I found if I don't multiply the MO coefficient for f+3 and f-3 shell by -1, my results would be totally wrong. After multiplying by -1, the pure f shell can be correctly transfer to cartesian f shell.

So my question is, where is this -1 from?

Best regards,
Jingbai

Last edited by lijingbai2009 (2018-03-07 18:54:34)

I_was_a_baby

Member

Registered: 2017-09-11

Posts: 29

Re: Question about MO coefficients for f+3 and f-3 shells

Dear forum,
I've been checking the molecular orbitals loaded from molden file and got confused a little bit.

For example, my test job was a single Pt ion with +2 charges using pbe0/def2-svp in ORCA 4.0.0. The f shell coefficient in MO 29 were "-0.000014, 0.000002,-0.391068, -0.000025,-0.000011, -0.000016, 0.919986 ". When I used Multiwfn to print coefficient matrix, it showed "-0.000014, 0.000002,-0.391068, -0.000025,-0.000011, 0.000016, -0.919986 ". Just look at the last two numbers, they are from f+3 and f-3 shell. The first is the second multiplying -1.
After I saved to another molden, the MO coefficients for f+3 and f-3 in the new molden file turned back as "-0.000016, 0.919986".

Then I checked the transformation from spheric harmonic type GTO to cartesian ones. I found if I don't multiply the MO coefficient for f+3 and f-3 shell by -1, my results would be totally wrong. After multiplying by -1, the pure f shell can be correctly transfer to cartesian f shell.

So my question is, where is this -1 from?

Best regards,
Jingbai

To Jingbai:

I don't know where the "-1" come from, so the reasonable conjectures are following:

There is a serious trouble that the molden files generated by ORCA can be problematic when the higher type basis functions may have been obtained in current basis set. Another problem is that, in Multiwfn, the EDF information of all elements involving pseudopotential can be replaced by inner-core electron density. It is an automatically process when you DO NOT change the setting.ini file of Multiwfn. So, the issue of “-1” may be caused by the reasons mentioned above.

The suggestions I give to you are that, the first one, by using Multiwfn program, check sanity of wavefunction of your molden files. It can be obviously seen in Appendix 5 of Multiwfn manual to consult how you can do. If the result is negative, please using the Molden2aim to normalize your molden files. On the other hand, if you do not want Multiwfn read EDF information from this file but from the built-in EDF library, you can change "readEDF" in settings.ini from 1 to 0. This will be consulted in Appendix 4 of Multiwfn manual.

Best wishes
wawa

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,763

Website

Re: Question about MO coefficients for f+3 and f-3 shells

Dear lijingbai2009,

The reason of introducing the factor of -1 is that the convention used by ORCA is very weird. For F(+3), F(-3) and G(+3) G(-3) G(+4) G(-4), the basis functions are normalized to -1 rather than 1, therefore in order to make analysis result meaningful, the corresponding LCAO coefficient must be multiplied by -1. I don't know why ORCA uses such strange convention.

Best wishes,

Tian

lijingbai2009

Member

Registered: 2018-03-07

Posts: 2

Re: Question about MO coefficients for f+3 and f-3 shells

Dear lijingbai2009,

The reason of introducing the factor of -1 is that the convention used by ORCA is very weird. For F(+3), F(-3) and G(+3) G(-3) G(+4) G(-4), the basis functions are normalized to -1 rather than 1, therefore in order to make analysis result meaningful, the corresponding LCAO coefficient must be multiplied by -1. I don't know why ORCA uses such strange convention.

Best wishes,

Tian

Dear Tian Lu,

That's the exact reason I wanted to know. ORCA does have many tricks inside. Now everything are clear for me. Thanks a lot again.

Best regards,
Jingbai