Multiwfn forum

michaelp

Member

Registered: 2018-02-06

Posts: 21

NCP localisation for actinide compounds

Hello,

I have been using MultiWFN for teaching and research and I am very happy with the program. I recently stumbled upon a problem though. I have been calculating simple uranyl complexes with orca (all electron ZORA or DKH, trying also different DFT functionals) and in each case MultiWFN could not find the NCP of the uranium atom. I used a molden file generated by orca which I translated into a .wfn file by means of the molden2aim code. I also tried using the latest version of MutliWFN with a molden file generated by orca directly, no luck there either. I Would appreciate help in the matter.

Best wishes,
Michael

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,830

Website

Re: NCP localisation for actinide compounds

Dear Michaelp,

In topology analysis module, enter "-1 Set CP searching parameters", select below two options and then set corresponding convergence threshold more loose,
3 Criteria for gradient-norm convergence: 1.00000E-06
4 Criteria for displacement convergence: 1.00000E-07
Then the NCP should be able to be located when you use "2 Search CPs from nuclear positions".

The reason of difficulty of locating these NCPs is that for very heavy atoms, there is a extremely sharp peak at nuclear position, so it is not easy to meet default convergence criteria during Newton iterations due to numerical aspects.

Best wishes,

Tian

michaelp

Member

Registered: 2018-02-06

Posts: 21

Re: NCP localisation for actinide compounds

Dear Tian,

thank you for your fast reply! Loosening the convergence criteria did the trick. I tried to tighten the convergence criteria - that was the wrong way to go obviously. It finally worked with gradient norm convergence set to 1.0e-04 and displacement convergence at 1.0e-05.

Best wishes,
Michael