Multiwfn forum

giovannabruno

Member

Registered: 2021-10-29

Posts: 3

Topological analysis on .cube files

Good morning,
I have a problem in running CPs search by Multiwfn 3.4.1 on density .cube files generated by ORCA.  Regardless the option i choose for the search of CPs (starting from nuclei positions or from a given staring point o defining a sphere ecc) the program doesn't find anything because everytime it finds points on which the hessian is singular. This seems strange so my question is if Multiwfn can operate this kind of analysis on .cube files instead of .wfn files. I guess it's possible as when i run the serach by using .wfn files the program also generates .cube files of densities from them. Could you please help me?
Thank you very much,
Giovanna Bruno

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,830

Website

Re: Topological analysis on .cube files

There is no any reason to perform topology analysis of electron density based on .cub file. For ORCA user, you can use either .molden or .wfn file as input file for this kind of analysis, they can be generated by orca_2mkl and orca_2aim, respectively.

It is not impossible to perform topology analysis based on .cub file, since Multiwfn can calculate all needed information (function value, gradient and Hessian) by interpolation based on loaded grid data, however this is not recommended because the accuracy must be evidently poorer than directly calculating them based on wavefunction file.

Best regards,

Tian

giovannabruno

Member

Registered: 2021-10-29

Posts: 3

Re: Topological analysis on .cube files

Good morning,
The codes orca_2aim and orca_2mkl work only on .gbw, .qro, .uno ecc files which contain orbitals information at the SCF level. However, I need to analyse post-HF densities, more precisely CC densities. The latter are stored into .mdcip (electron density) and .mdcir (spin density) which can be easily transformed into .cube files by orca_plot, that's why i need to perform topological analyses directly on .cube files. There's no way in ORCA to obtain a .wfn file at the post-HF level, at least not at canonical o local CCSD(T) level. Besides, I tried to perform the topological analyses on a simple UHF wavefunction generated by Gaussian and i found some critical points already using the low resolution option for the grid; then, I generated from the same .wfn file the electron density .cube file using the same level of resolution for the grid in Multiwfn. When I tried to perform the topological analyses on this .cube file I couldn't find any CP. I can't understand the problem.
Thank you for your help,
All the best
Giovanna

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,830

Website

Re: Topological analysis on .cube files

Hello,

Using Multiwfn to analyze CCSD wavefunction produced by ORCA is feasible and very easy.

After performing CCSD calculation by ORCA, you will have .mdci.nat file in current folder, which contains natural orbitals of CCSD density matrix. You just need to change its suffix to .gbw, then use orca_2mkl to convert it to .molden file, the resulting .molden file will contain CCSD natural orbitals. If you use it as input file of Multiwfn, all analyses will be conducted for CCSD wavefunction.

ORCA is unable to produce CCSD(T) density, but PSI4 can. See Section 4.A.8 of Multiwfn manual on how to perform wavefunction analysis at CCSD(T) quality in combination with PSI4.

Please do not confuse topology analysis with basin analysis. The former is realized by main function 2 and does not directly rely on grid; the latter is realized by main function 17 and fully performed based on grid data, it is able to (crudely) locate position of attractors (i.e. (3,-3) type of critical points) but unable to locate other types of critical points such as (3,-1) ones.

Best regards,

Tian

giovannabruno

Member

Registered: 2021-10-29

Posts: 3

Re: Topological analysis on .cube files

Dear Tian,
thank you for your suggestion, I can definitely use .mdci.nat file for this analyses, I didn't think about that. I also know that ORCA does not produce the fully relaxed density, however one can still use the OOCCD method and obtaining a good approximation of that, which i'm interested to test. However, I'll follow your advice and I'll check the performances of PSI4 which i didn't know. Thank you again!
All the best,
Giovanna