[CP2K-user] [CP2K:13855] Mayer bond orders

[hut... at chem.uzh.ch]

Hi

I just wanted to add some information on this issue.

Mayer Bond Orders are not implemented in CP2K, at least not
the straightforward evaluation that is based on Mulliken
population analysis.

The MAO_ANALYSIS is a much more difficult procedure that is
reducing the basis set dependence in population analysis.
As part of this analysis, a Bond Order is calculated in the
MAO basis. You also get a shared electron number, which is an
improved measure for bonds. 
This as been tested for molecules with a large HOMO-LUMO
gap. It will not work for k-points and might have problems for
systems with small band gap and or if you use smearing.
For these systems more work is needed.

There are also other possibilities using external programs, e.g.
chargemol.

Coming back to the simple Mayer Bond Order. Implementation would be
easy but currently not at the top of my list.

best regards

Juerg Hutter

--------------------------------------------------------------
Juerg Hutter                         Phone : ++41 44 635 4491
Institut für Chemie C                FAX   : ++41 44 635 6838
Universität Zürich                   E-mail: hut... at chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp... at googlegroups.com wrote: -----
To: cp... at googlegroups.com
From: "Anton S. Lytvynenko" 
Sent by: cp... at googlegroups.com
Date: 09/10/2020 02:04PM
Subject: Re: [CP2K:13855] Mayer bond orders

                   Dear Miguel,
     Unfortunately, I have no idea how to fix your issues, I am       writing that just to confirm that I've encountered similar issues       when I once tried to calculate Mayer bond orders in CP2K. It looks       like Mayer bond order calculation in CP2K could be faulty or at       least documented so obscurely that it is hard to understand the       right way to perform it.
     Yours,
     Anton
     
31.08.20 09:05, Miguel Steiner пише:
                        Dear CP2K users,
       
       I want to use the Mayer bond order analysis available in CP2K and       first wanted to try it on the Si example that is included in the       tutorial (see input file attached).
       However, I ran into the problem that I cannot find the right       parameters for a successful analysis.
       In general the section looks like this
       
  !-----------------------------------------------------------------------------!
                                LOCALIZED MINIMAL BASIS ANALYSIS
                        W.C. Lu et al, J. Chem. Phys. 120, 2629 (2004)
  !-----------------------------------------------------------------------------!
        Total Number of Atomic Basis Set Functions         :                              104
        Total Number of Minimal Basis Set Functions        :                                  x
        Total Number of Molecular Orbitals available       :                                    y
       
       From my understanding and trials, I have to give the number of       MAOs and the number of virtual MOs in the input for a successful       analysis. I tried a few things and noticed that the MAOs per kind       manipulate number x in the output as expected and the added MOs       change number y in the output.
       I now noticed the following behavior:
       
       If I choose MAOs to be too low, I get an error with invalid number       of columns in a matrix, which makes sense if the minimum number       given is lower than the actual physical minimum number determined       through the number of electrons.
       
       If x < y and y <= 104:
       I get "WARNING: Only a subset of MOs is available: Analysis       depends on MOs"
       And then I get the error: "Hotelling inversion did not converge"
       
       if x > y and y < 104:
       I simply get "Localized Minimal Basis Analysis not possible" and       no warning and errors are raised.
       
       if x > 104 and y <= 104:
       CPASSERT failed in minbas.f:158
       
       
       Therefore, no matter which numbers I plug in, I cannot get a       successful calculations and it seems to me that I tried the whole       range of possible inputs.
       
       In general, I would need to do this analysis in a robust way       unsupervised for multiple systems (rather surfaces than crystals),       while also staying computationally feasible.
       So I would very much appreciate a rather safe combination of       inputs or a formula to calculate the necessary inputs for each       structure. 
       
       Best regards and thanks for your help,
       Miguel       -- 
       You received this message because you are subscribed to the Google       Groups "cp2k" group.
       To unsubscribe from this group and stop receiving emails from it,       send an email to cp... at googlegroups.com.
       To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/29e456a0-390a-47c5-ba66-6bdd498f80d0n%40googlegroups.com.
           
  -- 
 You received this message because you are subscribed to the Google Groups "cp2k" group.
 To unsubscribe from this group and stop receiving emails from it, send an email to cp... at googlegroups.com.
 To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/f421f220-a68b-a92c-5e62-182bb792216e%40gmail.com.