[CP2K-user] [CP2K:13333] Pseudopotential Optimization for Minniesota Functionals

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

Hi

you can find input files for the GTH PP optimization for almost
the full periodic table at

https://github.com/juerghutter/GTH

These inputs are for the standard form (meaning number of Gaussians
and number of projectors) used in most applications.
You can just change the functional in the input and rerun.
However, except for first row atoms convergence will not be easy.
I usually run multiple restarts until the overall error is
small enough. You can see what final errors have been achieved
in the summary files for each functional.

When restarting you can play with the following input parameters

     STEP_SIZE  0.0002
     STEP_SIZE_SCALING  0.95
     MAX_INIT   25
     MAX_FUN    500

You need to build up some experience what works. There is no obvious rule
how to proceed.

This input will do 25 restarts of Powell optimizations, each with a maximum of
500 function evaluations. The original search radius is 0.0002 and at each
restart the radius is multiplied by 0.95.

Start your search with the parameters from a similar functional (GGA or hybrid ...).

Good luck

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: "cp2k" <cp... at googlegroups.com>
From: "jt Yang" 
Sent by: cp... at googlegroups.com
Date: 05/18/2020 07:02PM
Subject: [CP2K:13333] Pseudopotential Optimization for Minniesota Functionals

Hi CP2K community,

I want to use M06 functional in cp2k, and encounter with PP Opt problem like others, 

1. I checked several posts, there are many parameters one can set and little explanation, so I'm not sure if the input file is proper for my aim.

1-1. if I change Ti to H, C, N, O, is there some parameters special should I change?

1-2. some parameters: "WEIGHT*" "MAX_ANGULAR_MOMENTUM", how they effect the result?

1-3. I may use other Minniesota functionals in future, so please let me know if something about PP Opt should change in M06-2X or M06-L.

2. Some posts say testing the result is important, then what kind of test should I do? How do I know the PP is good or bad? 

2-1. If the result is bad, what should I do, which parameters should reset?

here is my input file, just change the functional of Hutter's version.

Best wishes,
jtYang  
  -- 
 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/edd6774f-bad8-4aed-98de-6f89822e4af0%40googlegroups.com.
 

[attachment "jtyangTi.inp" removed by Jürg Hutter/at/UZH]