Top Page > Browsing
ASE-OPENMX for NEB calculation
Date: 2020/03/06 18:31
Name: Reza

H
As you may know, ASE could be linked with different calculators such as siesta, openmx, vasp, ...
I use ASE-NEB combining with siesta, openmx, and FHI-aims.
I also use these codes' neb without ASE.So, I am familiar with them.

ASE-NEB has climb imaging and different algorithms. Combining openmx with ASE is good,
but has some problem that some of these problems belong to the ASE codes and some others goes back
to openmx. I know that openmx developers try to develop linking their code. So, in addition to the
ASE forum, I comment theses problems here:

1- ASE 3.19.0 version automatically changes k-points to 444, and this is a bug that should be corrected.

2- openmx should use old density matrix for each image in neb to increase its speed of SCF convergence.
Because in each NEB iteration, only small changes in images structures happen, it is logical to save the
old density matrix and use it in converging new SCF.It seems siesta do this behavior.

3- Some printing output are unnecessary during NEB calculation. For example, printing different cube files
are only time-consuming and not needed. I think it goes back to ASE codes. I used openmx neb, but didn't see
these unnecessary cube files.

4- In openmex all eigenvalues and orbital decompositions are calculated in each image and every iteration.
Please have an option that user can prevent it.

Thanks
Reza

メンテ
Page: [1]

Re: ASE-OPENMX for NEB calculation ( No.1 )
Date: 2020/03/06 18:35
Name: Reza

Also, linking ASE with new version of DFT-DATA19 and version 3.9 has a bug. In openmx.py (I think) '13'
should be changed to '19' to use new DATA19. However, I could not use 3.9 because it says that
there is not property energy in calculations! This is ASE bug not for OpenMX.
メンテ
Re: ASE-OPENMX for NEB calculation ( No.2 )
Date: 2020/03/08 08:51
Name: T. Ozaki

Hi,

Thank you for your suggestion.

> 2- openmx should use old density matrix for each image in neb to increase its speed of SCF convergence.
> Because in each NEB iteration, only small changes in images structures happen, it is logical to save the
> old density matrix and use it in converging new SCF.It seems siesta do this behavior.

During the NEB calculation, OpenMX uses the restart files to accelerate the SCF convergence.

> 4- In openmex all eigenvalues and orbital decompositions are calculated in each image and every iteration.
> Please have an option that user can prevent it.

It might be difficult to skip the calculations of eigenvalues.

As for the ASE issues, I hope that the ASE team will fix them.


Regards,

TO
メンテ

Page: [1]

Thread Title (must) Move the thread to the top
Your Name (must)
E-Mail (must)
URL
Password (used in modification of the submitted text)
Comment (must)

   Save Cookie