Top Page > Browsing
Parameters for running large systems
Date: 2021/09/18 09:32
Name: MChen

Dear developer,

I am wondering if there is any parameter for the calculations of large systems. In VASP, one can use NPAR which switches from k-point
parallelization to band parallelization. What is the parameter using openmx? I am running electronic structure calculations for a system of about 1000 atoms. Any suggestion to accelerate the calculation? Thank you very much.

Best regards,

MChen
メンテ
Page: [1]

Re: Parameters for running large systems ( No.1 )
Date: 2021/09/18 11:17
Name: T. Ozaki

Hi,

If the unit cell size is large enough, it would be better to set keywords as follows:

scf.EigenvalueSolver cluster
scf.Kgrid 1 1 1

In this case, the eigenvalue solver is switiched to that for a real matrix case,
which is faster than the counterpart of complex matrix case.

Regards,

TO
メンテ
Re: Parameters for running large systems ( No.2 )
Date: 2021/09/23 11:21
Name: MChen

Dear Prof. Ozaki,
Many thanks for your reply. The parameter sounds for cluster systems. Will this be working for the band structure calculations? Thank you very much.

Best regards,

MChen
メンテ
Re: Parameters for running large systems ( No.3 )
Date: 2021/09/23 20:36
Name: T. Ozaki

Hi,

The cluster routine works for the band calculation with the gamma point, where the periodic
boundary condition is properly considered.

Regards,

TO
メンテ
Re: Parameters for running large systems ( No.4 )
Date: 2022/02/28 14:09
Name: Gurung

Hello Professor Ozaki,
I calculated the band structure for the SrTiO3 using cluster and band as eigenvaluesolver for simple test. I found very different results when comparing the band structure resulting from these two methods. The band structure from the VASP was comparable with that one using band eigenvaluesolver.

I used all the parameter to be same for both cluster and band method. I was expecting minor differences but the shape of the band structure changes drastically along some k path and the energy difference between some of the bands was hugely reduced. What might be the possible source of such differences? Maybe for them to be same, I need to change the cutoff energy as suggested in the known problems in the openmx manual.

***Version used is openmx3.9***

Best regards
Gurung
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