SCF setting about geometry optimization or variable cell optimization

Top Page > Browsing

SCF setting about geometry optimization or variable cell optimization

Date: 2021/06/01 12:55
Name: Z: Dear developers and users of OpenMx,

Hi, everyone! I am a beginner of DFT and OpenMx. If I want to get a MD.Opt faster when I calculate a large system, could I set the SCF section as follow:

scf.energycutoff 100 # default=150 (Ry)
scf.criterion 1.0e-3 # default=1.0e-6 (Hartree)

even:

scf.maxIter 10 # default=40

Whether the scf convergence setting will affect geometry optimization or variable cell optimization? If optimization involves only the calculation of interatomic forces, I think such a setting might be efficient. Could someone give me some guidance?

Regards,
Z

Page: [1]

Re: SCF setting about geometry optimization or variable cell optimization ( No.1 )

Date: 2021/06/01 13:17
Name: T. Ozaki

Hi,

I am wondering that the setting looks crude, and may cause some oscillation during optimization
due to insufficient convergence especially for the SCF convergence.
Instead, I might suggest the following setting:

scf.energycutoff 140
scf.criterion 1.0e-5
scf.maxIter 40

Of course, you may find a proper setting for your specific system by looking at results performed
with a different parameter setting.

Regards,

TO

Re: SCF setting about geometry optimization or variable cell optimization ( No.2 )

Date: 2021/06/01 16:56
Name: Z

Dear T. Ozaki,

Thanks for your help and advice.

Regards,
Z

Page: [1]