Importance of scf.fixed.grid in phonon calculations

Top Page > Browsing

Importance of scf.fixed.grid in phonon calculations

Date: 2023/05/30 12:17
Name: Masanobu Miyata <m-miyata@jaist.ac.jp>: Dear Ozaki-sensei,

　When running a phonon calculation, should I explicitly state "scf.fixed.grid 0 0 0" for all displacement pattern input files in the process of calculating total energy for multiple displacement patterns?　
　Referring to the OpenMX manual, it explains: "The total energy depends on the atomic coordinates and the relative position between the grids. When calculating interaction energies or energy curves as a function of atomic coordinates, it is important to maintain this relative position in all calculations required for those series of calculations, because it is important to reduce errors".
　It seems more accurate to align the Grid Origin by explicitly stating "scf.fixed.grid 0 0 0" in all files.　For simple systems such as Si, no difference occurred, but for systems where weak interactions are important, the phonon bands changed slightly.

Best regards,
Miyata

Page: [1]

Re: Importance of scf.fixed.grid in phonon calculations ( No.1 )

Date: 2023/05/30 14:07
Name: T. Ozaki

Hi,

It would be better to fix the grid origin by the keyword:

scf.fixed.grid 0 0 0

"0 0 0" may be proper when the fractional coordinates are used to
specify the position of atoms.

For the cases:
Atoms.SpeciesAndCoordinates.Unit Ang # Ang|AU

you may get the grid origin from the standard output for
an input file. Then, it would be better to keep the same grid origin
for all the other cases by the keyword.

Regards,

TO

Re: Importance of scf.fixed.grid in phonon calculations ( No.2 )

Date: 2023/05/30 14:39
Name: Masanobu Miyata <m-miyata@jaist.ac.jp>

Dear Ozaki-sensei,

Thank you for your prompt reply.

I understand that I should use scf.fixed.grid to fix the origin.

In the case of Cartesian coordinates, I will read the Grid Origin from the standard output of the input file and fix the origin of all displacement pattern files at those coordinates.

Thank you very much for your answer.

Best regards,
Miyata

Page: [1]