Wrong phonon dispersion with OpenMX

[Youhaojen]

Dear Prof. Tadano,

I tried to run ALAMODE with OpenMX to calculate the phonon dispersion of GaN. I also used the --offset tag when getting the force set, but the fitting error of force constant is still very high (Fitting error (%): 86.0102). As a result, I ended up with imaginary phonon dispersion.

only_band.pdf

Could you please help me identify what might have gone wrong?
I’ve attached both the input and output files below.

input_output.zip

Additionally, I’ve also included the phonon dispersion of GaN calculated using Phonopy with VASP for comparison.

band.pdf

Sincerely,
Hao-Jen You

[ttadano]

For the anphon code, you will need to use the primitive cell lattice vectors as

&cell
8.07570871302273692286
 0.0 0.5 0.5
 0.5 0.0 0.5
 0.5 0.5 0.0
/

However, this modification did not solve the imaginary frequency issue.
I did not find any obvious errors in the inputs (except the above one), so I guess you will need to increase the cutoff energy in the OpenMX.

The total energy of the displaced configuration (openmx-1.out) is lower than that of the undisplaced configuration (openmx-0.out). This also indicates a numerical issue with the OpenMX calculation.

If increasing the cutoff energy does not work, please consider using a larger atomic displacement (0.03 Angstrom).

[quynhxinhdep]

Dear Dr. Hao-Jen
Did you fix your problem
I get the same problem. even I increase energy cutoff and using a larger atomic displacement (0.03 Angstrom), my results become worse