RESCU+ Band structure calculation

Kan

Dear Nanoacademic Team,

The band structure calculation would require a lot of time if the model contains thousands of atoms. Is there some parameters to improve the calculation speed?
Besides, the .log for band structure calculation only display the total elapsed time. It is hard to estimate the total elapsed time when the job start to run unless the time for every band calculation can display in the .log file.

Sincerely,
Kan

siaodani

Hi have you tired increasing the number of cores by setting
calc.solver.set_mpi_command("mpiexec -n 16")
where 16 can be replaced with 32 , 64 , 100, 200, etc ...

plz take a look at
https://docs.nanoacademic.com/rescupy/tutorials/etot/tutorial_etot/
https://docs.nanoacademic.com/rescupy/tutorials/bs/tutorial_bs/

Kan

siaodani Hi, I've tried calc.solver.set_mpi_command("mpiexec -n 320"), and calc.solver.eig.inclband=[1,4780]; calc.solver.eig.target_irange=[4680,4780], and I'd like to know is there some other parameter could affect the calculation speed.
Kan

siaodani

Kan what's the number of k-points and number of atomic orbitals?

Kan

siaodani Usually I set calc.system.set_kpoint_path(special_points=["G","M","K","G"],grid=200), and the number of atomic orbitals is default. I use the target_irange because the tutorial says this command would only converge the index range of the energy bands, but it seems just make little difference.