Problem with routine scf calculation exploding into large energies

[Abdulrahman_Allangawi]

Dear All,

I am attempting to run what should be typical Rutile TiO2 (110) optimization, however, during the first ionic step, the first scf iterations volatiles and eventually explodes into a very large number, causing the run to crash. It is very confusing to me as the same genetic setting worked with Rutile IrO2. I tried many stuff such as changing the KPOINTS, ENCUT, spearing width, ALGO, number of bands, spin-polarization, etc... and nothing worked. I attached the relevant filed, does anyone have an idea about such a problem?

Regards,

[alexey.tal]

Dear Abdulrahman_Allangawi,

Thank you for your question.

I just tried to run your calculation on 16 MPI ranks and the first ionic iteration converged in just under 1 min without any issue. I see that you are running this calculation on 768 MPI-ranks, but I don't have access to such a large system to test if I can reproduce this issue on the same number of ranks. However, this system appears to be too small to scale to that many ranks anyway. Could you please try to run the same calculation on a fewer ranks?

Best wishes,
Alexey

[Abdulrahman_Allangawi]

Dear,

Thanks for your response.

I lowered the number of ranks, but it still did not converge. I changed the mixing parameters, this seem to have worked.