Discrepency between MPP and Pcrytsal

Rams

Hi,

I am trying to simulate a slab model of thin film on substrate. When I ran the job, there was a convergence problem in the SCF cycle. My initial suspect was the interface, since I don't have extract geometrical configuration it was only a pure guess.

So I ran a calculation with only the substrate, but still it did not convergence in SCF cycle with MPPcrystal. I reran the same calculation of only substrate with Pcrystal it converged in 20 cycles. So why is there discrepancy between both versions. I am attaching the input and output for both calculation here.
Substrate_MPP.out
Substrate_MPP.d12
Substrate_Pcrystal.out
Substrate_Pcrystal.d12

To check if there is any issue with the MPP compilation, I added the REPLDATA tag in MPP input, so that it runs as Pcrystal, this time it again converged in 20 cycles.

I would like to use MPPcrystal for my final structure of thin film to speedup the calculation and avoid memory issues, so could you let me what is the issue I am encountering with both the versions.

Thanks & Regards,
Rams

GiacomoAmbrogio

Hi Rams,

I’ll look into this issue further, but in the meantime, could you please try running the MPP version with the same input, just without the DIIS keyword, and let me know if that converge?

Rams

Hi Giacomo,

Thanks for your reply. I did some test by removing the entire DIIS part, and keeping only the Fock mixing.

with a FMIXING of 95% the SCF converged in 84 cycles.
with 70% in 17 cycles
with the default 30%, it doesn't converges - also the atomic charges were distributed not properly, charges are fully either at the top / bottom layer, and this oscillates between cycle (screenshot below)

--> But this works only with the substrate, for the entire structure with thin film, FMIX of 70% or 95% gives a very similar behavior of charge distribution in 3-4 scf cycles. Only with 99% it moves properly, but that actually means I am practically forcing the system towards the path right ? it may not be a real ground state.

Looking forward towards your opinion and suggestions

Thanks & Regards,
Rams

GiacomoAmbrogio

Hi Rams,

The need for relatively high values of FMIXING is not unusual, so what you are observing is expected. That said, you are correct that if FMIXING is set extremely high (close to 100%), there is a risk that the SCF procedure may reach converge before reaching the "true" ground state. This is not unique to your system, it is a general trade-off with strong damping.

To improve robustness while still guiding the calculation toward the correct solution, you can try:

Increasing TOLDEE: this tightens the SCF energy convergence criterion and will allow the SCF to continue reaching the minimum even with high FMIXING. Keep in mind, however, that this will typically require many more SCF cycles.
Combining mixing with other stabilization strategies, such as level shifting (LEVSHIFT) or electronic smearing (SMEAR) (if applicable to your system), which often reduce the need for such extreme damping.