DIIS unstable and no convergence

[mdsimula...@gmail.com]

Hi

I'm using DIRECT_P_MIXING and when DIIS kicks in, things become unstable.  Any ideas what would be causing that?  I've included my input.  The system is copper oxide and I'm using PBEsol/pc-1 in the GAPW method.  Thanks! 

[Anton Lytvynenko]

Hi,

19.09.2022 18:25, mdsimula...@gmail.com пише:

> Any ideas what would be causing that?

Whatever. Convergence is often unstable, especially when transition
metals are present.

1. Check again if your starting geometry is reasonable.

2. Try OT if you can.

3. If not, try to add smearing and/or try to change mixing method
(kerker_mixing with high beta was very useful in my hands when HOMO-LUMO
gap was low, try to check).

4. Let us know if it helps.

Btw outer_scf has limited relevance with diagonalization (with OT it
allows to restart the preconditioner). Anyway, if you put it, please
make sure that eps_scf in the inner and outer cycles match).

Maybe something is also wrong with your GAPW settings, but I can't check
this due to lack of experience.

Yours,

Anton