SCF convergence - CoO for XAS Calculation

[Börries von Seggern]

Hey guys,

I`m intrested in calculating XAS Spectra of various transition metal oxides and wanted to do some reference calculations. Unfortunately my calculation for Cobalt oxide (CoO) fails to converge completely using the OT method and is very very demanding using the DIAG method. Am I missing something or is this really standard behaviour?

Please find some input and output files attached, the OT calculation has been performed using 12 cores of a Xeon Skylake 6130, while the DIAG calculation has been performed on 24 cores of 2 Xeon Skylake 6130s.

Thanks a lot for any advice,

Börries

[Marcella Iannuzzi]

Dear Börries, 

The calculation in CoO_calc_DIAG.out also did not converge. 

Actually the SCF shows a rather weird behaviour. 

The difference in number of ERIs  integrals- and then in timings- between the two calculations (OT and DIAG) is rather large, and it is caused by the setting of

EPS_SCHWARZ 1.0E-5 

in the OT calculation. This value is absolutely too large, and might also be one of the reasons why the ground state electronic structure does not converge.

It is not recommendable to start the XAS calculation before obtaining a well converged ground state. 

Regards

Marcella

[Börries von Seggern]

Dear Marcella,

thanks a lot for the advice!

I`ll try it again with the standard for EPS_SCHWARZ.

Kind regards,

Börries