[ADF-LIST] FDE calculations with RELAX and spinorbit ZORA never converging

[gosia olejniczak]

hi Erik,

Thanks a lot for email!


The check whether the FDE freeze-and-thaw are converged looks at the number
> of SCF cycles it takes ADF to converge.
> In case of spin-orbit ZORA this check may be broken, because the number of
> SCF cycles might be always too large.
> I think you should be able to calculate the FDE_NMR_relax example with
> spin-orbit ZORA such that it is in practice FDE converged, although ADF
> says "FDE not converged".
>

In my case the energy  ('Complete subsystem DFT energy') oscillates (but it
could be considered as ok), but even if i ignore the 'FDE not converged'
and proceed with NMR, the shielding values are totally wrong,

also: SCF converges for each fragment; scalar ZORA goes without problems
(FDE converges, NMR shielding values are ok)
i'll then take your advice and send the inputs to the support, thank you!



> It is not possible to set convergence thresholds for freeze-and-thaw
> calculations.
>

ok, thank you!



> Remark:
> If you use relativistic scalar ZORA or relativistic spin-orbit ZORA and
> use the Create mode to make the atoms, make sure that you use a ZORA basis
> set. If you use the Basis key that should have been taken into account
> already.
>

yes, thank you!
i'm generating inputs with pyadf, all that seems to be covered,

thanks again for all suggestions!
best wishes,
gosia



>
> Best regards,
> Erik van Lenthe
> SCM
>
> hi,
>
> i'm writing about FDE calculations with freeze-and-thaw and spin-orbit
> ZORA: are these calculations possible in ADF2019?
>
> in my calculations involving heavy elements i run into 2 problems:
> freeze-and-thaw calculations are not converging ("FDE NOT converged" and i
> checked that 'Complete subsystem DFT energy' oscillates) which leads to
> completely wrong shielding values (calculated in post-fde step for active
> subsystem)
>
> i checked on a simpler system, such as this test:
> https://www.scm.com/doc/ADF/Examples/FDE_NMR_relax.html
> and this test (as it is) is passing OK, but if i modify this input
> slightly, i.e. add spinorbit ZORA, then i also see that freeze-and-thaw
> calculations are not converging (no matter how many iterations i ask for);
>
> additional question is about setting convergence thresholds for
> freeze-and-thaw calculations - is it possible (it's not in the manual)
>
> i'm using ADF2019
>
> thanks! and best wishes,
> gosia
>
> _______________________________________________
> ADFlist mailing listADFlist at scm.comhttps://lists.scm.com/mailman/listinfo/adflist
>
>
> _______________________________________________
> ADFlist mailing list
> ADFlist at scm.com
> https://lists.scm.com/mailman/listinfo/adflist
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.scm.com/pipermail/adflist/attachments/20200611/64d6abde/attachment-0001.html>