[CP2K-user] Converging plane-wave cutoff, with and without density smoothing

[Luc Labarrière]

Dear  All,
I would like to do molecular dynamics using BLYP with the GPW method. My 
system is an organic molecule in methanol so i followed the procedure to 
converge the plane-wave cutoff (the example with water molecules, I used 
the exact same input files provided to test if I was doing things 
correctly).
I tried: without density smoothing, with NN50 density smoothing + 
NN50_SMOOTH, NN10 density SMOOTHING + SPLINE2_SMOOTH, and NN10 + SPLINE2. 
The total charge density converges at 340 Ry without any problem in all 
cases, however, when looking at the energy things go wrong.
(All examples use a REL_CUTOFF of 60)
The NN10 and NN50 methods don't converge even at 2000 Ry.
Without density smoothing, the energy is approximately converged at 660 Ry, 
however there are oscillations:
660Ry: -551.02598 au
680Ry: -551.02598 au 
700Ry: -551.02598 au
720Ry: -551.02598 au

740Ry: -551.02489 au

760Ry: -551.02586 au
up to 920 Ry

940Ry:  -551.02611 au
up to 1040 Ry

and it keeps going (I went up to 2000 Ry)

(I also tested PBE without density smoothing, which converges to around 
10^-6 au, before oscillating in the same way)

Does any of you know if I'm doing anything wrong ? I tested the procedure 
on CP2K 7.1 that I compiled on my personal computer, as well as on CP2K 7.1 
and 5.1 on a national super computer.

Thanks in advance and best regards,
Luc Labarrière
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20210316/56488f6e/attachment.htm>