[QE-developers] Possible Bug in Tkatchenko-Scheffler

[c.h.bermudez at cox.net]

Hello,

 

         I may be encountering a bug in vers. 6.4.1 in connection with
relaxation using vdW-TS. The attached files contain input and output for 2
different runs looking at adsorption of NO2 on an MoS2 monolayer. Everything
is the same (I believe), and the input structures are very similar. In fact,
the input to #245329 is the final-step output from #244653. The one (known)
difference is that #244653 uses 140/560 cut-offs and #245329 uses 100/400
cut-offs. The disturbing finding is that the TS dispersion corrections are
very different for the two (approximately -0.71925 Ryd for one and -0.72919
Ryd for the other). This is playing havoc with my attempts to obtain
adsorption energies. 

         I've done similar comparisons for different cut-offs in other runs
using vers. 6.2 and 6.4.1 and not seen differences anywhere near this large,
except in one case where it may have been half this big. In other words,
these large differences in TS dispersion corrections for nominally-similar
conditions appear to be sporadic. I should add that I've tested the Hamann
ONCV pseudopotentials that I'm using and find that they give very good
results for my systems of interest. I've also tested the convergence with
respect to the Monkhorst Pack grid, and that's very good. 

         One other point, if I may. The 6.4 release notes say that vc-relax
with TS wasn't working in earlier versions. Before being made aware of this
I was using vc-relax with TS in vers. 6.2 with no obvious problems. The
stress, forces and total energy all converged nicely, and the final scf
(when it didn't crash) was very close to the last vc step. In what way was
vc-relax + TS not working in vers. 6.2 ? 

         Any information, especially advice about a work-around if this is a
real bug, would be greatly appreciated. 

 

Best Wishes,

Vic Bermudez

bermudez at alum.mit.edu <mailto:bermudez at alum.mit.edu>  

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20191206/93b9bfe8/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: IO_244653.zip
Type: application/x-zip-compressed
Size: 58316 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20191206/93b9bfe8/attachment-0002.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: IO_245329.zip
Type: application/x-zip-compressed
Size: 184268 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20191206/93b9bfe8/attachment-0003.bin>