[Pw_forum] Another SCF convergence problem

Matteo Cococcioni matteo at umn.edu
Wed Oct 3 01:20:14 CEST 2007

Dear Vidur

I gave a look to your input and also to the crystal structure of your 
system (xcrysden --pwi swnt_bndl.pw.inp).

I noticed few things that could give troubles and it looks like some of 
the input parameters probably were not converged.

are you sure that 20 Ry is enough for a NC pseudopotential?

degauss = 0.5 looks to me like a large number. even if your calculation 
were converged it would have converged to a wrong
result. Keep in mind this is Ry.

your k-point mesh is probably too small. Though your system is quite 
large on the xy plane I don't think 1 point is enough.
two points in z directions are definitely too few. In your case also the 
second point looks equivalent to the first one  to me.
I'm not 100% sure actually because you gave it in tpiba units and I'm 
not sure what the code uses as a in this case (probably celldm(1)).
Anyhow if the second point is not equivalent to the first one it is in a 
very weird position.
I would go for an automatically generated mesh (e.g. 228 or 448).


Vidur Parkash wrote:
> Hi
> I'm trying to perform a SCf calculation on a carbon nanotube cluster
> (metallic system), and no matter what i do the results do not
> converge.
> I've tried changing the mixing factor, add a smearing, tried a
> different type of charge mixing, change number of k points etc.
> The input file I've attached is the closest it ever reached towards
> convergence with accuracy better than 0.1 Ry...but ultimately failing
> to converge. 
> Can you please suggest some solutions. The input file is attached
> this time.
> Thanks
> Vidur Parkash
> Michigan Tech University
> 716 EERC, 1400 Townsend Drive
> Houghton MI-49931, USA
> Tel: 1-906-487-0023
> ------------------------------------------------------------------------
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum

-------------- next part --------------
A non-text attachment was scrubbed...
Name: matteo.vcf
Type: text/x-vcard
Size: 294 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20071002/44e45abc/attachment.vcf>

More information about the users mailing list