[Smeagol-discuss] What's wrong with the Pulay mixing?

[Xiaohong Zheng]

Hello, Dear All,

Pulay mixing is a convergency accelerator. However, in a very simple
test, it was found that it slowed down the convergence. The
"DM.NumberPulay" was set to 5 and "DM.MixingWeight" is set to 0.1. In
the scf process, we can see that for every 5 steps, "dDmax" does not get
better, but gets worse. It is oscilating!

However, if I set "DM.NumberPulay 5" and "DM.MixingWeight 0.2" or just
set "DM.NumberPulay 1", then it works very well and converges with 5 or
6 steps. Thus, I think there must be some problem with the Pulay mixing.
Does anybody know why?

The input and output files and the pseudopotential file is attached.
Thank you very much.

Xiaohong
-------------- next part --------------
A non-text attachment was scrubbed...
Name: lead2.fdf
Type: application/vnd.fdf
Size: 2114 bytes
Desc: not available
Url : http://lists.tchpc.tcd.ie/pipermail/smeagol-discuss/attachments/20100617/1bb80315/attachment-0001.bin 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: lead2.out
Url: http://lists.tchpc.tcd.ie/pipermail/smeagol-discuss/attachments/20100617/1bb80315/attachment.pl 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Li.psf
Url: http://lists.tchpc.tcd.ie/pipermail/smeagol-discuss/attachments/20100617/1bb80315/attachment-0001.pl