[Pw_forum] Problem with convergence parameters forc_conv_thr and etot_conv_thr

Gabriel Pagola gpagola at df.uba.ar
Tue Jun 19 06:27:03 CEST 2012


Dear all,
I am trying
to run an optimization of one crystal system using the keyword ‘wc-relax’. 
 
I used the
following values as convergence parameters:
 
    forc_conv_thr
= 3.D-2
    etot_conv_thr=5.0D-1
 
I have put
very “high” values in order to be sure that the calculations finish in a few
iterations. But, as you can see in the output attached, the QE continues making
iterations although the conditions of convergence are clearly fulfilled.
For example
seeing the total energy in first iterations:
 
!    total energy              =   -1159.31135867 Ry
!    total energy              =   -1158.59596269 Ry
!    total energy              =   -1160.05552255 Ry
!    total energy              =   -1159.98617122 Ry
!    total energy              =   -1160.17541568 Ry
!    total energy              =   -1160.14018816 Ry
!    total energy              =   -1160.22211554 Ry
!    total energy              =   -1160.22852050 Ry
!    total energy              =   -1160.23360361 Ry
!    total energy              =   -1160.23596633 Ry
 
The parameter
of convergence etot_conv_thr has been fulfilled since the 7th or the 8th
iteration. The same has happened with the force an forc_conv_thr
 
I don’t
know why this happens. Perhaps I have misunderstood the manual of QE.
 
I have
attached the entire input and output
 
Any help
would be appreciated.  Thank in advance
Gabriel Pagola    

Physics Department
FCE. y N, Universidad de Buenos Aires
(1428)Ciudad Universitaria-Pab. I
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20120618/ba952ad3/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: IMP_OUT.zip
Type: application/zip
Size: 124212 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20120618/ba952ad3/attachment.zip>


More information about the users mailing list