[gpaw-users] ValueError: Davidson eigensolver does not support band parallelization?

[Pu ZHANG]

Yes, it works well after I specify the eigiensolver.

Thanks, Ask! Although something's wrong in the tutorial, at least I get
help in very short time :)

Best regards, Pu Zhang
--
Postdoc
The group of Structured Electromagnetic Materials
Department of Photonics Engineering
Technical University of Denmark
Ørsteds Plads, Building 345v, 166
2800, Kgs. Lyngby
E-mail: puzha at fotonik.dtu.dk
Homepage: http://www.dtu.dk/Service/Telefonbog/Person?id=91944&tab=0
Phone: 45 25 36 43
Fax: 45 93 65 81


On Mon, Jul 7, 2014 at 3:28 PM, Ask Hjorth Larsen <asklarsen at gmail.com>
wrote:

> Hello Pu
>
> Unfortunately there's a bug in the IO code so that the eigensolver
> information is not saved in the gpw file, which I reported to the list
> recently.
>
> When using PW mode and rmm-diis which supports band parallelization,
> then saving and restarting, it reverts to davidson and crashes.
> Specify eigensolver='rmm-diis' when restarting to avoid the error.
> It's a bit unfortunate that the standard tutorial examples don't work,
> but really it is the IO code which should be fixed.
>
> JJ (CC'd), are you sure we shouldn't fix this now before it generates
> too much confusion?
>
> Best regards
> Ask
>
>
> 2014-07-07 14:56 GMT+02:00 Pu ZHANG <puzhang0702 at gmail.com>:
> > Hi, all!
> >
> > When I run the tutorial example Structure optimization (see
> > https://wiki.fysik.dtu.dk/gpaw/tutorials/H2/optimization.html), I get
> some
> > error. From the error messages I can see something like:
> >
> >
> ****************************************************************************
> > * hwloc has encountered an out-of-order topology discovery.
> > * An object with cpuset 0x0000000a was inserted after object with 0x0
> > * Please check that your input topology (XML file, etc.) is valid.
> >
> ****************************************************************************
> >
> > librdmacm: Fatal: no RDMA devices found
> >
> > ValueError: Davidson eigensolver does not support band parallelization.
> > This calculation parallelizes over 2 band groups.
> >
> > Besides, I attach the error log with the email.
> >
> > I just copied the script from the tutorial. Is there anything wrong with
> the
> > eigensolver?
> >
> > Thanks!
> >
> > Best regards, Pu Zhang
> >
> > --
> > Postdoc
> > The group of Structured Electromagnetic Materials
> > Department of Photonics Engineering
> > Technical University of Denmark
> > Ørsteds Plads, Building 345v, 166
> > 2800, Kgs. Lyngby
> > E-mail: puzha at fotonik.dtu.dk
> > Homepage: http://www.dtu.dk/Service/Telefonbog/Person?id=91944&tab=0
> > Phone: 45 25 36 43
> > Fax: 45 93 65 81
> >
> > _______________________________________________
> > gpaw-users mailing list
> > gpaw-users at listserv.fysik.dtu.dk
> > https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20140707/61ebe836/attachment-0001.html>