[Octopus-users] Writing of restart info

[Conn O'Rourke]

Hi Matthieu,

Thanks for the advice - I'll try the gradually lowering the tolerance like
you suggest.

Cheers,
Conn

On Mon, Feb 22, 2010 at 4:22 PM, matthieu verstraete <
Matthieu.Verstraete at ulg.ac.be> wrote:

> There are no SCF iterations, in the sense that for nonSCF runs (unocc)
> the hamiltonian is fixed (because the density is). It looks like
> OutputEvery is linked to this loop over SC evaluation of the
> hamiltonian. Formally, an unocc run is like a single gs iteration (so
> outputevery 1 anyway).
>
> Your only recourse is to limit the number of steps (I believe
> MaximumIter  also governs the unocc run mode, although the doc claims
> it's only for SCF: try with MaximumIter = 1 to see if it stops
> immediately) and set FromScratch=no for the next round.
>
> Or try your luck with an initial high tolerance on the unocc states,
> then lower it for the next tries of convergence.
>
> Matthieu
>
> On Mon, Feb 22, 2010 at 4:40 PM, Conn O'Rourke
> <conn.orourke at googlemail.com> wrote:
> > Hi All,
> >
> > So I have tried Emily's suggestion of using OutputEvery to get restart
> > information written at regular intervals when running an unocc
> calculation,
> > but it doesn't seem to work.
> >
> > In the variables list it says that OutputEvery works for gs and td runs,
> so
> > does this mean that the variable doesn't apply to unocc calculations? It
> > seems to only want to write the restart info at the very end of the
> > calculation.
> >
> > Any suggestions would be appreciated!
> >
> > Cheers,
> > Conn
> >
> > On Mon, Feb 22, 2010 at 11:42 AM, Emily <listemily at eml.cc> wrote:
> >>
> >> my pleasure to help!
> >>
> >>
> >> ----- Original message -----
> >> From: "Conn O'Rourke" <conn.orourke at googlemail.com>
> >> To: "Emily" <listemily at eml.cc>
> >> Date: Mon, 22 Feb 2010 09:15:44 +0000
> >> Subject: Re: [Octopus-users] Writing of restart info
> >>
> >> Hi emily,
> >>
> >> Thanks for the reply - I'm surprised I couldn't find this. Thanks for
> >> saving
> >> me the bother!
> >>
> >> All the best,
> >> Conn
> >>
> >> On Mon, Feb 22, 2010 at 1:15 AM, Emily <listemily at eml.cc> wrote:
> >>
> >> >
> >> > I believe the variable OutputEvery controls that.
> >> >
> >> >
> >> >
> http://www.tddft.org/programs/octopus/doc/3.2.0/html/vars.php?section=Output&name=OutputEvery
> >> >
> >> > You can't specify how many minutes/hours go by before Octopus writes
> the
> >> > restart files, but you can at least specify how many iterations. You
> >> > just have to guess and check how many iterations you get in how much
> >> > time.
> >> >
> >> > :) Emily
> >> >
> >> >
> >> > ----- Original message -----
> >> > From: "Conn O'Rourke" <conn.orourke at googlemail.com>
> >> > To: octopus-users at tddft.org
> >> > Date: Sun, 21 Feb 2010 13:15:48 +0000
> >> > Subject: [Octopus-users] Writing of restart info
> >> >
> >> > Dear Octopus users,
> >> >
> >> > I'm running some unocc calculations on a cluster for which there is a
> >> > limited time for jobs.
> >> > I would like to be able to restart the jobs if they run over the time
> >> > limit,
> >> > but there sometimes is no restart information written to
> unocc_restart.
> >> >
> >> > Is there a variable which controls the writing of the restart
> >> > information?
> >> > (Apologies if there is one in the variable list, as I have looked
> >> > through
> >> > and really can't find one)
> >> >  Or could someone suggest where to look in the code to force the
> restart
> >> > information to be written at regular intervals.
> >> >
> >> > Thanks in advance,
> >> > Conn
> >> >
> >
> >
> > _______________________________________________
> > Octopus-users mailing list
> > Octopus-users at tddft.org
> > http://www.tddft.org/mailman/listinfo/octopus-users
> >
> >
>
>
>
> --
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Prof. Matthieu Verstraete
>
> Universite de Liège
> Institut de Physique, Bat. B5, 3/7
> Allée du 6 aout, 17
> B- 4000 Sart Tilman, Liège
> Belgium
>
> Associate member European Theoretical Spectroscopy Facility (ETSF
> http://www.etsf.eu)
>
> Phone : +32 4 366 90 17
> Fax   : +32 4 366 36 29
>
> Mail : matthieu.verstraete at ulg.ac.be
>          matthieu.jean.verstraete at gmail.com
>
> _______________________________________________
> Octopus-users mailing list
> Octopus-users at tddft.org
> http://www.tddft.org/mailman/listinfo/octopus-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.tddft.org/pipermail/octopus-users/attachments/20100222/a6adf07a/attachment-0001.html