[Pw_forum] About restarting ph.x calculation
William Parker
wparker at anl.gov
Tue Oct 16 20:52:42 CEST 2012
Hi James,
I didn't see anyone reply to your questions yet so I thought I'd take a shot.
To put _ph0 in a different place than (prefix).save would require a little coding. As you noticed, outdir is where the input, output and temporary files all reside.
If your _ph0/(prefix).phsave/ folder contains all of the data-file.(qpt #).(irr #).xml files, then recover=.true. should spit out all of dynamical matrix files properly. Reputedly, if you run with nimage > 1, ph.x should find these files in all of _ph(image #)/(prefix).phsave/ directories. This does not work for me in 5.0.1 and earlier versions--I must manually (in my script) copy the files to the _ph0/(prefix).phsave/ directory before running with recover=.true., otherwise ph.x starts re-calculating all of the representations not in _ph0/(prefix).phsave/
Only the run with the first representation at Gamma does the electric fields calculation.
Check out the "ADDITIONAL INFORMATION" section at the end of PHonon/Doc/INPUT_PW.txt for additional information.
--William
On Oct 12, 2012, at 3:38 PM, James Mao wrote:
> Hi everyone,
>
> I need some help for restarting ph.x calculations. I searched archives and
> read most posts about it, but still not sure about several things.
>
> Basically I need calculate IR and Raman for a pretty big crystal cell. It generates
> huge scratch files, and of course takes quite a long time. Calculations definitely
> will be terminated from time to time due to my limited run time on cluster.
>
> So my question:
>
> 1) Is there any way to assign folder "_ph0" to a different disk from "prefix.save"?
> I think "prefix.save" from relax calculation is for ph.x to read, and ph.x will not write
> anything into it, right? "_ph0" is where huge scratch data will be kept and if possible
> I would like to assign it to local disk on cluster, while keeping "prefix.save" on host.
> I think "outdir" will assign both to a same location.
>
> 2) To restart a ph.x calculation (recover=.true.), which data are necessary? (do I need
> prefix.save folder? do I need all files under _ph0 folder or just _ph0\prefix.phsave folder?)
>
> 3) Can I calculate several representations at same time by submitting several calculations
> at same time (assigning differet start_irr and last_irr)? I noticed before calculating
> representations, electric fields calculation is done first. If I split the job, will each job
> repeat same electric fields calculation? Any way to skip it if I already have it done once?
>
> Thanks first for any help!
>
> btw: is our archives website still working well? It seems recently some posts I received in
> email cannot be found in the website.
>
> Best Regards,
> James Mao
> -------------------
> Postdoc, Department of Chemistry, University of Pittsburgh
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
*********************************************************
William D. Parker phone: +1 (630) 252-3388
Argonne Scholar
Argonne Leadership Computing Facility
Building 240, Office 1-D.26
Argonne National Laboratory
9700 S. Cass Ave.
Argonne, IL 60439
U.S.A.
*********************************************************
More information about the users
mailing list