[Pw_forum] phonon parallel calculations

Andrea Dal Corso dalcorso at sissa.it
Thu Feb 26 09:49:21 CET 2015


The problem is that 'image parallelization' and 'GRID approach' are two
complementary approaches. I think that the two together have not been
tested.
In any case your idea to use only_init and then images is interesting,
but probably will need some tuning.

Andrea


On Thu, 2015-02-26 at 00:44 +0100, Carlo Nervi wrote:
> I did some other test with the "GRID" approach above outlined, and I cannot
> understand where I'm wrong.
> I tried different libraries during the compilation process, but while the
> "normal" phonon calculations runs smoothly (but slower, about 1 hour), the
> "GRID" approach always gave me bad (negative) frequencies (but much faster,
> about 10 minutes). I'm using ecut=60 and erho=480
> 
> The phonon results from without using the "-ni 8" approach is:
>      freq (    1) =      -0.219517 [THz] =      -7.322283 [cm-1]
>      freq (    2) =       0.237288 [THz] =       7.915064 [cm-1]
>      freq (    3) =       0.412515 [THz] =      13.760017 [cm-1]
>      freq (    4) =       0.888921 [THz] =      29.651216 [cm-1]
>      freq (    5) =       1.818322 [THz] =      60.652688 [cm-1]
>      freq (    6) =       3.267016 [THz] =     108.975924 [cm-1]
>      freq (    7) =       3.944444 [THz] =     131.572497 [cm-1]
>      freq (    8) =      13.702502 [THz] =     457.066270 [cm-1]
>      freq (    9) =      15.986666 [THz] =     533.257782 [cm-1]
>      freq (   10) =      21.306349 [THz] =     710.703307 [cm-1]
>      freq (   11) =      29.251729 [THz] =     975.732641 [cm-1]
>      freq (   12) =      31.330252 [THz] =    1045.064724 [cm-1]
>      freq (   13) =      40.094614 [THz] =    1337.412380 [cm-1]
>      freq (   14) =      91.247143 [THz] =    3043.677078 [cm-1]
>      freq (   15) =      94.196285 [THz] =    3142.049865 [cm-1]
> 
> If I use the "-ni 8" (GRID) I got:
>     freq (    1) =     -14.549865 [THz] =    -485.331266 [cm-1]
>      freq (    2) =     -10.853880 [THz] =    -362.046453 [cm-1]
>      freq (    3) =      -9.921965 [THz] =    -330.961126 [cm-1]
>      freq (    4) =      -9.680609 [THz] =    -322.910351 [cm-1]
>      freq (    5) =      -9.593462 [THz] =    -320.003431 [cm-1]
>      freq (    6) =      -9.461406 [THz] =    -315.598531 [cm-1]
>      freq (    7) =      -8.724432 [THz] =    -291.015732 [cm-1]
>      freq (    8) =      -4.051543 [THz] =    -135.144927 [cm-1]
>      freq (    9) =       8.067478 [THz] =     269.102094 [cm-1]
>      freq (   10) =      19.972800 [THz] =     666.220885 [cm-1]
>      freq (   11) =      29.137064 [THz] =     971.907843 [cm-1]
>      freq (   12) =      30.784477 [THz] =    1026.859612 [cm-1]
>      freq (   13) =      39.868296 [THz] =    1329.863222 [cm-1]
>      freq (   14) =      91.702737 [THz] =    3058.874040 [cm-1]
>      freq (   15) =      94.287387 [THz] =    3145.088688 [cm-1]
> 
> A lot of negative frequencies appear and sometimes (with a bigger system) I
> got a ************ in the matdyn file (in this case an error is printed).
> Any idea?
> Anybody else had this kind of problem?
> Thank you,
> 
>  Carlo
> 
> >
> > --
> 
> ------------------------------------------------------------
> Prof. Carlo Nervi carlo.nervi at unito.it  Tel:+39 0116707507/8
> Fax: +39 0116707855      -      Dipartimento di Chimica, via
> P. Giuria 7, 10125 Torino, Italy.    http://lem.ch.unito.it/
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum

-- 
Andrea Dal Corso                    Tel. 0039-040-3787428
SISSA, Via Bonomea 265              Fax. 0039-040-3787249
I-34136 Trieste (Italy)             e-mail: dalcorso at sissa.it





More information about the users mailing list