[Pw_forum] Visualization of Local DOS
degironc at sissa.it
Sat Sep 15 08:34:27 CEST 2007
Brad Malone wrote:
> Actually, when I tried plotting with the pp.x program I got the
> following error:
> task # 0
> from set_fft_dim : error # 1
> input nr1 value not allowed
the complaint comes from the fft part.
It clearly says that the first fft dimension you are using (45) is not
allowed by the FFT library you are using.
Different libraries accept different values for their dimensions.
Have you generated the file on one machine (using a given library) and
try to analyse it on a different machine (with a different library)?
If yes, the problem is likely the incompatibility of the requirements of
the two libraries.
Use the input variables nr1,nr2,nr3 in pwscf to overwrite the values
that the code choose by default so as to find a value that is accepted
by both libraries: if 45=3*3*5 is not accepted try 48=3*2*2*2*2 ...(46
or 47 are likely universally bad choices...).
If the code and the post-processing have been run on the same machine
with the same library then there is something strange. In such a case
please provide more information on what you are doing so as to make it
hope this helps,
stefano de Gironcoli,
SISSA and DEMOCRITOS
> The top of my local density of states file is:
> 45 45 108 45 45 108 24 2
> 0 1.00000000 0.00000000 0.00000000 0.00000000
> 0.00000000 0.00000000
> 10.75226179600000 -5.1673100000000005E-004
> -5.1773999999999996E-004 10.75226235600000
> 8.8574731000000000E-002 -8.8575347000000004E-002
> 4.0528473457 4.0000000000 40.0000000000 10
> So the progam has an issue with the number '45' it appears. Is there
> some reason why this shouldn't be allowed? I saw that the
> requirements for nr1 to be 'allowed' or not is the same in v2 and it
> is in v3, looking at the set_fft_dim.f90 file. Is there perhaps
> another difference as to why the FFT value isn't allowed?
> I could of course do the entire calculation over on the system with
> v3, but it takes a decent amount of time so I'd prefer to not if I can
> get away with it, especially since I don't know for sure if that would
> fix the problem.
> Thanks again,
> UC Berkeley
> Pw_forum mailing list
> Pw_forum at pwscf.org
More information about the users