[ase-users] GULP calculator in ASE with periodic boundary conditions

[Ask Hjorth Larsen]

Hi James,

2018-01-17 17:47 GMT+01:00 James Aston <J.C.Aston at sms.ed.ac.uk>:
> Hi,
>
> Thanks for the response. My output for my periodic crystal has 'Final
> internal derivatives' in units of eV instead. I have tried to change the
> output of this by searching for different gulp keywords but couldn't find
> any solutions. Do you think it would wise for me to proceed by parsing the
> gulp output file for the internal derivatives and using this information for
> my phonon calculations?

What we need is to generalize the parser, yes, since the current
calculator does not pick up the necessary outputs.

Either ASE can give a keyword to make Gulp always output them in the
right way (this may interfere a bit with how predictable/simple/clean
the interface is), or it needs to parse all the possible ways of
formatting the output.

Most likely it is easy enough to parse all the possible outputs in this case.

I have the example outputs from the online GULP documentation, grepped
among them like this:

cat *.got | grep -i derivatives | sort | uniq  | grep -v "Calculation of"

Gives a not-too-big output which includes the following:

  Final Cartesian derivatives :
  Final cell parameters and derivatives :
  Final derivatives for region 1 :
  Final internal/Cartesian derivatives :
  Final internal derivatives :
  Final surface cell parameters and derivatives :

I am not sure how things work with "regions" and how ASE can deal with
that, but the other cases are probably simple to accommodate in the
parser.

One could go so far as to generalize the parser as an IO format.  Some
calculators do that.

Oh right - come to think of it, I have a parser lying around for the
GULP "marvin" format which I can contribute to ASE at any time.  Then
it would be possible for the ASE calculator to e.g. always request a
marvin file, and hence always be able to parse it.  Would this be
interesting?  Then we don't have to worry so much about the text
format, probably.

Best regards
Ask


>
> Best wishes,
> James
>
>
> Quoting Ask Hjorth Larsen <asklarsen at gmail.com> on Tue, 16 Jan 2018 18:41:48
> +0100:
>
>> Hi,
>>
>> 2018-01-16 18:05 GMT+01:00 James Aston via ase-users
>> <ase-users at listserv.fysik.dtu.dk>:
>>>
>>> Hello everyone,
>>>
>>> I am a PhD student at the University of Edinburgh and a new user of ASE.
>>> I
>>> am having trouble using the GULP calculator to determine the forces and
>>> phonon dispersion of a periodic crystal structure.
>>>
>>> I am currently trying to use the 8-Si example given here
>>>
>>> (https://github.com/atztogo/phonopy/blob/master/example/ase/8Si-phonon.py).
>>> I have set the calculator to calc = GULP(library='tersoff.lib',
>>> keywords="conp, gradient, fractional, nosymmetry").
>>
>>
>> (The examples in ASE use "conp gradient ...." without commas.  I infer
>> that probably it is supposed to work anyway though.)
>>
>>>
>>> The code runs when periodic boundary conditions are set to false, albeit
>>> with negative phonon frequencies. When I set periodic boundary conditions
>>> to
>>> true I get an error from get_forces():
>>>
>>> 'ase.calculators.calculator.PropertyNotImplementedError: forces not
>>> present
>>> in this calculation'
>>>
>>> Does anyone know how to fix this? Thank you in advance.
>>
>>
>> Probably the forces are printed differently so the parser does not find
>> them.
>>
>> The parser looks only for "Final Cartesian derivatives".  Does your
>> output have that?
>>
>> Best regards
>> Ask
>>
>>>
>>> Best wishes,
>>> James
>>>
>>>
>>> --
>>> The University of Edinburgh is a charitable body, registered in
>>> Scotland, with registration number SC005336.
>>>
>>>
>>> _______________________________________________
>>> ase-users mailing list
>>> ase-users at listserv.fysik.dtu.dk
>>> https://listserv.fysik.dtu.dk/mailman/listinfo/ase-users
>>
>>
>>
>
>
>
> --
> The University of Edinburgh is a charitable body, registered in
> Scotland, with registration number SC005336.
>
>