[ase-users] system_changes always indicates everything has changed

[Ask Hjorth Larsen]

Hi,

2018-01-08 19:29 GMT+01:00 James Kermode via ase-users
<ase-users at listserv.fysik.dtu.dk>:
> Hi Ellad
>
> I think I’ve observed this behaviour before when I’ve forgotten to call the base-class method, Calculator.calculate() from the subclass method.

Yes... we should actually improve this situation.  Presently one has
to memorize all aspects of the base Calculator in order to know what
to do, because there is no logical way to infer that it behaves like
that.  Imagine rather that a particular calculator implementation can
explicitly 'subscribe' to some services, such as caching of
parameters, caching of results, tracking changes between invocations,
managing restart/prefix/label and so on.  Each of these should be
relatively small and predictable mechanisms.

Right now you subclass Calculator and you get the whole package.  This
somewhat bogs down the interface because we cannot change it without
subtly breaking many calculators.

But probably I won't get time to try improving this any time soon.

Best regards
Ask

>
> Best wishes
> James.
>
>> On 8 Jan 2018, at 17:31, Ellad Tadmor via ase-users <ase-users at listserv.fysik.dtu.dk> wrote:
>>
>> Hello--
>>
>> According to the documentation on adding new calculators (https://wiki.fysik.dtu.dk/ase/development/calculators.html), the argument "system_changes" contains a "list of what has changed since last calculation. Can be any combination of these six: ‘positions’, ‘numbers’, ‘cell’, ‘pbc’, ‘initial_charges’ and ‘initial_magmoms’."
>>
>> However, I am finding that system_changes always contains all of the six possibilities listed above whether or not they have changed. Is this a known issue, or could I be doing something wrong?
>>
>> Thanks,
>>
>> Ellad Tadmor
>> _______________________________________________
>> ase-users mailing list
>> ase-users at listserv.fysik.dtu.dk
>> https://listserv.fysik.dtu.dk/mailman/listinfo/ase-users
>
>
> _______________________________________________
> ase-users mailing list
> ase-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/ase-users