[ase-users] Calculator interface proposal

Jens Jørgen Mortensen jensj at fysik.dtu.dk
Fri Feb 15 11:39:59 CET 2013


Hi!

I think the proposal is almost there now.  There is also an 
implementation of the base-classes that has been tested on the EMT, 
NWChem, Abinit and GPAW calculators.

https://wiki.fysik.dtu.dk/ase/development/proposals/calculators.html
https://trac.fysik.dtu.dk/projects/ase/browser/branches/aep1/ase/calculators/calculator.py

What is still not quite there is the way we handle restarting old 
calculations and how we specify names of files.  The proposal and 
implementation currently works like this:

   calc = ABC(label=..., iomode='rw', output=None)

The meaning of label can be "restart from" or "write to" depending on 
iomode ('r', 'w' or 'rw'). The output parameter can be used to write to 
a different file than the one you are restarting from. This may be a bit 
confusing - I don't know what you guys think? What happens if the input 
file is missing or corrupt?

An alternative that I think is easier to understand is this:

   ABC(restart=None, output=None, ignore_bad_or_missing_input=False)

I think this should work for everyone.  A restart would go like this: 
ABC('abc') and fail if the file is bad or missing.  A fresh calculation 
can be done with ABC(output='abc') and so on ...

A shorter name for ignore_bad_or_missing_input would be nice!

Please tell me what you think.  I've been staring at this for quite some 
time, so if you can think outside the box and come up with something 
better then that would be great.

Jens Jørgen





More information about the ase-users mailing list