Hannes Loeffler

Ok, that's certainly a plan but the practical problem is that this affetcs one single file which in turn means conflicts in parallel development. For the near future this will...

Ok, I have created a branch 'refactor'. Done so far. - dynamic loading of parsers: need to discuss the interface - replaced optparse with argparse - eliminated globals in main()

The interface i.e. inputs and outputs. At the moment P is passed in which is, frankly, a rather "heavy" data structure. It also seems to be (ab?)used to also contain...

I think I have an idea how units should be handled. Essentially, the parsers fill up dhdlt and u_klt and report back what units they are in. The parsers should...

Following the duscussion on https://github.com/pandegroup/openmm/issues/680 it doesn't look to me that the current maintainer is much interested in making this module available in a sensible way (i.e. standalone). Pint has...

I had a quick look into pint and it looks workable. Possible downsides: this approach means replacing all 'raw' values with unit objects. This could mean more memory and also...

I have addressed most of these except for th unit problem. The harder part will now be to separate the current functions into more useful chunks (probably by means of...

My main worry was that this wouldn't work well with the numpy arrays that we use now. A numpy array can only contain certain data types (a "raw" number) and...

Yes, I have started with some light-weight changes in the code and I would think we should use this as a basis for discussion.

Yes, let's got with email to arrange a discussion in October.