[QE-developers] Notes for developers' meeting (was Re: Developers' meeting?)

[Nicola Marzari]

On 21/12/2018 18:12, Paolo Giannozzi wrote:
> Makes sense to me as well. I have updated the tentative agenda 
> accordingly (same link as before: 
> http://www.quantum-espresso.org/news-events/news/qe-developers-meeting-2019).



Dear All,


I've collected a few of the desiderata from here and from other groups,
about some of the more pressing needs for QE (maybe something to keep in 
mind also for MaX, given that the funding is meant to support a user
community?).

As you can see from the list below, they have a lot to do with using
the code(s) in reliable ways, minimizing the external changes from one 
release to the other.

Also, note that it's very difficult for groups outside Trieste
to have a sense of what will/will not happen to the codes, and this
doesn't help the broader developer community (gitlab has improved things
a lot, but there is still the sense that changes and developments aren't
coming from bottom-up requests, or even a broader discussion - having 
regular meetings of the QEF foundation, or pre-planned developers' 
meetings where people can connect in remore, or public hackatons on 
certain goals, would all help - especially if adveertised/planned
early).

Here is the list:

-----
1) XML output - it would be great to have some public reference document
or roadmap (e.g. on gitlab) that could summarize

  - current status of XML output, open issues, and which will be
    addressed in the following release

  - which PP tools do/do not support the new XML

2) roadmap for pw and for other codes: what are the planned
developments, especially for what concerns the input and output
(this is e.g. essential for us, to make sure we do not lose
functionality of all the workflows developed)

3) hdf5 file format (with documented schema): what is the status?
it would be great to have it for all binary data, in particular for
wavefunctions, density and potential, encoding all auxiliary data needed
for interoperability and reproducibility (e.g. g-vector ordering, cell
vectors, etc..).

4) k-point generation (and symmetries): standardize generation, for
example by adopting spglib, which is increasingly becoming a de facto
standard and has wide functionality.

5) pseudopotentials (less of an issue): which format to use upf, json,
xml ? upf is a poor approximation of xml.
-----

Thanks a lot,

				nicola



----------------------------------------------------------------------
Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL
Director, National Centre for Competence in Research NCCR MARVEL, EPFL
http://theossrv1.epfl.ch/Main/Contact http://nccr-marvel.ch/en/project