Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Hans Dembinski
    @HDembinski
    Filling the smaller one takes negligible time
    Luke Kreczko
    @kreczko
    sweet
    Hans Dembinski
    @HDembinski
    Sorry to tell you folks that the collaboration with the main HepMC3 author is falling apart.
    There are now three Python packages for HepMC3 on PyPI, https://pypi.org/project/HepMC3/ and https://pypi.org/project/pyhepmc-ng/ and https://pypi.org/project/pyhepmc. It looks like there is still an agreement to merge pyhepmc-ng into pyhepmc, so there will be two competing packages.
    The HepMC3 bindings are based on pybind11, but autogenerated with binder. They do not offer a Pythonic interface, just a plain mapping of the C++ interface.
    Eduardo Rodrigues
    @eduardo-rodrigues
    Thanks for the info, and sad to hear it. It makes no sense given your previous discussions with the HepMC3 people! Weird reaction from their side!
    Hans Dembinski
    @HDembinski
    I got some private feedback that it is probably not my fault this went south...
    Anyway, my plan is still to continue pyhepmc for the time being.
    Hans Dembinski
    @HDembinski
    Oh no, I just accidentally uninstalled the WIP App for Scikit-HEP on Github
    Will try to revert this...
    Looks like it can be reinstalled here: https://github.com/apps/wip
    But I will wait for some feedback before messing up more
    Henry Schreiner
    @henryiii
    Welcome into the WIP app uninstallation club! :)
    We can actually add it for select Scikit-HEP repositories. It’s not ideal (since it has to be enabled one-by-one, rather than opt-out, which would be better). I have never liked that it adds a check to the list of checks; “draft” mode in GitHub is better but cannot be enabled after the fact. When working with someone else’s PR (like a summer of code student), it can be useful to add a WIP after the fact.
    Henry Schreiner
    @henryiii
    Put it back for now.
    If yet another admin uninstalls it, we need to set up a proper club. :)
    Hans Dembinski
    @HDembinski
    Thank you very much and sorry for that ...
    Henry Schreiner
    @henryiii
    You were not the first, I’ve done it as well. Other people in Scikit-HEP do use it, though. The ability to add WIP back to a PR is nice.
    Hans Dembinski
    @HDembinski
    This looks very interesting, https://github.com/google/jax
    Has anyone played with this and can share experience?
    In my understanding, you can auto-generate the gradient for a score function with this.
    It looks like normal numpy code, so I don't have to learn tensorflow to use it
    It also jit-compiles the Python code and can run it on the GPU
    Jonas Eschle
    @mayou36
    Yes, I did, It is in fact very nice and works quite well for some things (yes, gradient, gpu etc) but is not yet production ready. E.g. using it with scipy does often not work, also you cannot do certain things syntax wise (AFAIR). So it's a really nice tool, mostly for existing code. But since TF and PyTorch are very similar, I personally prefer one of those, they seem more explicit to me and are built from ground up for this kind of things.
    Hans Dembinski
    @HDembinski
    i see, thanks @mayou36
    Jonas Eschle
    @mayou36
    pyhf (AFAIK) tries to use it as an (additional) backend, on top of their numpy implementation, maybe they can comment more
    Matthieu Marinangeli
    @marinang
    FYI the package scikit-stats has just been renamed hepstats :D https://github.com/scikit-hep/hepstats
    Henry Schreiner
    @henryiii
    Better. :)
    Jim Pivarski
    @jpivarski
    @marinang I'm glad: giving the right impression about the scope will save complaints later. I know we took a Scikit name for Scikit-HEP, but that one seemed appropriate because the HEP community is a well-defined scope. Some have complained that "HEP" doesn't describe neutrino experiments and direct detection dark matter searches—I agree; I think that "particle physics" is a better-defined group of related interests—but there isn't a nice acronym for "particle physics." I hope the neutron magnetic moment guys forgive us for using "HEP" everywhere!
    Eduardo Rodrigues
    @eduardo-rodrigues
    Yep, Scikit-PP would not be so recognisable.
    Indeed we've always said "...aim of providing Particle Physics at large..." :-)
    on the website..
    Matthew Feickert
    @matthewfeickert
    As many of you know, Giordon, Lukas, and I are happy to announce that pyhf has now joined Scikit-HEP. Our new home is over at https://github.com/scikit-hep/pyhf :)
    Hans Dembinski
    @HDembinski
    Cool!
    Matthew Feickert
    @matthewfeickert
    Hi All. Can all the project maintainers please add the #scikit-hep topic tag to their GitHub repos? I think it would be quite nice to have this topic become more discoverable on GitHub: https://github.com/topics/scikit-hep
    Matthieu Marinangeli
    @marinang
    Done !
    Matthew Feickert
    @matthewfeickert
    :)
    Eduardo Rodrigues
    @eduardo-rodrigues
    Good idea. I've just added a whole bunch of tags.
    Matthew Feickert
    @matthewfeickert
    Great! Thanks Eduardo.
    I'm asking around to see if I can get someone to submit a PR for us to github/explore to get the tag recognized. We can't submit one as one of the checks is "I have no affiliation with the project I am suggesting (as a maintainer, creator, contractor, or employee)."
    Henry Schreiner
    @henryiii
    Affiliated packages are now listed properly on the main page for the Scikit-HEP website :)
    Hans Dembinski
    @HDembinski
    Cool
    Matthew Feickert
    @matthewfeickert

    I was thinking it would be nice if we designed a Scikit-HEP badge that would link to the project webpage that all projects would have in their READMEs. While this might first seem redundant as they are all under the Scikit-HEP GitHub org, as the READMEs end up getting put up on PyPI during publishing it could be a useful way to get a curious user from PyPI to the project quickly.

    Thoughts? Also, as I rely on GitHub Issues to get anything done if there are no complaints from @henryiii I'll open an Issue on the website repo to further discuss this.

    Henry Schreiner
    @henryiii
    I'm in favor of it.
    Jim Pivarski
    @jpivarski
    Me, too.
    Matthew Feickert
    @matthewfeickert
    great
    Matthew Feickert
    @matthewfeickert
    Made an Issue for further targeted discussion: scikit-hep/scikit-hep.github.io#36
    Eduardo Rodrigues
    @eduardo-rodrigues
    Good idea.
    Hans Dembinski
    @HDembinski
    There seems to be a clear consensus already, but I also think this is great.
    Henry Schreiner
    @henryiii

    Here's what I'm thinking (also posted in the issue):

    Scikit-HEP

    Scikit-HEP