Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Hans Dembinski
    @HDembinski
    I got confused by your original example, where you computed the sum of weighted values. That's a profile in ROOT language, so I explained how to make a profile. But you actually want just a sum of weights and then you need to use the WeightedSum storage as others explained. In the latter case, there are no values to be weighted, just a sum of weights is collected and the variance of that sum is given by sum of weights squared, which this storage computes for you.
    Henry Schreiner
    @henryiii
    @matthewfeickert The work on uproot is here, and the issue is scikit-hep/boost-histogram#423. I realized on inspection of Jim’s work that I’ve really muddled some of the values for different storages, so I’ll be tweaking and clarifing after the AS meeting today.
    Henry Schreiner
    @henryiii
    I’m probably about to push to make Boost-Histogram 1.0, probably sometime this month. I’ve discussed numbering and Python versions a bit, but I’m beginning to think the following plan might be best: I’ll make all the changes for 1.0, and make a release (say, 0.12). Then I’ll strip Python 2 and Python 3.5 support, and deprecated features, and make an otherwise API compatible 1.0 release. That way, going forward we don’t have to maintain support for Python 2 for most new features, and reduce the large number of wheels we have to push for a release, but the “final” API of 1.0 is available in 0.x, so if we do pick up major Python 2 users, like experiment stacks, they are not using an outdated API (well, unless they ignore FutureWarnings, that is ;) ). If we really had to, we could even release 0.13 along with 1.1 (would need serious and desperate users, but the option would be open) @HDembinski I think this is closer to your original idea, as well; thoughts? @jpivarski & @eduardo-rodrigues ?
    Hans Dembinski
    @HDembinski
    Sounds good to me. We settled the Plottable protocol and now also how slices work, was there anything else that was controversial for the interface?
    Henry Schreiner
    @henryiii
    I don’t think so. The list here is pretty complete: https://github.com/scikit-hep/boost-histogram/milestone/7 In fact, indexing a list of categories will likely wait till Boost.Histogram has support, so that will move to post 1.0. Addable weight view already is implemented, AFAIK.
    Actually, there is one more unification that I have that you might like, but I have to play with it a bit to make sure it works.
    So expect one more issue or PR. But it doesn’t really change the interface, but unify some things behind the scenes and enable a little bit more than before.
    Hans Dembinski
    @HDembinski
    ok
    If the implementation can be simplified or made more uniform, great
    Eduardo Rodrigues
    @eduardo-rodrigues
    Hi @henryiii, I think your proposal sounds very reasonable.
    Henry Schreiner
    @henryiii
    I’m giving GitHub Discussions a try! Feel free to open discussions at https://github.com/scikit-hep/boost-histogram/discussions
    Hans Dembinski
    @HDembinski
    @henryiii Could you perhaps try it out with another project? In terms of community management, it is best to point people to one place for feedback.
    By adding https://github.com/scikit-hep/boost-histogram/discussions you are splitting the community and I don't want that.
    Alexander Held
    @alexander-held
    What is that one place you describe, gitter? To me gitter does not seem suited for focused discussions - the history is hard to read, even more so when one comes back after a month and tries to find a topic discussed previously.
    Henry Schreiner
    @henryiii

    I don’t think it’s really a replacement for gitter; it’s more of a replacement for a) stack overflow, and b) opening issues for help instead of a bug. It’s also all in one place - GitHub, so less fragmented. It’s really just a bit more orginisation on issues, with some minor added features. An issue closes - you aren’t supposed to care about the issue anymore when it’s fixed. But if it was opened as a discussion, or moved to a discussion, in the Q&A category, then it just gets an answer, but remains easy to find, so it becomes less likely to become a recurring quesiton. The “showcase” topic looks great - people can post thier most intersting and exciting examples of boost-histogram.

    And while it’s new to me, other members of our comunity have been trying them out already (the PyHF repo has had them on for over a month, for example).

    This Gitter is mostly a few dedicated poeple (mostly devs), and while gitter has threaded convos, nobody really uses them. And there’s no reasonable search. It’s rather useful for announcements, but it’s taking off for the type of help discussions I would have hoped to see. Pybind11 is a great example of a vibrant Gitter community - but there, the same set of questions tend to get asked over and over.
    Alexander Held
    @alexander-held
    Oh, is it possible to move an issue over to make it a discussion instead? That is very nice.
    Henry Schreiner
    @henryiii
    Yep, a ton of Awkward issues got moved over. And a few Uproot ones. You can even move them by label. :) Also, you can move them to a different repo if they were opened in yours but really were for a library higher up the chain.
    Alexander Held
    @alexander-held
    This is really useful, I like that a lot.
    Henry Schreiner
    @henryiii
    Imagine someone opens a feature request (called Ideas) discussion in boost-histogram. It’s for something we don’t think is basic/universal enough for boost-histogram, so we move it to Hist. :)
    Hans Dembinski
    @HDembinski

    What is that one place you describe, gitter? To me gitter does not seem suited for focused discussions - the history is hard to read, even more so when one comes back after a month and tries to find a topic discussed previously.

    Yes, I was referring to Gitter as the place to communicate with users.

    Alexander Held
    @alexander-held
    I see usefulness in both Gitter and GH Discussions:
    • Gitter for issues that require some back-and-fourth for a short amount of time and are likely very user-specific (e.g.: my compilation fails and I don't know how to fix it), if that gets lost in the history it is probably ok since other users may not have the same question
    • Discussions for questions that lots of users may have (and then in the future find there answered, like StackOverflow), or questions that take a long time to answer (to have one consistent thread there instead of being spread out in Gitter between other topics). I also think the "Show and tell" category could be quite nice, and a message like that would disappear quickly in Gitter. And I personally like the idea of everything in one place (repository) a lot.
    Henry Schreiner
    @henryiii
    I’d see gitter as more for announcements and for developer discussions (scroll back in the history, it’s pretty much the same 4-5 people with maybe a couple of user questions), while most users usually open an issue (there are more question issues than questions in gitter, I believe). Discussions gives them a safer, nicer place to open these things that hopefully won’t be as large of a barier, and will be more discoverable after they are answered. But they should be seen mostly as better orginization and a few extra features like threads and answers on top of issues.
    Andrzej Novak
    @andrzejnovak
    I haven't had the chance to use discussions much in mplhep yet, pretty quiet these days, but I am a big supporter. It's a great place for "not quite issue"'s. Gitter still works when one has a question that's more random, but it might as well not keep history past few days
    Henry Schreiner
    @henryiii
    With gitter you also tend to “subscribe” to the whole thing, rather than just quickly ask a single question and follow just that.
    Hans Dembinski
    @HDembinski
    Fine, what I am not interested in is proliferation of places that I have to check for user feedback.
    Jim Pivarski
    @jpivarski
    I set them all up with email notifications. That way, I don't have to explicitly remember to check.
    Hans Dembinski
    @HDembinski
    @jpivarski That's good advice, thanks
    Michael Eliachevitch
    @meliache
    Hi. Thanks for the cool package, I am trying to use it for my new plots. Is there any canoncial/convenient way to do stacked histogram plots, similar to plt.histogram(stacked=True)? So far I just used that to plot my boost histograms:
    image.png
    Michael Eliachevitch
    @meliache

    My histogram is having a Regular and a strCategory axis and I want to show the cumulative sum of the categories, but with the different components stacked on top of each other.

    I found out I can just use np.cumsum(axis=<category axis>) to calculate the cumulative sum and create a new histogram from that. In had also managed to draw a stacked histogram by plotting multiple plt.bar with different bottom=... arguments taken from the cumulative sums. But I dislike that visualy, because I want to have a black outline for my components, but no outlines for the individual bins, as seen in the screenshot.

    Andrzej Novak
    @andrzejnovak
    @meliache you could check if the mplhep.histplot works for you it should handle stacking reasonably
    though admittedly i have not tested is super thoroughly with bh, but as especially if exported to numpy arrays it should be trivial to plot stuff like this
    Henry Schreiner
    @henryiii
    This is going to be something I’ll make sure works in the near future - propagating the Protocol so we can easily plot weighted histograms is my main priority currently.
    (And the correct way to do it will be the function in mplhep, or possibly mpl’s new stairs plot)
    Andrzej Novak
    @andrzejnovak
    @henryiii I have it on the list to implement it in mplhep, but I am waiting form mpl 3.4 with stairs
    Will probably bump minor version for mplhep with it
    Alexander Held
    @alexander-held

    Hi, I used to be able to modify histogram contents like this

    import boost_histogram as bh
    import numpy as np
    
    bins = [0, 1, 2]
    hist = bh.Histogram(bh.axis.Variable(bins), storage=bh.storage.Weight())
    yields = [3, 4]
    var = [0.1, 0.2]
    hist[...] = np.stack([yields, var], axis=-1)
    
    hist.view().value /= 2

    That worked until boost-histogram version 0.11.1. In the current master it does not anymore, I think scikit-hep/boost-histogram#475 changes the behavior:

    Traceback (most recent call last):
      File "test.py", line 10, in <module>
        hist.view().value /= 2
      File "[...]/boost_histogram/_internal/view.py", line 57, in fset
        self[name] = value
      File "[...]/boost_histogram/_internal/view.py", line 49, in __setitem__
        raise ValueError("Needs matching ndarray or n+1 dim array")
    ValueError: Needs matching ndarray or n+1 dim array

    Is there another way to achieve this now?

    Henry Schreiner
    @henryiii
    That’s supposed to work, it’s just been generalized to work in more places - must be a bug!
    Alexander Held
    @alexander-held
    Thanks! I'll open it in an issue then.
    Henry Schreiner
    @henryiii
    Perfect, thanks, I’ll get right on it.
    Henry Schreiner
    @henryiii
    Ahh, the breakage happens with the final line, not the earlier part. Good, faith in my testing has been restored. I was sure I tested somethign like the hist[…] line. Okay, interesting; this code wasn’t supposed to have triggered here. Working on fixing now.
    Technically, this is still “unimplemented” behavior - see scikit-hep/boost-histogram#276 - but if it used to happen to work, it should continue to. And it was planned anyway. Sorry, thought you were using hist.view() /= 2.
    Henry Schreiner
    @henryiii
    Found the bug. 🤦
    Henry Schreiner
    @henryiii
    Static typing would have caught this. (Even just by forcing me to be aware of the possible arguments to __setitem__)
    Alexander Held
    @alexander-held
    Thanks a lot Henry!
    Henry Schreiner
    @henryiii
    Boost-histogram 0.12 is out. It’s mostly a bugfix release, fixing several important bugs, with a few minor things. The current develop branch has PlottableProtocol, which needs a little more time before being released, but feel free to try it out if you like the bleeding edge!
    Hans Dembinski
    @HDembinski
    :-D
    Hans Dembinski
    @HDembinski
    Hi all, there is currently a push in scipy to include Boost as a dependency. This is initiated by the wish to replace current implementations in scipy.special and scipy.stats with those in Boost.Math. Someone else (not me!) then brought up that one could then also base histograms and the scipy.stats.binned_statistics on Boost.Histogram.
    I then told them about boost-histogram and the possible performance increases. Not anything I could work on anytime soon, but exciting to see interest in Scipy about this sort of thing.