-
-
Notifications
You must be signed in to change notification settings - Fork 1.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Added theta parameter to GaussianKernel2D #3605 #4687
Conversation
@aniketk21 - yes, you definitely need to update the docs and add a changelog entry. Note that this should go under the changelog section for v1.2, as it's a new feature. Also, it looks to me like the discussion in #3605 settled on expecting the arguments to be exactly the same as Otherwise this looks pretty good to me. |
Oh, one other worry, though: it's good that this is mostly backwards compatible, but there's a subtle change that the first argument is no longer called |
@eteq I'll add the docs and changelog. |
@aniketk21 Thanks for working on this! Yes, definitely add the docs and and a changelog entry. I agree with @eteq to repeat the parameter names from I think the additional Optional: |
…el. Added a deprecation warning for stddev. #3605
@eteq @adonath - just a general question about these kinds of changes (to |
@astrofrog I'm not really sure what the best option is. I know that Sherpa avoids this kind of problem, by using a different parametrization of the models, that includes an asymmetry parameter I think in general it's not a big problem, if there are multiple parametrizations of a model available. The only issue I see is possible code duplication and a crowded model name space. The later could be avoid with a clever nomenclature of the models. E.g. let the different parametrizations of the model still start with the same model name such as Another option could be introduce a short-cut for direct coupling of parameters, so one could do:
But this still leaves the issue which fitter can be used. |
Hmm, I think I'm slightly in favor of having multiple models i.e., I would say it's reasonable to have one be the subclass of the other - i.e. |
Oh and I'm -1 to saying "only |
I also think it would make sense to have more model classes, both in e.g. the trapezoid case and gaussian case. |
@eteq The Adding more models with different parametrizations and explicit names seems to be a good option to me as well. |
Ohh, gotcha, yeah that makes sense, although it might be good to avoid |
The changes do not look very controversial. Was there a reason why this is stalled? @aniketk21 , do you still wish to wrap this up? You need to:
|
Hi humans 👋 - this pull request hasn't had any new commits for approximately 1 year, 6 months. I plan to close this in a month if the pull request doesn't have any new commits by then. In lieu of a stalled pull request, please close this and open an issue instead to revisit in the future. Maintainers may also choose to add If you believe I commented on this issue incorrectly, please report this here. |
This looks potentially useful but abandoned. I might pick this up and finish it but it is pretty far down my to-do list right now. Anyone else interested feel free to take over. |
Same issue here, it was on my radar, but fallen to the bottom of the todo. (I remember going back to this during pyastro, but then got side-tracked and only got to the point of rebasing it.) |
An attempt to continue this work is at #6748 |
@astrofrog @embray Please review.
Should I change the documentation for GaussianKernel2D and add an entry to CHANGES.rst?