You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Turns out doing a PhD takes a lot of time and effort and it turns out my PhD has nothing to do with nested sampling, so it's been hard to find time to devote to this project. I would be very happy if someone is eager to take over this package with their own improvements and ideas. In the mean time, I want to stop feeling guilty about not checking my GitHub notifications because I spread myself too thin with open-source projects, so I just want to be clear that I cannot be a reliable maintainer of this package currently. I can manage straightforward compathelper PRs and version bumps, but that's about it right now.
There are a couple of thoughts I have about the current state of this package-
Integrator rewrite + JOSS paper
So I wrote a whole JOSS paper about this package that is pretty much ready to go #78 . I've been stalled on it because the integrator in this package does not seem to have the same statistical performance as the state integrator in dynesty #80 . I've spent lots of time trying to figure out any differences in the code between the two packages, but I haven't found the culprit yet. It takes me a lot of time to make progress on this front because I have to spend ~a day relearning the nested sampling math and rereading the dynesty and NestedSamplers.jl code to figure out what's happening.
If someone wants to take charge of that PR it would be a huge step for this package, I think that would be a reasonable place to stamp a v1 and submit that JOSS paper.
Ellipsoidal fitting
This package is one of my earlier Julia packages, so some of the code in the ellipsoid fitting routines could be better. I had thought of reworking these routines with the idea that each "Bound" is really just a distribution that I am fitting. In theory, this could simplify the code in this package by out-sourcing parts of the random number generation and fitting to other packages (e.g., adopting Distributions.jl and sub-typing AbstractBounds under a distribution type).
The text was updated successfully, but these errors were encountered:
thank you for reaching out @Astro-mh! I'll post some status updates and what I can collect of my thoughts over in that thread for you sometime this week.
Hi everyone,
Turns out doing a PhD takes a lot of time and effort and it turns out my PhD has nothing to do with nested sampling, so it's been hard to find time to devote to this project. I would be very happy if someone is eager to take over this package with their own improvements and ideas. In the mean time, I want to stop feeling guilty about not checking my GitHub notifications because I spread myself too thin with open-source projects, so I just want to be clear that I cannot be a reliable maintainer of this package currently. I can manage straightforward compathelper PRs and version bumps, but that's about it right now.
There are a couple of thoughts I have about the current state of this package-
So I wrote a whole JOSS paper about this package that is pretty much ready to go #78 . I've been stalled on it because the integrator in this package does not seem to have the same statistical performance as the state integrator in dynesty #80 . I've spent lots of time trying to figure out any differences in the code between the two packages, but I haven't found the culprit yet. It takes me a lot of time to make progress on this front because I have to spend ~a day relearning the nested sampling math and rereading the dynesty and NestedSamplers.jl code to figure out what's happening.
If someone wants to take charge of that PR it would be a huge step for this package, I think that would be a reasonable place to stamp a
v1
and submit that JOSS paper.This package is one of my earlier Julia packages, so some of the code in the ellipsoid fitting routines could be better. I had thought of reworking these routines with the idea that each "Bound" is really just a distribution that I am fitting. In theory, this could simplify the code in this package by out-sourcing parts of the random number generation and fitting to other packages (e.g., adopting Distributions.jl and sub-typing
AbstractBounds
under a distribution type).The text was updated successfully, but these errors were encountered: