Releases: EdwardBerman/shopt
Sync with Zenodo
v3.0.1 .
Simultaneous AJ and JOSS Companion Paper Acceptances
This is the state of the program for the official camera ready acceptances
arxiv-AJ Release
This is the version of the software as it was when we submitted our paper to the preprint arxiv and concurrently to the Astronomical Journal for review.
Paper Revisions
No major software has been developed. Paper has been updated based off of internal review. Contributors JEM and I are happy to be submitting to JOSS at this stage, with acknowledgements to COSMOS-Web collaborators.
Paper Submission Release (JOSS)
With this project nearing submission for peer review, I wanted to release a new version to make it clear what was worked on before and after feedback from the reviewers! New to this version is multi-threading support, easier star catalog input, a new pixel grid fitting mode (PCA), and the skeleton of a planned Chisq mode (this last one would be a good first issue for any new contributors!) We may make a few edits between now and submission, but these will be mostly minor cleanup or documentation. Having gone through the instillation process on a few different computers now, we'd like to address some of the quirks that might be different from machine to machine. Open to feedback if any of our instructions on the README.md are unclear! You may also see edits to the paper.md, paper.bib files that we are submitting to JOSS. Regardless,
Happy ShOpt-ing =] !!
User Friendly Release
Previously, ShOpt.jl was the minimum viable product in terms of output. Since then, a number of improvements have been made. Conjugate Gradient has been replaced by LBFGS everywhere to increase speed. Tunable hyperparameters are more easily tuned in the YAML, including polynomial fits for an arbitrary degree polynomial. The data cleaning process has been more streamlined with an output checkplot to give some visual information as to how and how many stars are being filtered based off of signal to noise (with an additional checkplot planned). More configuration to follow in the YAML.
Perhaps most importantly, there is now an easy way to reconstruct the PSF at an arbitrary location in the sky. In the julia REPL or your code editor, obtain the necessary functions using include("reader.jl"). Then run PolynomialMatrix, degree = read_shopt("path/summary.shopt"); p(u,v, PolynomialMatrix, degree) to reconstruct the PSF at an arbitrary (u,v). I have a separate repository, sigma-eta-shopt-reader with plans to expand this tool to languages outside of Julia, but that is not a top my priority list. I chose sigma-eta because its analogous to "sh" as in shopt and I wanted to make it clear that the repository was different than shopt itself, and greek letters are always fun.
Official First Version
ShOpt is now out of the development phase and into the testing phase. After fitting analytic and pixel grid fits and interpolating across the field of view ShOpt returns appropriate diagnostics plots as well as a fits file titled "summary.shopt". We are now ready to use our existing code to compare ShOpts performance in both fidelity and speed to other software such as PIFF and PSFex and begin exploring new ways to compare them!
Shopt Beta
Shopt has enough functionality in its current state to be useful for lots of Lensers! There are known issues and missing features I plan on implementing (namely PSF Interpolation across the Full Field of View). However, Shopt currently gives Analytic and Pixel Grid Fits using Conjugate Gradient Descent with reparameterization /retraction and an autoencoder model for the Pixel Grid Fits. In this stage the easiest way to get started is to fork or pull the repo and tune it to your specific catalog fits file. Also, please be aware that the software is currently hard coded to only run on the first 25 sources for testing purposes. This is still a pre-release, and so the lack of ease in getting started is something I am aware of and looking forward to streamlining!