reword

paper.md

@@ -51,7 +51,7 @@ Empirical PSF characterization tools like Point Spread Function Extractor (PSFEx
 # State of the Field
 The JWST captures images at high resolution and at wavelengths of light that have been previously unexplored [@Gardner_2006]. With these images we are seeing farther into the early universe than we ever have before. The difficulties of producing good PSF models for the JWST are emblematic of a larger problem: Our data sets are getting bigger and existing software was not built to scale. That is to say, the advancements in software are falling behind the advances in instrumentation. Not only does `ShOpt` produces PSF models for JWST NIRCam images, it also sets the precedent for designing software that scales.  
 
-There are several existing empirical PSF fitters, in addition to a forward model of the JWST PSFs developed by STScI [@Jarvis_2020 ; @2011ASPC; @2014SPIE ; @2012SPIE]. We describe them here and draw attention to their strengths and weaknesses to further motivate the development of `ShOpt.jl`. As described in the statement of need, `PSFex` was one of the first precise and general purpose tools used for empirical PSF fitting. However, the Dark Energy Survey collaboration reported small but noticeable discrepancies between the sizes of `PSFex` models and the sizes of observed stars. They also reported ripple-like patterns in the spatial variation of star-PSF residuals across the field of view [@Jarvis_2020], which they attributed to the astrometric distortion solutions for the Dark Energy Camera.
+A number of JWST PSF libraries already exist, using both empirical and forward-modeling approaches. We describe them here and draw attention to their strengths and weaknesses to further motivate the development of `ShOpt.jl`. As described in the statement of need, `PSFex` was one of the first precise and general purpose tools used for empirical PSF fitting. However, the Dark Energy Survey collaboration reported small but noticeable discrepancies between the sizes of `PSFex` models and the sizes of observed stars. They also reported ripple-like patterns in the spatial variation of star-PSF residuals across the field of view [@Jarvis_2020], which they attributed to the astrometric distortion solutions for the Dark Energy Camera.
 
 These findings motivated the Dark Energy Survey's development of `PIFF` (Point Spread Functions in the Full Field of View). `PIFF` works in sky coordinates on the focal plane, as opposed to image pixel coordinates used in `PSFex`, which minimized the ripple patterns in the star-PSF residuals and the PSF model size bias. (Based on the DES findings, `ShOpt` also works directly in sky coordinates.) `PIFF` is written in Python, a language with a large infrastructure for astronomical data analysis, for example Astropy [@2022ApJ] and Galsim [@rowe2015galsim]. The choice of language makes `PIFF` software more accessible to programmers in the astrophysics community than `PSFex`, which was first written in `C` in 2007 and much less approachable for a community of open source developers. One of the motivations of `ShOpt` was to write astrophysics specific software in `Julia`, because `Julia` provides a good balance of readability and speed with its high-level functional paradigm and just-in-time compiler [@Stanitzki_2021]. Julia ranks behind Python, IDL, Matlab, and Fortran in full-text mentions in astronomical literature [@TheAstropyCollaboration_2022]. We are optimistic that `ShOpt` will demonstrate that Julia is an appealing choice for programming in astronomy despite its low adoption to date. There is also recent work on using `PSFr` for PSF reconstructions that has been applied to JWST data [@2022ApJ939L28D; @2022ApJ938L17Y; @2022ApJ938L14M; @2023ApJ942L27S]. Similar to `ShOpt`, `PSFr` begins its PSF modeling by stacking stars to form an initial guess. However, rather than using polynomial interpolation to address spatial variations, `PSFr` employs an iterative process of shifting the pixels in its PSF model. This process continues until the model can adequately represent all of the stars in the catalog. Finally, the PSF fitter `STARRED` [@michalewicz2023starred] has been shown to produce PSF models competitive with `PSFr` and `PSFex`. Like `ShOpt`, `STARRED` puts an emphasis on computational efficiency and uses the JAX package to achieve just-in-time compilation. There is future work to be done to benchmarking all of these empirical approaches on JWST NIRCam data.