From 21c2065e2d16499e4f9bcc1ac1f43fac29bd82bd Mon Sep 17 00:00:00 2001 From: Jammy2211 Date: Thu, 23 Jul 2026 22:26:32 +0100 Subject: [PATCH] docs: rectangular mesh users must cite Enzi et al. 2026 (RTU grids) Adds a "Rectangular Mesh" section to docs/general/citations.md and the Enzi2026 bibtex entry (arXiv:2606.30620, MNRAS submitted) to files/citations.bib: the adaptive rectangular meshes implement the paper's ray-guided transformed uniform (RTU) grid formulation and must be cited when used; RectangularUniform is exempt, and the regularization differs from the paper's Gaussian-process prior. Part of PyAutoLabs/PyAutoArray#402. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_01XK2ZZagTywsBZC4CZL66eB --- docs/general/citations.md | 26 ++++++++++++++++++++++++++ files/citations.bib | 11 +++++++++++ 2 files changed, 37 insertions(+) diff --git a/docs/general/citations.md b/docs/general/citations.md index 3367f6ebd..2616c7fe8 100644 --- a/docs/general/citations.md +++ b/docs/general/citations.md @@ -72,6 +72,32 @@ also be cited: The package is available at . +## Rectangular Mesh (Pixelized Reconstructions) + +If you reconstruct a galaxy using the adaptive rectangular meshes (`RectangularAdaptDensity` or +`RectangularAdaptImage`), you **must cite** the following paper under citation +key `Enzi2026`. The mesh's adaptive coordinate transform implements the ray-guided transformed uniform (RTU) +grid formulation this paper introduces, which is what makes the pixelization adaptive and fully +auto-differentiable: + +```bibtex +@article{Enzi2026, + author = {Enzi, Wolfgang J. R. and Krawczyk, Coleman M. and Li, Tian and Collett, Thomas E.}, + title = {Gaussian processes on ray-guided transformed uniform grids for fast, flexible, and auto-differentiable adaptive source reconstruction in lens modelling}, + journal = {MNRAS, submitted}, + eprint = {2606.30620}, + archivePrefix = {arXiv}, + primaryClass = {astro-ph.GA}, + year = {2026}, + url = {https://arxiv.org/abs/2606.30620}, +} +``` + +Note that **PyAutoGalaxy** pairs the RTU grid with its own regularization schemes (e.g. `reg.Constant`, +`reg.Adapt`) rather than the Gaussian-process prior used in the paper, so quantitative results are not +directly comparable between the two implementations. The `RectangularUniform` mesh performs no RTU transform, +so this citation is not required when only the uniform mesh is used. + ## Dynesty If you used the nested sampling algorithm Dynesty, please follow the citation instructions [on the dynesty readthedocs](https://dynesty.readthedocs.io/en/latest/references.html). diff --git a/files/citations.bib b/files/citations.bib index d7bc603ab..8318f3a56 100644 --- a/files/citations.bib +++ b/files/citations.bib @@ -367,3 +367,14 @@ @article{dynesty4 publisher={Zenodo}, author={Sergey Koposov and Josh Speagle and Kyle Barbary and Gregory Ashton and Ed Bennett and Johannes Buchner and Carl Scheffler and Ben Cook and Colm Talbot and James Guillochon and et al.}, year={2023}, month={Jun} } + +@article{Enzi2026, + author = {Enzi, Wolfgang J. R. and Krawczyk, Coleman M. and Li, Tian and Collett, Thomas E.}, + title = {Gaussian processes on ray-guided transformed uniform grids for fast, flexible, and auto-differentiable adaptive source reconstruction in lens modelling}, + journal = {MNRAS, submitted}, + eprint = {2606.30620}, + archivePrefix = {arXiv}, + primaryClass = {astro-ph.GA}, + year = {2026}, + url = {https://arxiv.org/abs/2606.30620}, +}