<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Forward-Mode Differentiation of Maxwell's Equations · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/1908.10507/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/1908.10507/feed.xml" rel="self" type="application/rss+xml" /><item><title>Forward-Mode Differentiation of Maxwell's Equations — HN submission</title><link>https://news.ycombinator.com/item?id=20875167</link><guid isPermaLink="false">urn:hackerpapers:submission:1908.10507:20875167</guid><pubDate>Wed, 04 Sep 2019 08:36:46 GMT</pubDate><description>&lt;p&gt;Simulating Maxwell&amp;#x27;s Equations with Automatic Differentiation&lt;/p&gt;&lt;p&gt;We wrote core Maxwell&amp;#x27;s Equations solvers using the HIPS autograd package, and defined custom primitives for performing backpropagation through electromagnetic simulations. This allows one to effortlessly perform gradient-based optimization of optical devices, as well as plug electromagnetic simulations into a machine learning model. For more details, our github repo: https://github.com/twhughes/ceviche and arxiv preprint: https://arxiv.org/abs/1908.10507&lt;/p&gt;&lt;p&gt;3 points; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/1908.10507/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>physics.optics</category></item></channel></rss>
