<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>A Topological Version of Schaefer's Dichotomy Theorem · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2307.03446/</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/2307.03446/feed.xml" rel="self" type="application/rss+xml" /><item><title>A Topological Version of Schaefer's Dichotomy Theorem — mentioned by nyrikki</title><link>https://news.ycombinator.com/item?id=39471253</link><guid isPermaLink="false">urn:hackerpapers:mention:2307.03446:39471253</guid><pubDate>Thu, 22 Feb 2024 18:42:37 GMT</pubDate><description>&lt;p&gt;The known lower bound for self attention is n^(2-epislon) PH, or the polynomial hierarchy is the collapse I was talking about, but note I was talking about global and not individual or families of instances. Schaefer&amp;#x27;s Dichotomy Theorem shows that HORNSAT is solvable, given those constraints. I found this paper useful as it reframes it as topology which is helpful to build it intuitions, at least me. https://arxiv.org/abs/2307.03446 Of the proposed linear self attention like solutions I have seen, they tend to approach it more from what can be reduced the clustering problem. A lens for that…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2307.03446/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.CC</category></item></channel></rss>
