David Savitt

Math department chair at Johns Hopkins University and a director of Canada/USA Mathcamp.

[David Savitt @dsavitt]

@prz_chojecki I see the behavior you mean, but have had a couple of successes where GPT 5.4/5.5 was able to grind something out (essentially combinatorial) in 30-40 prompts.

[David Savitt @dsavitt]

@littmath I think it may have proved something for me a week or two ago but the writing is so impenetrable that I'm starting over and adding "you may introduce new notation but not new terminology" to the prompt. Hoping that doesn't handicap it too much.

[David Savitt @dsavitt]

@NoahJSnyder @EllynBriggs I'm seeing them play Cote D'Ivoire in Philly! Super-excited for that one.

[David Savitt @dsavitt]

@littmath @florianederer Right

[David Savitt @dsavitt]

@littmath @florianederer Hao Huang got some nice prizes and was able to move from Emory to NUS. That seems like a reasonable comparison to me.

[David Savitt @dsavitt]

@NoahJSnyder @dzackgarza I think you could also entertain the possibility that they are testing their models against all sorts of problems, and combinatorial/analytic number theory is where the low hanging fruit has turned out to be.

[David Savitt @dsavitt]

@ericneyman @YafahEdelman No real clue but would definitely find it weird if they were not.

[David Savitt @dsavitt]

@ericneyman @YafahEdelman That seems a touch pessimistic to me, but a better guess than all but one by 2032. Unless Hodge gets knocked off by being false, say.

[David Savitt @dsavitt]

@marcusgnt I mean I'm pretty sure it solved the problem I asked it to solve (will only know for sure when the formalization is done) but this is horrifyingly unreadable. I'll have to try turning the formalization back into natural language to see if that's any better.

[David Savitt @dsavitt]

Well, good luck everybody.

[David Savitt @dsavitt]

@GregHBurnham Might as well associate them with bagels.

[Sapsterr @Sapsterr_]

Adam Wylde from SDPN reported on the Steve Dangle Podcast that he was told by "multiple sources" that Evan Gold was OFFERED the GM position, but was lowballed "Below the $1M, $750K" range.

He also stated that The Canucks "do not have the same R&D, analytics, and computing systems for prospects and current players as most of the league"

Later on he mentioned how the future buyout of Paolo Aquilini may have affected this.

If this is even REMOTELY close to true, then holy crap man. What a joke.

[David Savitt @dsavitt]

@littmath @AcerFur "I must be honest: this does not finish the whole theorem, but it narrows the remaining obstruction. The remaining problem is no longer the [made-up terms] case, it is the [more made-up terms] case."

[David Savitt @dsavitt]

@chaumian This is my experience too except that once, about 30 continuations in, it closed the problem.

[David Savitt @dsavitt]

@ColmezPierre And some of us disappear entirely, but for more pleasant reasons.

[David Savitt @dsavitt]

@tonylfeng @littmath Maybe the cleanness of a result is a reasonably universal value where one could take steps towards making it objective?

[David Savitt @dsavitt]

@littmath @tonylfeng The three of us have already-formed mathematical tastes, and the statements we find interesting will evolve but only slowly. Or maybe what I mean is that statements we once found interesting/appealing will stay so. For someone who's 10 years old today, though....

[Nature is Amazing ☘️ @AMAZlNGNATURE]

Squirrel eating fermented berries

[Breakthrough @brkthroughprize]

Congratulations to Mingjia Zhang for winning a 2026 Maryam Mirzakhani New Frontiers Prize for contributions to the theory of Shimura varieties. https://t.co/hl9WFS5GDA @Princeton

[Breakthrough @brkthroughprize]

Congratulations to Vesselin Dimitrov @Caltech and Yunqing Tang @UCBerkeley on winning the 2026 New Horizons in Mathematics Prize for work in Diophantine geometry, including the proof of the Atkin-Swinnerton-Dyer unbounded denominators conjecture and new irrationality results for special values of Dirichlet L-series.
https://t.co/awmRtAWxGl