Jared Duker Lichtman

534 posts

Jared Duker Lichtman banner
Jared Duker Lichtman

Jared Duker Lichtman

@jdlichtman

number theorist, szegő assistant professor of mathematics at @Stanford

Katılım Temmuz 2024
0 Takip Edilen2.6K Takipçiler
Sabitlenmiş Tweet
Jared Duker Lichtman
Jared Duker Lichtman@jdlichtman·
It is remarkable what a small group of aligned individuals can achieve. Huge congrats to the team for cooking on this one.
Math, Inc.@mathematics_inc

Today, at the @DARPA expMath kickoff, we launched 𝗢𝗽𝗲𝗻𝗚𝗮𝘂𝘀𝘀, an open source and state of the art autoformalization agent harness for developers and practitioners to accelerate progress at the frontier. It is stronger, faster, and more cost-efficient than off-the-shelf alternatives. On FormalQualBench, running with a 4-hour timeout, it beats @HarmonicMath's Aristotle agent with no time limit. Users of OpenGauss can interact with it as much or as little as they want, can easily manage many subagents working in parallel, and can extend / modify / introspect OpenGauss because it is permissively open-source. OpenGauss was developed in close collaboration with maintainers of leading open-source AI tooling for Lean. Read the report and try it out:

English
0
1
14
1.2K
Jared Duker Lichtman retweetledi
Jonathan Gorard
Jonathan Gorard@getjonwithit·
Lots still to figure out about how we integrate with the thought processes of human mathematicians, to better capture the pedagogical purposes of proofs, not merely the epistemological ones. But I'm excited to be a part of this journey with the rest of the @mathematics_inc team.
Math, Inc.@mathematics_inc

Today, at the @DARPA expMath kickoff, we launched 𝗢𝗽𝗲𝗻𝗚𝗮𝘂𝘀𝘀, an open source and state of the art autoformalization agent harness for developers and practitioners to accelerate progress at the frontier. It is stronger, faster, and more cost-efficient than off-the-shelf alternatives. On FormalQualBench, running with a 4-hour timeout, it beats @HarmonicMath's Aristotle agent with no time limit. Users of OpenGauss can interact with it as much or as little as they want, can easily manage many subagents working in parallel, and can extend / modify / introspect OpenGauss because it is permissively open-source. OpenGauss was developed in close collaboration with maintainers of leading open-source AI tooling for Lean. Read the report and try it out:

English
4
12
130
10.2K
Lawrence Paulson
Lawrence Paulson@LawrPaulson·
Formal Verification of Axiom-Free Gödelian Ontological Argument and Trinity Necessity Proof in Isabelle HOL Kim, Yong-Dock 📧 We reconstruct metaphysical and ontological arguments within a framework of computational formalization and machine verification. isa-afp.org/entries/Axiom_…
English
3
6
44
2K
Helen Toner
Helen Toner@hlntnr·
As someone publicly associated with EA, it was very funny to watch SV tech world try to use it to destroy my reputation in DC - because the main association policy people had with EA, if they'd heard of it at all, was thinking EAs are shills for big tech 😂
Ben Brody@BenBrodyDC

It continues to confuse me that the Valley thinks its strong feelings about EA resonate in DC. I go to every congressional vote to talk to lawmakers about tech. Not one has ever brought up EA

English
6
6
196
14.5K
Chris
Chris@chatgpt21·
@jdlichtman I don’t think I would do well in a math class in Stanford
English
2
0
4
1K
Chris
Chris@chatgpt21·
I was grinding chatGPT 5.4 on hard calculus for hours and it didn’t hallucinate once on any of the calculations, - rounded a couple times which had me question it.. It’s so interesting how the world just quickly moves on. I know Sam mentioned this phenomenon a lot but as someone who truly loves math it still shocks me how consistent it can be as someone who used to deal with countless math hallucinations just a couple of years ago.
English
30
16
324
17.5K
scoot
scoot@scootykins·
@punished_daniel I do not trust strivers. The future belongs to double digit screen time slackers
English
3
3
74
2.4K
Daniel Yang
Daniel Yang@punished_daniel·
There's 2 emergent cultures coming out of the zoomer generation and it's striver wasians vs permanent underclass tiktok wireheads
a16z@a16z

"There's something about this culture of young people coming up where they're not afraid of hard work. They're not afraid to pop a Zyn and work at the factory all day." Why @KTmBoyle is bullish on Zoomers: "The best quote that summarizes why I'm so bullish on the Zoomers is Alysa Liu after winning her gold. She said, 'I love to struggle. It makes me feel alive.'" "It's the opposite of the morose theater kid vibes that we got from the millennial generation, where everything was very different in how they operated." "Like Jack Hughes—they get their teeth knocked out, they come back and say, 'It's not even a question. Of course I got my teeth knocked out. It's hockey.'" "And that means we're seeing totally different companies than we saw out of the Facebook diaspora—which was very much the Harvard dorm room—I like to work on my computer, I like to build apps. It's a totally different style of founder." "The next generation is so patriotic and bullish on the American project. I think this generation cares a lot about the country. And it shocked us. @davidu and I talk about this all the time—for some of these young people, they were not born on September 11th. They have no recollection of the things that the millennials remember, or anyone older than us remembers, but they care about the country." "They look up to people like @elonmusk, to people like Alex Karp. They look up to people who've been doing the hard thing for 20, 30 years and they want to do it too." "It's a different generation of founder that we've had the privilege of seeing very, very early on. I think the rest of the country is going to define tech and Silicon Valley by these people for the next 10–20 years." From @nypost

English
14
12
688
41.2K
agniv
agniv@agniv_s·
yeah, it's all coming together now
agniv tweet media
English
2
0
9
532
Yiqing Xu
Yiqing Xu@xuyiqing·
Call me naive, but I’m optimistic about what this AI tsunami 🌊 means for social science and the liberal arts. Once workflows stabilize and humans adapt to rapidly advancing machine capabilities, we may return to what matters---capturing **importance** in complex society---with much higher efficiency and accuracy. I think it’s a bright world ahead ✨ but also a more competitive one, especially for junior scholars.
English
9
10
126
8.8K
FleetingBits
FleetingBits@fleetingbits·
garry tan and gary marcus live in a kind of superposition on two sides of the metr graph; each makes the other look almost reasonable while the metr timelines increase exponentially between them
FleetingBits tweet media
English
2
0
24
2.1K
sarah
sarah@atheorist·
Today my colleagues and I released a paper on the formalization of d=4 Quantum Field Theory in Lean. We believe that formal verification has the potential not only to radically reshape how mathematical research is conducted, but also to transform research in theoretical physics.
English
35
70
637
45.4K
Jared Duker Lichtman
Jared Duker Lichtman@jdlichtman·
Indeed! One of the main objectives of the project with Viazovska is make the proof efficient, modular, and useful for future works. In fact, Gauss first proved Sphere Packing in 500,000+ lines of Lean, based on 20+ cited papers it ingested. Then after the proof was discovered, Gauss realized that much in the supporting papers weren't directly needed, and so compressed the proof down to ~200,000 lines. I very much view the 2 tasks as separate: finding proofs, then compressing them. And in practice, the latter seems to become easier once the former is done!
English
0
0
12
341
Bartosz Naskręcki
Bartosz Naskręcki@nasqret·
I believe you. Just wondering how this knowledge will be organized in the long run. It would be great to make some preliminary decisions. But possibly it does not matter that much once things are formalized. You can indefinitely recode any formally compiled code to the shape that you like. It sounds a bit outlandish to think about mathematics in such a situation.
English
1
0
6
973
Jared Duker Lichtman
Jared Duker Lichtman@jdlichtman·
I proposed to Buzzard taking on complete FLT back in the fall after we did the Prime Number Theorem. He laughed me out of the room. After Sphere Packing, he's not laughing anymore.
Daniel Litt@littmath

@unsorsodicorda Yes, I’m curious to see what kind of progress will be made. Buzzard had an ongoing project (pre-autoformalization) which aimed to formalize large parts of the argument “by hand” in 5 years, which gives a sense of the magnitude.

English
3
10
91
11.7K