After all the agony in stony places about AIxMath, what have you actually understood? That's the only thing that matters in math. Can you actually give an impromptu talk about what you have understood to an audience of 5, on a whim? If you can't you're not a mathematician. Sorry.
Epic interview | 31:48
Alvaro: How do you feel personally?
Ravi: Some ppl are finding "wait I'm not going to like being a mathematician in this new world." Others saying "this is the most exciting thing ever." I don't know where I am
I think you should ask him again in 1 year
Ravi Vakil is an incredibly busy person (Professor at Stanford University, President of the AMS, member of AGMAI) so I'm incredibly grateful that he made time to sit down with me for this interview. He's such a level-headed, kind, and knowledgeable person, that I really hope many of you take some time to hear what he has to say about the present and future of the mathematical profession, about the Labs, about AGMAI, about education... It was a great conversation and I'm really happy to be able to share it with all. Please share with students!
youtu.be/0trpmJKKpFM?is=pIjJ…
This is not a drill🚨I'm teaming up with @ElliotGlazer to formalize his crazy proof that Z is diophantine in Q
We announced the initiative this week during a talk at the @Caltech@math_hackathon
Our repo is already 4M+ lines of Lean on top of TauCeti, FLT, @axiommathai Stacks
This is not a drill🚨I'm teaming up with @ElliotGlazer to formalize his crazy proof that Z is diophantine in Q
We announced the initiative this week during a talk at the @Caltech@math_hackathon
Our repo is already 4M+ lines of Lean on top of TauCeti, FLT, @axiommathai Stacks
~AI Health Warning~
Upon further auditing, code quality is definitely hit-or-miss
Repo is a slop pit. There is code duplication, vendoring across folders, branches. That's kinda the point at this stage. It will be cleaned up when the job is done
⚠️ PROCEED AT YOUR OWN RISK ⚠️
Over the past week, following Elliot's original paper, I've managed to reduce the main theorem to ~30 results previously known in the literature
WE ARE NOW ACCEPTING OPEN SOURCE CONTRIBUTIONS
Point your agent at this issue. It will know what to do 🤝
github.com/Deicyde/diophanti…
This is not a drill🚨I'm teaming up with @ElliotGlazer to formalize his crazy proof that Z is diophantine in Q
We announced the initiative this week during a talk at the @Caltech@math_hackathon
Our repo is already 4M+ lines of Lean on top of TauCeti, FLT, @axiommathai Stacks
The final effort that is being crowdsourced via this repo is formalization of Version 1 of the manuscript, led by Jack McCarthy (x.com/Jack______M), who has decomposed the remaining-to-formalize part of the proof into 45 statements.
I have a question:
what the hell has LeanFRO and Mathlib Initiative teams been doing with millions $ from XTX? Mathlib exists for a decade now and we have nothing remotely like this useful interactive map tracking the developments?!
Very cool stuff at @andTruthPBC! Way to go!
Your readme claims to "resolves the spectral gap without requiring non-constructive choice axioms" but your comparator.json clearly permits Classical.choice
Btw it wouldn't even be possible to state their challenge.lean unless you first rewrite all of Mathlib to be constructive
#OpenAI’s #Lean4#YangMill repo (`OpenAI-YM-20261028`) NO pos. Mass Gap relies with 100k+ lines of non-constr. Class. choice lattice limits
Since Issues&PRs are locked,
HERE the working Lean 4 proof with #TONE MassGap.lean)+full struct. audit via via j²=1:
github.com/frintroper/TONE-v…
Peter Woit on the Shepherds: "Now that they have released the Lean code, their plan to keep secret the 1147 page document explaining the proof and devote great efforts to understanding it and explicating it themselves makes no sense."
Indeed. This makes no sense!
math.columbia.edu/~woit/word…
lol wtf is this repo setup? You expect people to download a random zip file and run it? Well I'm stupid enough to do that, ya got me
The entire proof is ≤10kb with more axioms than theorems 🤦♂️ But don't worry guys, the proof is "Compiled in Lean 4, Zero Sorry, with MathLib" :))
Peter Woit calls for the Yang-Mills Sheparding Group to "shut themselves down"
He says that their plan of only releasing the Lean code while holding onto OpenAI's natural language proof "makes no sense" in light of informalization efforts like those of @scottnarmstrong