@comohology @HarmonicMath Can you share your project ID? Is it OK if I take a look?
English
Tudor Achim
225 posts
Tudor Achim
@tachim
CEO @HarmonicMath. We're hiring! https://t.co/JjakO5nV9a
Menlo Park, CA Katılım Ekim 2007
180 Takip Edilen2.9K Takipçiler
.@HarmonicMath's Aristotle fails an elementary geometry puzzle, says the answer is the opposite corner.
Prompt: "On the surface of a 1x1x2 box, what's the furthest point away from the corner in terms of surface distance?"
English
QME
@tachim @HarmonicMath @axiommathai @mathematics_inc @CarinaLHong @KenOno691 Aristotle theorem prover from @HarmonicMath worked overnight and is still running. Currently at 59% progress proving the Bilateral Collapse Theorem. It was at 49% when first checked at 8:23 a.m.
#LeanProver #OpenScience #RiemannHypothesis #ComputationalMathematics
English
🔨 Lean 4 proof in progress - Bilateral Collapse Theorem:
(a·P1 + b·Q1)·(b·P1 + c·Q1) = −2·b·(a+c)·e₀
For any a, b, c - the sedenion product is forced entirely into the scalar channel. All 15 vector components identically zero.
@HarmonicMath working it.
#Sedenions #LeanProver
English
@aztecsungod @HarmonicMath @axiommathai @mathematics_inc @CarinaLHong @KenOno691 sure -- just run it on aristotle aristotle.harmonic.fun
English
@HarmonicMath @axiommathai @mathematics_inc @CarinaLHong @KenOno691 @HarmonicMath, @axiommathai, @mathematics_inc
Anyone up for a challenge and capable of closing this?
#LeanProver #MathChallenge #BilateralCollapseTheorem #Sedenions
English
This is a huge upgrade -- Aristotle can handle much larger and more complex projects than before. Stay tuned for what's coming next! 🚢🚢🚢
Harmonic@HarmonicMath
🦾Meet Aristotle Agent, the world’s first autonomous mathematician — live and currently free of charge. We designed Aristotle Agent to solve and formalize the world’s most challenging mathematical research problems. It is now: ☑️#1 in Formal Math: We’re the #1 formal math model according to ProofBench, by @ValsAI, ahead of the closest competitor by 15%. Aristotle Agent can autonomously prove/formalize for up to 24 hrs without human intervention. ☑️Fully Agentic: Give it an English problem and it will prove/formalize from scratch, or it can work and edit files directly inside your Lean project / repository. ☑️Github-ready: Aristotle agent produces repo-quality code; project leads are increasingly merging Aristotle-drafted PRs with no modifications. Now live across both web, CLI, and API. 🔥
English
Tudor Achim retweetledi
1/8 CompPoly is a library of machine-checked mathematics for concrete polynomial computation over rings and finite fields—the algebra behind modern cryptographic proof systems. @HarmonicMath Aristotle has recently made significant contributions to help close Phase 1 of the roadmap. @vladtenev @tachim
English
Tudor Achim retweetledi
"Lean in my view is the best programming language ever created." 😍
Great interview with @tachim and @vladtenev on @CogRev_Podcast!
🔗youtube.com/watch?v=-01-Mc…
YouTube
English
It’s rare to see a model blow out the Pareto frontier like this — huge congrats to @StefanoErmon and team!
Stefano Ermon@StefanoErmon
Mercury 2 is live 🚀🚀 The world’s first reasoning diffusion LLM, delivering 5x faster performance than leading speed-optimized LLMs. Watching the team turn years of research into a real product never gets old, and I’m incredibly proud of what we’ve built. We’re just getting started on what diffusion can do for language.
English
Daniel also has helped me massively improve debilitating post viral / Long COVID symptoms over the last month.
I believe he has the workings of an entirely new framework through which we can view immune and autonomic nervous system health that can massively improve these difficult to treat and brutal conditions.
Ian Goodfellow@goodfellow_ian
I'd like to thank @daniel_rossett for his help in my recovery from the POTS version of Long COVID. Daniel was key in bringing me back from highly disabled and suffering to being able to do what I want to again. This X account is mostly focused on ML / AI. From that point of view, many of you know that in December 2024, I wasn't able to do the test of time award talk at NeurIPS, even by video call. Daniel started working with me in March 2025. By April, I started to have days of no POTS symptoms, by June I was off all heart rate lowering medications, by September I was back to work. I'm back to full exercise, running, lifting weights, mountain biking, and have even done things I hadn't done before I got sick, like riding Whistler Mountain Bike Park. I'm now getting the word out to help Daniel build a company that will bring this approach to more people.
English
Tudor Achim retweetledi
V12 is now live for open beta. It can:
- Find valuable bugs
- Generate working, runnable PoC
- Generate patch and test the PoC against it
In our testing during audits at Zellic, Zenith, and Code4rena we've been consistently impressed.
Best of all: it's free. (Don't abuse it!)
pashov@pashov
@claudeai Impressive. Very nice. Now do this, but for smart contracts
English
Fun discussion with @vladtenev and @labenz
The Cognitive Revolution Podcast@CogRev_Podcast
@vladtenev & @tachim, co-founders of @HarmonicMath , join @labenz on @CogRev_Podcast to discuss building Mathematical Superintelligence — and why formally verified AI may be the only path to AI we can actually trust. They cover: Mathematics IS reasoning — making it the ideal domain for AI to go superhuman first How Aristotle won IMO Gold using Lean, Monte Carlo Tree Search & lemma-guessing — while massively outpunching its weight vs. the big labs Why the formal vs. informal debate is settled — no human can read a 5,000-page AI-generated proof How Lean replaces prestige-based peer review with computational certificates — the GitHub moment for math Why hallucinations are a feature, not a bug — entropy is what lets the system explore paths no human has tried The 2030 vision: multiple competing grand unified theories of physics, all self-consistent — waiting only on collider experiments to tell them apart
English
I remember talking to @Leonard41111588 in 2023 as we were getting Harmonic started -- really proud that we're able to support the FRO in a real way today.
Harmonic@HarmonicMath
Today we're donating $300k to @leanprover as the inaugural sponsor! We believe the future of mathematical reasoning lies in formal verification. Our model, Aristotle, uses Lean to eliminate errors and verify results. We're thrilled to support the tools and people that make safe, accurate Mathematical Superintelligence possible.
English
RamaX unlocks rapid and scalable validation of protein designs, which has been the key bottleneck for protein generative modeling ever since @namrata_anand2 's original breakthrough on diffusion models. Now, you can use it downstream of SOTA models in one click!
Namrata Anand@namrata_anand2
Announcing a unified pipeline for end-to-end protein binder design and experimental validation at scale. 🤖🤝🏾🧬 DiffuseSandbox ➡️ RamaX
English
Tudor Achim retweetledi
I'm giving a talk about automatic formal verification of computer hardware using @HarmonicMath's Aristotle system at the University of Cambridge at their CS department on Thursday 26 February 2026 at 2PM in room FW26. The talk is open to the public.
talks.cam.ac.uk/talk/index/243…
English
Tudor Achim retweetledi
Thanks for the thought-provoking piece.
My main critique is that you are overemphasizing flashy but low probability events like “left-handed bacteria,” while merely giving lip service to the risk of extreme economic concentration of power, which is very real and materializing as we speak.
Anthropic is reportedly raising funds at a $350B valuation, and the wealth created thus far has been concentrated into a few hundred (perhaps more like dozens) high net worth individuals / institutions. It’s looking increasingly likely to me that none of the leading AI labs will IPO until they reach valuations in the trillions, at which point retail investors will finally be able to get shares. In order for retail to get a 100x return on these investments, which was achievable for Apple, Microsoft, Amazon, and Google, the valuations of the AI labs will need to reach hundreds of trillions of dollars, meaning it’s likely too late for a more equitable redistribution of wealth.
Simply put, you are currently exacerbating the problem. The consequences of this are that voters may take matters into their own hands and push for either or both 1) more aggressive / nonsensical forms of redistribution — the CA Founders’ Tax is just the beginning or 2) a drastic knee-capping of the AI industry in America, which make the CCP dominance scenario more likely.
The solution is to enable retail ownership now, increasing the number of Americans with economic exposure to Anthropic and other AI labs from hundreds of people to millions.
Dario Amodei@DarioAmodei
The Adolescence of Technology: an essay on the risks posed by powerful AI to national security, economies and democracy—and how we can defend against them: darioamodei.com/essay/the-adol…
English
@hkhanna Is thinking 6, 12, 24, 48 months into the future that hard?
More ICE drama => polling impacts => mid terms => impacts agenda completion / non-completion => Vance / Newsom => etc.
English
Ugh. And now parents have another high profile instance of “don’t put yourself in this position” with their kids.
Peter Kazanjy@Kazanjy
If you're a parent, you have two very important video examples of what not to do to discuss with your children. First is the bar fire in Switzerland, and how to act and not watch an indoor fire. Second is the Minneapolis ICE shooting, and how not to put yourself in a situation where you're confronting armed men in a high stakes situation. Discuss this with your children today so they remember it 10, 20, 30 years in the future.
English
@kylekabasares Huh that's pretty promising. It means you hit your budget. Run it a few more times and see if it finishes -- if it does then you're golden
English
As per @tachim's suggestion I did use Aristotle to try and build off of what GPT-5.2 Pro came up with and it produced...something? after 12+ hours of running.
If someone wants to look at it and see if anything of interest is there, I just opened the .lean file in Google Docs which you can access here
docs.google.com/document/d/139…
English
Seeing that throwing open math problems into SoTA LLMs is the thing nowadays, I got curious this morning and I asked GPT-5.2 Pro to try and solve Erdős problem 295 (currently open, I just literally just clicked "Random Open" on the erdosproblem site).
For full transparency, I'm not a research mathematician and honestly have no clue whether or not whether the 53 minutes GPT-5.2 Pro spent on it actually produced something valuable. I specifically asked it to state at the end whether or not it actually solved the problem (spoiler: it claimed it did not), and it did end up claiming a "potential novelty" for its approach, but again, I have no way of verifying if this is accurate or not.
I'm not going to be able to spend a lot of time on this, but I thought I'd at least share the chat so people with better mathematical skill than myself can evaluate or even build upon what GPT 5.2 Pro produced if it made any progress on this problem at all.
Regardless, I'm just amazed we live in a day and age where one can literally prompt a SoTA LLM an open problem, eat lunch (literally what I did), and possibly have an answer, or a new perspective on the problem. It feels like a new era of "math crowdsourcing" where non-experts can use these tools, see what the LLMs return, and collaborate with professional mathematicians to see if any progress was made. The new challenge for the pros will be to filter and distinguish genuinely new results from the hallucinations and nonsense. Still, if SoTA LLMs could solve open problems at even a low (1-5%) success rate, who knows what that could lead to?
Link to Erdos 295: erdosproblems.com/295
Link to GPT 5.2 Pro Chat: chatgpt.com/share/69728d23…
English
Mathematical Superintelligence is near, and we're stoked to support the researchers helping us get there. Apply now!
Harmonic@HarmonicMath
We’re excited to announce $1,000,000 of sponsorships directly to students and researchers to encourage further exploration using AI and formal verification. More details 👇
English