logoalt Hacker News

Levent Alpöge claims proof of the Generalized Riemann Hypothesis [pdf]

2041 points • by iamronaldo • today at 2:16 PM • 1385 comments • view on HN

Comments

moezd • today at 6:39 PM

Even when it turns out to be incomplete or wrong, it just shows how much sprawl we have in humanity's collective knowledge. Maybe we are just beginning to coalesce it into a manageable hierarchy, and if nothing else, that's worth spending time and money on, instead of staying in our knowledge silos and pushing forward without the full picture. A pessimistic look would assume that wouldn't be enough, but imagine what could be done once you can start with a clean slate and a clearer learning path.

pjungwir • today at 4:43 PM

Nearly a year ago I asked ChatGPT to show me a proof of the Goldbach Conjecture, just to see what it did. Of course it said no such proof exists. But after a couple turns of pressing, it gave me two or three proposed outlines for a proof, with a several-pages-long research program for how you could attempt to solve the gaps still needed. I couldn't understand any of it, but it was pretty impressive. At that level it probably wasn't nonsense: I bet it was half-copied from some grad student blogs. If you asked it to dig into the details and keep updating the status of the outline, you could probably make a lot progress.

I think AI will become more autonomous as it becomes able (or just willing) to think more abstraction levels deep. Each level can be expressed in language, so it should be able to engage with it. It reminds me of a chess program being able to think many moves deep. For a while grandmasters like Kasparov could still see big-picture things the computer couldn't (e.g. with a locked-up pawn formation and an eventual piece sacrifice to promote a queen, where the kings can run in circles forever). AI seems like that now.

1970-01-01 • today at 3:35 PM

If true, the infinite agent theorem needs to supersede the infinite monkey theorem. Just launch 1,000,000 agents with 10hr windows at one open math problem per day and see what happens. I know OpenAI is sort of trying this already for marketing purposes but I'm thinking a more formal, BOINC approach to problems.

https://en.wikipedia.org/wiki/Infinite_monkey_theorem#Histor...

trostaft • today at 5:01 PM

I really love this writeup, and I think the approach of posting the lean proof and leaving the long process of digestation for future work is a perfectly acceptable approach. It's the difference between closing fields of research (openai) and opening them.

yodsanklai • today at 3:17 PM

> The argument is incredibly long at the moment, and while its formalization is finished and it is routine to produce a huge, unreadable document containing sufficient detail, in my view of my path I have just taken a step. I intend to do my best to understand and impart understanding of this argument and its certain simplifications, to speak about it, to write it properly.

Next step, make the proof intelligible for humans.

➕ show 1 reply
galoisrepresent • today at 4:32 PM

This (https://github.com/tzu-chen/slim) can be used to navigate the lean source and search for key statements without building the whole release, which can take hours.

almostarockstar • today at 3:35 PM

I wonder how specifically he used Claude to do this.

Was it a case of having more dollars to spend by virtue of being an employee and just pointing it in the right direction, or did he provide some insight into the context that allowed it to produce superior results?

It must be the second - but if so, how? Was it as simple as spelling it out in chat and analysing output?

buzzy_hacker • today at 3:16 PM

If true, it seems like Goldbach's conjecture may fall shortly after, see this recent proof conditional on Generalized Riemann Hypothesis [0]. It has other conditions too though.

[0] https://mrc.sdu.edu.cn/english/info/1018/1514.htm

qnleigh • today at 5:41 PM

> The precise timing is horrible. If I could have, I’d have finished this all much earlier. Hell, strictly before this age would have sure been nice.

Is he over 40, and referring to being ineligible for the Fields Medal? It would be strange if you could get a fields medal for work done by an AI, but I suppose that is how many Fields Medals will be won from now on. I hadn't thought of that until just now.

➕ show 2 replies
bagacrap • today at 5:50 PM

Thread for to follow along with verification: https://leanprover.zulipchat.com/#narrow/channel/113488-gene...

(login seems to be required)

cvoss • today at 6:33 PM

> ... asking [Dario] about my ability to announce. We went and figured it out and the fact is that there was a constraint to respect, though I did ask about progressively more bizarre ideas.

Seems kind of cryptic. What is this referring to?

➕ show 2 replies
tensorific • today at 4:07 PM

I've downloaded the Lean formalisation. It's 175,187 Lean files, 17,374,215 lines of Lean code, and comes in at 0.96 GB of code. On my 20 core i7 box, it's projected to take about a day to build...

➕ show 1 reply
omnee • today at 3:36 PM

If proven then we can say with confidence that we now share the world with super intelligence, albeit in certain areas and indeed one that was endowed by humanity. I can only hope that in time the benefits are many and accrued to all of us.

pu_pe • today at 6:10 PM

The write-up is beautiful and poetic, but light on the details of how this was achieved from a practical standpoint. How many human and agentic hours were necessary? Was a breakthrough only possible after Fable 5.2 or another internal model?

andrewla • today at 7:08 PM

Okay, let's get Collatz next! I'm curious about two aspects of it -- one to confirm that the main version, the 3n+1, converges, but I would love to see the more generalized one, namely to show that the same framework with 5n+1 _diverges_ for some inputs, or frankly that any (2k+1)n+1 problem diverges for sufficiently large k.

RenRenJuan • today at 3:34 PM

An unreadable proof is basically none at all. Suggest you have a model produce a concise summary and work edge in from that. Apparently none of these problems has been verified and anybody working with models would be justly skeptical of these helium filled claims which are almost certainly not going to pan out.

Also apparently the Riemann claim is based on it compiling in Lean. This is classic because just the differences between an actual verification of a proof and it compiling in Lean are the standard model failure tropes, lack of judgement, confabulation, etc., to which this adds shortcomings of Lean.

cgearhart • today at 3:11 PM

This is such a wild ride. The conjecture itself is not so hard to understand, but approximately _all_ of the maths here are beyond me. The paper is written up to suggest "we knew which direction to go, but really had no idea how far the journey would be" which certainly makes it seem like the human expert was useful in this process. Perhaps that's just wishful thinking, but it's consistent with what I see elsewhere that AI is having an effect. The biggest effects accrue to the most effective humans. Random folks prompting is better than random folks toiling alone, but experts prompting is a different thing.

abetusk • today at 5:13 PM

> Does [Alpoge's proof] give anything about square roots mod pq (p,q prime)?

> Claude Notice: Fable 5.2's safeguards flagged this message.

Yikes

nomadpenguin • today at 7:14 PM

Does the proof produce novel mathematical concepts or structures? Will AI be able to pose the next GRH?

olwmc • today at 3:53 PM

This is an amazing result but I can't help but think about how much the world has changed since the controversy around the four color theorem's computer assisted proof.

sva_ • today at 3:41 PM

Formalized P!=NP proof by end of year?

ilitirit • today at 6:07 PM

I'm sure people have done it, but I haven't seen any news about it so I guess I'll ask here... has anyone asked thrown Inter-universal Teichmuller Theory at Claude?

FartyMcFarter • today at 5:27 PM

How tricky is it to translate the statement of the Riemann Hypothesis into Lean?

If we're confident in the accuracy of the Lean RH statement (which must be part of this proof), and if Lean is sufficiently bug-free, then we can trust the proof is correct.

➕ show 1 reply
claaams • today at 3:20 PM

I wonder if the money open ai and anthropic have spent just on solving these problems was spent on human mathematicians and resources would we have gotten these solved? Sooner?

➕ show 2 replies
thevalueisai • today at 4:17 PM

My hunch is that stuff like GRH and other problems we have will be childrens' toys in the future and truly novel, advanced math is not yet even within our grasp yet. We are struggling to walk here, we need machines to help us with this. We felt good about ourselves for a while because everyone was equally bad at it, but now reality is hitting us. Hard.

neuwirth • today at 3:22 PM

The shift now needs to be towards statements that can't be posed as a singular Lean statement (or, even, a collection of Lean or similar statements that is understandable by a human). Mathlib/NumberTheory/LSeries/RiemannZeta.lean already had the statement formalized in a way that us mortals can understand. The great open problems of the future will be the ones which cannot so easily be interpretably formalized in the first place, I fear.

➕ show 1 reply
mef51 • today at 4:58 PM

>>The argument is incredibly long at the moment, and while its formalization is finished and it is routine to produce a huge, unreadable document containing sufficient detail

How do we know the Lean formalization is an accurate representation of the incredibly long and unreadable argument?

quantum_state • today at 5:27 PM

We have been conditioned to be impressed. Should we update that conditioning now?

crovona • today at 6:56 PM

Mathematics is seeing its own revolution like physics did in the first half of the 20th century.

singularity2001 • today at 5:12 PM

There are certain math problems where the only guidance has been the prompt "you can do it". Now that we know it's possible, this should be a new benchmark. Probably too expensive for the next few years.

gadders • today at 5:00 PM

Out of interest, how do they use an AI to solve these maths problems? I assume they're not pointing the chat at the wikipedia page and saying "Go solve that for me."

Are they using code to do it? Is it a special maths AI harness>

➕ show 1 reply
Shitty-kitty • today at 5:40 PM

If this is true it would be astounding. Most Physicist's placed Navier–Stokes at the top of the list of what AI could quickly solve and the Riemann Hypothesis last.

hypendev • today at 3:16 PM

I kind of expected this to happen by EoY.

Combining the brightest mathematicians we have with intelligence-in-a-box, it was just a matter of time before the hardest math challenges fall.

Congratulations to Levant, but also to us as a species. We have managed to create machines made out of pure math that contain insurmountable knowledge and are helping us solve the mysteries of the universe.

FartyMcFarter • today at 2:54 PM

At some point I was looking at ways to try and find counter-examples computationally. I'm glad I didn't spend too much time on that...

freakynit • today at 5:21 PM

Software development and mathematics were two professions people used to say were the ones most stable for future .... I guess they couldn't have been more wrong :(

Hasz • today at 3:53 PM

Man this is so fucking cool, I cannot wait to see what other problems fall in the future (and what new ones this opens).

250, 000k pages is bonkers, I hope they will be able to condense it into something a bit more grokable.

I think this is a pretty clear indication that the hybrid approach (e.g amazing mathematician and great models with a huge token budget) can accomplish absolutely incredible things.

Math is the verifiable domain; I wonder if this ratio still holds in less verifiable domains as well, or it shifts more AI/more human (or how it changes in future!)

unknown-unknown • today at 5:42 PM

For example, a Field Medal-level Claude, if made available to the general public, could potentially enable people to design and develop something as technologically complex as SpaceX's Starship in a matter of months rather than years. What took hundreds of brilliant engineers at SpaceX years to accomplish could potentially be achieved in a fraction of the time, if the current pace of AI advancement continues.

Math is becoming the measure of intelligence.

me551ah • today at 5:06 PM

If the laws of AI scaling continue to be true, does this mean that in a year or two we would be seeing frequent revolutions in mathematics and physics?

zkmon • today at 8:21 PM

What's the direct impact of having this proof? Is anything going change or any changes needed in the real world things?

amai • today at 5:19 PM

The timeline given by Claude at the end of the note seems to jump back and forth in time. Are these just typos or what is going on there?

➕ show 1 reply
kdavis • today at 3:34 PM

Has anyone set up a GitHub repo with the Lean code a comparator and verified this?

➕ show 1 reply
ElijahLynn • today at 6:41 PM

What are some of the practical applications of this discovery?

➕ show 1 reply
gniv • today at 4:42 PM

Whoa. If they also solve P=NP I will really be scared.

torrent • today at 5:06 PM

The most important thing now is to find ways to visualize LEAN proves so that we can understand them faster by factors.

jiggawatts • today at 8:50 PM

Many people will be asking about the relevance of this, real world applications, etc. I think that's missing the point.

This is like climbing to the peak of Mt Everest.

There is no gold there, no riches, the view is grand but not that unique. Nobody will build a multi-billion-dollar business empire based on Everest tourism.

But it is a benchmark of capability! The locals that had lived there for millennia failed to do it. Getting there required technology, technique, and even some specific developments like electrically heated boots. The methods used are akin to going to space, with protective clothing (vaguely) comparable to a space suit, including oxygen tanks!

It's not about the goal, it's about the ability to get there, which is a mere milestone in the exponential improvement:

In 1953 a human stood on the peak of Everest.

In 1961 a human orbited the Earth.

In 1969 a human walked on the Moon.

In 2026 an AI solved the most famous unsolved maths problem.

In 203x ... ?

brandonpelfrey • today at 4:09 PM

Looking forward to the release of the approach and also someone to "distill" it down to the most intuitive and still-correct version.

fi-le • today at 6:06 PM

Seems like a great point to stop AI development at, internationally

wohoef • today at 3:38 PM

I really hope the major AI labs steer away from mathematics as a means to prove capability of their models, and start putting their tokens to real-world use cases.

aenis • today at 4:22 PM

So... its like paying technical debt in math, considering a lot of stuff assumed it was true, without a proof -- until today?

🔗 View 50 more comments