News for hackers. Stuff that matters.
Re:OpenAI claims Gold-medal performance at IMO 2025 by tlb on Saturday July 19, 2025 @10:03AM (#44615558)
I encourage anyone who thinks these are easy high-school problems to try to solve some. They're published (including this year's) at https://www.imo-official.org/problems.aspx. They make my head spin.
Re:OpenAI claims Gold-medal performance at IMO 2025 by xpressvideoz on Saturday July 19, 2025 @10:33AM (#44615783)
I didn't know there were localized versions of the IMO problems. But now that I think of it, having versions of multiple languages is a must to remove the language barrier from the competitors. I guess having that many language versions (I see ~50 languages?) may make keeping the security of the problems considerably harder?
Re:OpenAI claims Gold-medal performance at IMO 2025 by dmurray on Saturday July 19, 2025 @06:00PM (#44619815)
The problems are chosen by representatives from all the countries. So every country has someone who knows the full exam before the participants get it. Security is on the honour system, but it seems to mostly work.
Re:OpenAI claims Gold-medal performance at IMO 2025 by peteyreplies on Sunday July 20, 2025 @03:34PM (#44628564)
iirc, the IMO system automatically translates the questions into 50 languages, after they are entered in English.
Re:OpenAI claims Gold-medal performance at IMO 2025 by kappi on Saturday July 19, 2025 @03:42PM (#44618680)
It is tough for those who don't practise last years questions. If LLM is trained with last 30 years of IMO competitions, it is very predictable pattern for it to guess the next token.
Re:OpenAI claims Gold-medal performance at IMO 2025 by throw310822 on Saturday July 19, 2025 @03:49PM (#44618735)
I think you're joking, but you never know :)
Re:OpenAI claims Gold-medal performance at IMO 2025 by yongjik on Saturday July 19, 2025 @05:25PM (#44619566)
All the past IMO problems are known to public and contestants practice on them. If solving an IMO problem is the simple matter of "looking at all the past problems and apply the same pattern," you'd expect human contestants to do a lot better.
Re:OpenAI claims Gold-medal performance at IMO 2025 by kappi on Saturday July 19, 2025 @07:07PM (#44620311)
I think you haven't gone thru AMC8, AMC10, AIME competitions. If you are so confident, try giving an unsolved math problem outside the high school math competitions.
Re:OpenAI claims Gold-medal performance at IMO 2025 by saagarjha on Saturday July 19, 2025 @10:33PM (#44621472)
It's still tough, though practice obviously helps.
Re:OpenAI claims Gold-medal performance at IMO 2025 by mauriziocalo on Saturday July 19, 2025 @03:43PM (#44618691)
Related — these videos give a sense of how someone might actually go about thinking through and solving these kinds of problems:

- A 3Blue1Brown video on a particularly nice and unexpectedly difficult IMO problem (2011 IMO, Q2): https://www.youtube.com/watch?v=M64HUIJFTZM

-- And another similar one (though technically Putnam, not IMO): https://www.youtube.com/watch?v=OkmNXy7er84

- Timothy Gowers (Fields Medalist and IMO perfect scorer) solving this year’s IMO problems in “real time”:

-- Q1: https://www.youtube.com/watch?v=1G1nySyVs2w

-- Q4: https://www.youtube.com/watch?v=O-vp4zGzwIs

Re:OpenAI claims Gold-medal performance at IMO 2025 by nmca on Saturday July 19, 2025 @05:06PM (#44619404)
It takes Tim Gowers more than hour and a half to go through q4! (Sure, he could go faster without video. But Tim Gowers! An hour and a half!!)
Re:OpenAI claims Gold-medal performance at IMO 2025 by snewman on Sunday July 20, 2025 @10:11AM (#44625359)
For people who prefer reading to watching videos, I wrote a detailed account of my process for solving one of last year's IMO problems, along with thoughts on how this relates to AI:

https://secondthoughts.ai/p/solving-math-olympiad-problems

Re:OpenAI claims Gold-medal performance at IMO 2025 by bko on Saturday July 19, 2025 @07:47PM (#44620556)
I like watching youtube videos solving these problems. They're deceptively simple. I remember reading one:

x+y=1

xy=1

The incredible thing is the explanation uses almost all reasoning steps that I am familiar with from basic algebra, like factoring, quadratic formula, etc. But it just comes together so beautifully. It gives you the impression that if you thought about it long enough, surely you would have come up with the answer, which is obviously wrong, at least in my case.

https://www.youtube.com/watch?v=csS4BjQuhCc

Re:OpenAI claims Gold-medal performance at IMO 2025 by Tainnor on Saturday July 19, 2025 @08:46PM (#44620886)
This is slightly tedious to do by hand but there isn't really anything interesting going on in that problem - it's just solving a quadratic equation over the complex numbers.
Re:OpenAI claims Gold-medal performance at IMO 2025 by roenxi on Saturday July 19, 2025 @09:00PM (#44620954)
That isn't much of an argument; nothing in math is truly interesting if you take that approach. exp(i\pi)+1=0 could be said to be dis-interesting because it is just rotation on the complex plane. But it is the opposite - it is interesting because it turned out to be rotation on the complex plane but approached from summing infinite series.

Similarly you can say that solving a quadratic over complex numbers is dis-interesting, but it is actually an interesting puzzle because it is trying its best to pretend it isn't a quadratic. In many ways succeeding, it isn't a quadratic - there is no "2" in it.

Re:OpenAI claims Gold-medal performance at IMO 2025 by csomar on Sunday July 20, 2025 @12:01AM (#44621857)
It is not interesting because there is no “real” solution (pun intended).

If you go to the complex plane, you are re-defining the plane. If you redefine the plane, then you can do anything. The puzzle is about confusing the observer who is expecting a solution in a certain dimension.

Re:OpenAI claims Gold-medal performance at IMO 2025 by Tainnor on Sunday July 20, 2025 @07:32PM (#44630392)
It's true that it might be unexpected that there is no real solution. I also wouldn't have intuited that from the problem statement itself.

However, it's not like you have to go out of your way to look for the complex numbers in some creative way. At some point while solving the quadratic equation you'll have to take the root of a negative number. So the only choice is to reach for the complex numbers, your hand is kinda forced.

Re:OpenAI claims Gold-medal performance at IMO 2025 by Tainnor on Saturday July 19, 2025 @10:29PM (#44621442)
It's "not interesting" because no novel insight has to be used in order to solve this. It's immediately obvious how to solve it, just follow the textbook procedure.

This is distinct both from other typical IMO problems that I've seen and from research mathematics which usually do require some amount of creativity.

> exp(i\pi)+1=0

If your definition of "exp(i*theta)" is literally "rotation of the number 1 by theta degrees counterclockwise", then indeed what you quoted is a triviality and contains no nugget of insight (how could it?).

It becomes nontrivial when your definition of "exp" is any of the following:

- The everywhere absolutely convergent power series sum_{i=0}^\infty z^n/n!

- The unique function solving the IVP y'=y, y(0)=1

- The unique holomorphic extension of the real-valued exponential function to the complex numbers

Going from any of these definitions to "exp(i*\pi)+1=0" from scratch requires quite a bit of clever mathematics (such as proving convergence of the various series, comparing terms, deriving the values of sin and cos at pi from their power series representation, etc.). That's definitely not something that a motivated high schooler would be able to derive from scratch.

Re:OpenAI claims Gold-medal performance at IMO 2025 by roenxi on Sunday July 20, 2025 @01:51AM (#44622327)
Those definitions of exp are all immediately obvious and nearly the definition of textbook, every university calculus course covers them. That is the issue with defining interesting as novel - nothing generally known is novel any more. And they don't require any special maths - sum_{i=0}^\infty z^n/n! is literally just multiplication and addition.

The long and short of it is it just isn't possible to tell someone that their problem isn't interesting. Interest isn't an inherent property of an equation, it is the state of mind of the person looking at the equation. And in this case the x+y/xy is a classic interesting puzzle despite (really because of) how well known the solution is.

Re:OpenAI claims Gold-medal performance at IMO 2025 by Tainnor on Sunday July 20, 2025 @03:05AM (#44622653)
> those definitions of exp are all immediately obvious

no

Re:OpenAI claims Gold-medal performance at IMO 2025 by koakuma-chan on Saturday July 19, 2025 @08:05PM (#44620663)
How do those compare to leetcode hard problems?
Re:OpenAI claims Gold-medal performance at IMO 2025 by kenjackson on Saturday July 19, 2025 @08:24PM (#44620767)
Depends on how hard, but the “average hard” leetcode problem is much easier. These will be more like the ACM ICPC level questions, which I’d put at the “hard hard” leetcode level (also this is a collegiate competition rather than high school, but with broader participation).