Any CS student can understand this problem, because everyone learned what regular languages and regular expressions are. It's very easy to understand, but very hard to solve.

One OpenAI result is that multiplying two n*n matrices can be done in O(n^2.25) time. It's obvious that it can be done in O(n^3) time, and people have been pushing down the "3", but progress has…

Now this is very exciting

今早 OpenAI 干了件挺疯的事：往 GitHub 上一次性扔了 722 篇数学论文（372 组结果），全是一个内部从没发布过的模型写的。 我去把仓库翻了一遍，是真的。预印本 PDF、Lean 形式化证明、连模型的推理过程摘要都放出来了。选题也够野：黎曼 zeta 函数的零点自由区、CM 阿贝尔簇上的 Hodge 猜想，这种级别的题目都在里面。 但先别急着喊“数学家失业了”。OpenAI…

OpenAI's model claims to break computing's speed limits. Both come from today's 722-paper drop. • Matrix multiplication exponent: 9/4. The record sat near 2.37. • Integer multiplication: below n log…

Normalizable finite energy Bound States in BFSS with N=2? What the heck?

OpenAI made their big dump!

OpenAI: The list chromatic number of the K_t-minor-free graph is at most O(t) "Linear (list) Hadwiger Conjecture"

the maths and physics slopavalanche is here

here is the src :) > initial commit 57 min ago : yeah it's happening ....