Three Mathematicians Prove Talagrand's 30-Year Geometry Conjecture With a $2,000 Prize Still Unclaimed
Updated
Updated · Scientific American · May 19
Three Mathematicians Prove Talagrand's 30-Year Geometry Conjecture With a $2,000 Prize Still Unclaimed
2 articles · Updated · Scientific American · May 19
A new proof posted last week resolves Michel Talagrand’s 1995 convexity conjecture, showing that simple convex structures inevitably emerge even in vast, chaotic point sets across extremely high dimensions.
Antoine Song unlocked the problem by recasting it in probability theory, turning a long-stalled geometry question into one that quickly became tractable.
Song and student Dongming Hua briefly used ChatGPT to understand a missing step, but co-author Stefan Tudose produced a broader, more insightful proof, and the paper ultimately did not rely on the AI’s argument.
Talagrand, the 2024 Abel Prize winner, said he had doubted the conjecture was even true; he had long offered $2,000 for a solution and now expects the result to shape future work on high-dimensional data and machine learning.
A decades-old math proof unifies two abstract worlds. How will this breakthrough reshape the AI we use every day?
A 30-year math mystery is solved, but the AI's contribution was cut. Is human intuition still supreme in discovery?
A genius doubted his own theory for 30 years. What does this reveal about the nature of scientific breakthroughs?
Talagrand’s Convexity Conjecture Solved After 30 Years: Universal Number Three, AI’s Assist, and Far-Reaching Impact
Overview
A team led by Antoine Song, Dongming (Merrick) Hua, and Stefan Tudose has solved Talagrand's Convexity Conjecture, a problem that challenged mathematicians for 30 years. Their breakthrough shows that even in chaotic, high-dimensional spaces, order can emerge through a universal process involving Minkowski sums. By shifting the problem into probability theory, they proved that any subgaussian random vector can be written as the sum of three Gaussian vectors, explaining the universal number of sums needed. The proof also marks a milestone by explicitly citing AI for assistance in navigating complex literature, hinting at the growing role of AI in future mathematical discoveries.