Kaminer25.txt

File contents

Computer Algebra and Applications Seminar, Summer Semester 2025

Title: AI for Mathematics: From Symbolic Intelligence to Scientific Discovery 

Speaker: Prof. Dr. Prof. Ido Kaminer, Technion – Israel Institute of Technology


Time and Location: 
Fr, June 27, 2025, 10 a.m.; 
RISC Seminarroom, Castle of Hagenberg

Artificial Intelligence is beginning to transform mathematics—not only as a tool for calculation or automation but as a collaborative partner in discovery. In this talk, I will present an overview of the rapidly emerging field of AI for Math, with a special emphasis on our group's work with the Ramanujan Machine—an initiative that combines symbolic computation, machine learning, and distributed computing for discovery in mathematics. Inspired by experimental physics, we developed algorithms that autonomously generate conjectures about fundamental constants such as π, e, and values of the Riemann zeta function. These efforts have resulted in thousands of newly discovered formulas, surpassing the cumulative human discoveries over centuries. I will discuss these findings and their implications to unify mathematical knowledge. I will conclude with a vision for the next generation of AI-driven symbolic tools supporting mathematicians and scientists in solving long-standing open problems.

G. Raayoni, et al., The Ramanujan Machine: Automatically Generated Conjectures on Fundamental Constants, Nature 590, 67–73 (2021)

O. Razon, et al., Automated Search for Conjectures on Mathematical Constants using Analysis of Integer Sequences, ICML 202, 28809-28842 (2023)

R. Elimelech, et al., Algorithm-assisted discovery of an intrinsic order among mathematical constants, PNAS 121, e2321440121 (2024)

M. Shalyt†, U.Seligmann†, et al., Unsupervised Discovery of Formulas for Mathematical Constants, NeurIPS 37, 113156 (2024)

Beit Halachmi and I. Kaminer, The Ramanujan Library - Automated Discovery on the Hypergraph of Integer Relations, upcoming ICLR (2025); arXiv:2412.12361