Definition
Plain language
A famously hard university-level math competition in the U.S. and Canada.
As stated in the literature
The William Lowell Putnam Mathematical Competition, an undergraduate proof-and-problem contest whose problems serve as a high-difficulty benchmark for automated theorem-proving systems.
Also called: Putnam competition
Why it matters: Its notoriously hard problems make it a demanding yardstick for whether an AI can really do rigorous mathematics.
For example, a Putnam problem might ask you to prove a surprising fact about numbers that has no obvious starting point and stumps most strong math students.
Heard on the show
“… research benchmark — six hundred and seventy-two competition math problems, and not the easy ones, Putnam-level, the kind that humans train for years to attempt — and had an AI produce fully verified, machine-checked …”Episode 117 — How an Open AI System Verified 672 Hard Math Proofs for Under $300