Claude's Cycles: How Generative AI Shocked Donald Knuth

About this episode

This episode explores how generative AI, specifically Claude Opus 4.6, solved a complex mathematical problem that shocked Donald Knuth.

Legendary computer scientist Donald Knuth was recently in for a "shock" when an open mathematical problem he had been working on for several weeks was successfully solved by Anthropic's Claude Opus 4.6. In this episode, we dive into the fascinating story behind "Claude's Cycles," exploring how this generative AI showcased a dramatic advance in automatic deduction and creative problem-solving. The complex problem, intended for a future volume of The Art of Computer Programming , involved finding a general decomposition of a specific digraph's arcs into three directed Hamiltonian m^3-cycles. After Knuth solved it for m=3, his friend Filip Stappers challenged Claude to find a generalized solution. Guided by strict instructions to document its progress, Claude worked through 31 distinct algorithmic "explorations". Moving from simple depth-first search and simulated annealing to "serpentine patterns" and fiber decomposition, the AI eventually realized it needed "pure math" to discover a working solution for all odd values of m. Join us as we recount Claude's impressive 60-minute analytical journey, discuss the 760 perfectly valid…

People in this episode

Host: Dayan Ruben

Topics covered

• Generative AI
• Mathematics
• Problem Solving
• Computer Science
• Algorithmic Exploration
• Donald Knuth

Keywords

• Generative AI
• Claude Opus 4.6
• Donald Knuth
• mathematical problem
• algorithmic exploration
• Hamiltonian cycles
• computer science

Mentioned in this episode

Organizations: Anthropic

Products: Claude Opus 4.6

Books & works: The Art of Computer Programming, Claude's Cycles