OpenAI reports optimal-order mixing of the Thorp shuffle

For the Thorp shuffle on 2^d cards, the full permutation law after 1600d complete shuffles tends to uniform in total variation as d grows, uniformly over initial decks; a support bound requires at least 2d−O(1) shuffles, giving mixing time Θ(d).

Curated contributions 1

Understand the problem ↓

Understanding the problem

Read the background

For the Thorp shuffle on 2^d cards, the full permutation law after 1600d complete shuffles tends to uniform in total variation as d grows, uniformly over initial decks; a support bound requires at least 2d−O(1) shuffles, giving mixing time Θ(d).

Why it matters

If correct, establishes the optimal logarithmic order for full-deck mixing on dyadic deck sizes.

What counts as progress

Does the Thorp shuffle mix its entire dyadic deck in Θ(log deck size) shuffles?

Website timeline

Publication dates and editorial changes.