A Tensor Network View of Kitaev's Lower Bound for Strong Coin Flipping
summary
Recasts Kitaev’s lower bound for strong coin flipping as a tensor‑network duality game, showing Alice’s and Bob’s strategy sets become convex cones of positive operators and their inner product equals the protocol’s outcome; proves the product of optimal cheating probabilities is ≥ 1/2 so one party can bias to at least 1/√2 ≈ 0.707. Uses tensor networks, contractions, operator convexity and scaling of dual feasible points to reproduce the bound; constructs the network picture successfully but leaves explicit best‑response construction open.
AI-written summary · about