Parallel repetition with a threshold in quantum interactive proofs
In this note, we show that O(log (1/ϵ)) rounds of parallel repetition with a threshold suffice to reduce completeness and soundness error to ϵ for single-prover quantum interactive proof systems. This improves on a previous O(log (1/ϵ) loglog (1/ϵ)) bound from Hornby (2018), while also simplifying its proof. A key element in our proof is a concentration bound from Impagliazzo and Kabanets (2010).
READ FULL TEXT