Repeat-Until-Success circuits with fixed-point oblivious amplitude amplification

Abstract

Certain quantum operations can be built more efficiently through a procedure known as Repeat-Until-Success. Differently from other non-deterministic quantum operations, this procedure provides a classical flag which certifies the success or failure of the procedure and, in the latter case, a recovery step allows the restoration of the quantum state to its original condition. The procedure can then be repeated until success is achieved. After success is certified, the RUS procedure can be equated to a coherent gate. However, this is not the case when the operation needs to be conditioned on the state of other qubits, possibly being in a superposition state. In this situation, the final operation depends on the failure and success history and introduces a "distortion" that, even after the final success, depends on the past outcomes. We quantify the distortion and show that it can be reduced by increasing the probability of success towards unity. While this can be achieved via oblivious amplitude amplification when the initial success probability is known, we propose the use of fixed-point oblivious amplitude amplification to reduce this unwanted distortions below any given threshold even without knowing the initial success probability.