Qubit-efficient exponential suppression of errors

TL;DR

This paper introduces REQUEST, a qubit-efficient quantum error suppression method that achieves exponential error reduction using fewer qubits by leveraging active resets and a space/time trade-off, outperforming existing methods under realistic noise conditions.

Contribution

REQUEST adapts exponential error suppression to fewer qubits using active resets, enabling practical implementation on near-term quantum devices with limited qubits.

Findings

REQUEST reproduces exponential error suppression similar to virtual distillation.

REQUEST outperforms virtual distillation with fewer than 3N+1 qubits.

The required number of shots remains manageable for hundreds of qubits.

Abstract

Achieving a practical advantage with near-term quantum computers hinges on having effective methods to suppress errors. Recent breakthroughs have introduced methods capable of exponentially suppressing errors by preparing multiple noisy copies of a state and virtually distilling a more purified version. Here we present an alternative method, the Resource-Efficient Quantum Error Suppression Technique (REQUEST), that adapts this breakthrough to much fewer qubits by making use of active qubit resets, a feature now available on commercial platforms. Our approach exploits a space/time trade-off to achieve a similar error reduction using only $2N+1$ qubits as opposed to $MN+1$ qubits, for $M$ copies of an $N$ qubit state. Additionally, we propose a method using near-Clifford circuits to find the optimal number of these copies in the presence of realistic noise, which limits this error suppression. We perform a numerical comparison between the original method and our qubit-efficient version with a realistic trapped-ion noise model. We find that REQUEST can reproduce the exponential suppression of errors of the virtual distillation approach, while out-performing virtual distillation when fewer than $3N+1$ qubits are available. Finally, we examine the scaling of the number of shots $N_S$ required for REQUEST to achieve useful corrections. We find that $N_S$ remains reasonable well into the quantum advantage regime where $N$ is hundreds of qubits.