Evaluating the Resilience of Variational Quantum Algorithms to Leakage Noise

TL;DR

This paper investigates how leakage noise affects the performance of variational quantum algorithms, revealing that leakage generally impairs their effectiveness and highlighting the need for its suppression in practical quantum computing.

Contribution

It provides the first systematic analysis of leakage noise impact on VQAs using realistic hardware scenarios and benchmarks their performance on real-world tasks.

Findings

Leakage noise reduces the expressive power of VQAs.

Leakage noise negatively impacts training and outcomes in data fitting and classification.

VQAs are vulnerable to leakage noise, requiring effective suppression for practical use.

Abstract

As we are entering the era of constructing practical quantum computers, suppressing the inevitable noise to accomplish reliable computational tasks will be the primary goal. Leakage noise, as the amplitude population leaking outside the qubit subspace, is a particularly damaging source of error that error correction approaches cannot handle. However, the impact of this noise on the performance of variational quantum algorithms (VQAs), a type of near-term quantum algorithms that is naturally resistant to a variety of noises, is yet unknown. Here, {we consider a typical scenario with the widely used hardware-efficient ansatz and the emergence of leakage in two-qubit gates}, observing that leakage noise generally reduces the expressive power of VQAs. Furthermore, we benchmark the influence of leakage noise on VQAs in real-world learning tasks. Results show that, both for data fitting and data classification, leakage noise generally has a negative impact on the training process and final outcomes. Our findings give strong evidence that VQAs are vulnerable to leakage noise in most cases, implying that leakage noise must be effectively suppressed in order to achieve practical quantum computing applications, whether for near-term quantum algorithms and long-term error-correcting quantum computing.