Opportunities and challenges in scaling quantum error detection on hardware

TL;DR

This paper evaluates the potential and obstacles of scaling quantum error detection on hardware through benchmarking with real and simulated quantum computers, highlighting both challenges and promising results.

Contribution

It provides a comprehensive benchmarking study of quantum error detection codes on hardware, analyzing scalability and error thresholds with up to 74 qubits.

Findings

Quantum error detection shows promise for scaling despite exponential sample and processing costs.

Benchmarking with up to 74 qubits demonstrates the feasibility of error detection on current hardware.

Estimation of pseudothresholds maps the error detection frontier on quantum computers.

Abstract

Quantum error detection can produce unbiased expectation values that exponentially converge to noiseless results as the code distance is increased. Despite this, its performance as an error mitigation technique is relatively understudied on quantum hardware because of its two main drawbacks: (i) the number of samples increases exponentially in the circuit depth/noise level, and (ii) the classical processing generally grows exponentially in the code distance, though exceptions exist. Additionally, the constant (but often large) overhead of embedding the code and logical operations on hardware can make accuracy worse instead of better. In this work, we seek to provide a clear picture of these opportunities and challenges for scaling quantum error detection on hardware. We do so by performing a detailed benchmarking study on real and simulated noisy quantum computers, using the repetition code and triangular color code for memory experiments and logical computations with up to $74$ physical qubits. In addition to these benchmarks, we estimate the pseudothreshold of codes to map the frontier of error detection on current and future quantum computers. Despite the challenges, our results show strong promise for scaling quantum error detection on hardware.