Quantum Error Correction Scales Up, Paving the Way for Reliable Computers

Researchers are tackling a critical challenge in realising practical quantum computers: verifying the effectiveness of quantum error correction. John Zhuoyang Ye and Jens Palsberg, both from the University of California, Los Angeles, demonstrate a new, scalable testing approach that significantly improves upon current benchmarking methods. Current techniques, such as randomised fault-injection testing utilising tools like \stim, struggle to assess error correction performance at the distances necessary for fault-tolerant quantum computation. This work introduces a combination of stratified fault injection and extrapolation, enabling testing to scale to distance 17 with a physical rate of 0.0005 within a reasonable timeframe, and ultimately providing more confidence in estimating logical rates.

Scalable benchmarking of quantum error correction using targeted fault injection and extrapolation

Scientists have developed a new technique for rigorously testing quantum error correction, achieving a significant leap in scalability for assessing the reliability of quantum computations. The prevailing method, randomized fault-injection testing, faces limitations when applied to increasingly complex quantum systems, particularly those designed to correct errors.
This work introduces ScaLER (Scalable Logical Error Rate Testing), an approach that combines targeted fault injection with extrapolation to dramatically reduce the computational effort required to benchmark quantum error correction implementations. Researchers successfully scaled their tool to a distance of 17, with a physical error rate of 0.0005, completing the testing process within a two-hour timeframe on standard desktop hardware.

This represents a substantial improvement over existing tools, which struggle to achieve comparable results at such high distances and low error rates. For this specific configuration, ScaLER estimated a logical error rate of 1.51 × 10−11 with high confidence, a figure previously unattainable within the same time constraints.

The core innovation lies in the realization that not all potential error scenarios require exhaustive sampling. By strategically focusing on the most impactful faults and then extrapolating to estimate the behaviour of less frequent errors, ScaLER significantly reduces the number of simulations needed.

This allows for a more efficient and accurate assessment of logical error rates, a critical metric for evaluating the performance of quantum error correction algorithms. The team’s software is publicly available, facilitating further research and development in this crucial area of quantum computing. This breakthrough addresses a major obstacle in the field, as the testing required for high-quality quantum error correction algorithms is often computationally prohibitive.

The ability to rapidly and reliably estimate logical error rates is essential for advancing the development of fault-tolerant quantum computers, paving the way for practical applications in science and industry. ScaLER’s performance surpasses that of the state-of-the-art tool Stim, which, for the same two-hour time budget and a surface code with distance 7, estimated a significantly higher logical error rate of 5.95×10−6. The new method promises to accelerate progress towards realizing the full potential of quantum computation by streamlining the benchmarking process and enabling more thorough evaluation of error correction strategies.

Stratified fault injection and extrapolation for scalable quantum error correction benchmarking

A scalable approach to benchmarking quantum error correction underpinned this work, combining stratified fault injection with extrapolation techniques. Researchers addressed limitations in existing tools like \stim, which become inefficient for larger distances and low physical error rates. The methodology centres on efficiently sampling the fault space, then utilising extrapolation to complete the testing process, thereby reducing computational demands.

This enabled testing to scale to a distance of 17 with a physical error rate of 0.0005 within a two-hour timeframe on standard desktop hardware. The study implemented stratified fault injection, focusing initial sampling on fault weights that contribute most significantly to logical errors. High-weight samples, representing more complex error scenarios, were then tested selectively to refine the estimation process.

An algorithm was developed to pseudo-code the process, enabling systematic exploration of the fault space and efficient data collection. This algorithm prioritises faults based on their potential impact, reducing the number of simulations required to achieve statistical confidence. Implementation involved constructing a testing framework capable of simulating quantum circuits and injecting faults according to the stratified sampling strategy.

Evaluation compared the new tool, ScaLER, against \stim across multiple parameters, including scalability and accuracy. Specifically, ScaLER estimated a logical error rate of 1.51 × 10−11 for a distance-17 surface code with a physical error rate of 0.0005, a significant improvement over \stim’s best performance of 5.95×10−6 for a distance-7 code under the same conditions. The research team made their software publicly available to facilitate further investigation and validation.

Efficient logical error rate estimation using stratified fault injection and extrapolation

ScaLER, a new approach to benchmarking quantum error correction implementations, estimated a logical error rate of 1.51 × 10−11 for a surface code with distance 17 and a physical error rate of 0.0005, achieved within a two-hour time budget on a desktop computer. This estimation was performed using a combination of stratified fault injection and extrapolation techniques, allowing for scalable testing of larger distance codes.

The research demonstrates an ability to efficiently sample the fault space and then extrapolate to complete the testing process. For the same two-hour time budget, the state-of-the-art tool Stim estimated a significantly higher logical error rate of 5.95 × 10−6 for a surface code with distance 7. This represents a substantial improvement in the precision of logical error rate estimation.

ScaLER’s methodology involved injecting faults into a circuit and checking for logical errors, but with a focus on efficient sampling and extrapolation to overcome the limitations of traditional randomised fault-injection testing. The study utilized a repetition code with distance 3 as a preliminary example, demonstrating how Stim estimates logical error rates by injecting Pauli-X errors and tracking the propagation of these errors through the circuit.

Analysis of 11 samples, each with weight 1, revealed 2 logical errors, resulting in an initial logical error rate estimate of approximately 2 divided by 11. Further investigation with a surface code of distance 7 showed that samples with lower weights dominated the testing campaign, highlighting the importance of efficient sampling strategies. The number of samples recorded was 1e7, with the number of logical errors ranging from 0 to 80 depending on the weight.

ScaLER demonstrates enhanced surface code benchmarking through stratified fault injection and extrapolation

Researchers have developed a scalable approach to benchmarking implementations of quantum error correction, significantly reducing the testing effort required for high-quality algorithms. The new tool, ScaLER, combines stratified fault injection with extrapolation to efficiently sample the fault space and estimate logical error rates.

This methodology allows for testing at larger distances than previously possible with conventional randomised fault-injection testing, such as the state-of-the-art tool Stim. Specifically, ScaLER achieved estimation of the logical error rate with high confidence for a surface code with distance 17 and a physical error rate of 0.0005 within a two-hour time budget on a desktop computer, reporting a rate of 1.51 × 10−11.

In comparison, Stim estimated a significantly higher logical error rate of 5.95×10−6 for a surface code with a lower distance of 13, given the same computational resources. This improvement in scalability is crucial as testing becomes increasingly demanding for algorithms with lower logical error rates.

The authors acknowledge that the performance of ScaLER is dependent on the accuracy of the extrapolation methods used to complete the testing task. Future research directions include exploring more sophisticated extrapolation techniques and applying ScaLER to a wider range of quantum error correction codes and hardware platforms. These advancements will be essential for validating and optimising fault-tolerant quantum computing systems as they move closer to practical realisation and potential applications in science and business.