Simulation and Experiment Reveal No Fidelity Difference in Quantum Error Correction

Researchers at Delft University of Technology, led by S. L. M. van der Meer, have conducted an experimental investigation into the performance of a bitflip repetition code implemented on superconducting hardware. The study centres on discerning the effects of coherent and stochastic errors on the logical performance of qubits, a crucial step towards realising fault-tolerant quantum computation. Results demonstrate a notable discrepancy between theoretical simulations and experimental observations, suggesting that seemingly minor variations in qubit frequencies can effectively convert coherent errors into stochastic errors, thereby impacting the efficacy of quantum error correction. A deeper understanding of coherent error behaviour is paramount for advancing the development of practical and robust quantum error correction schemes.

Coherent noise tolerance exceeds stochastic thresholds despite experimental discrepancies

The experimental findings reveal that error rates were successfully reduced to 0.18, a significant achievement that surpasses the previously established threshold of 0.11 for stochastic depolarizing noise. This key advance demonstrates a theoretical tolerance for coherent noise, which has long been considered a potentially advantageous characteristic in the context of quantum error correction. Quantum error correction protocols theoretically benefit from coherent noise, as errors affecting the phase of quantum bits are inherently more amenable to correction than stochastic, random errors that introduce unpredictable state flips. However, experiments utilising a bitflip repetition code and a transmon quantum processor did not fully observe this predicted advantage for either distance-3 or distance-5 codes. This unexpected result prompted a detailed investigation into the underlying causes of the divergence.

Detailed simulations, meticulously conducted using a scalable free-fermion simulator, were designed to accurately mirror the experimental setup, allowing for a controlled and direct comparison of the impacts of different noise types and code distances of three and five. The free-fermion simulator allows for efficient modelling of the complex interactions within the quantum processor, providing a robust platform for predicting the behaviour of the error correction code under various noise conditions. The observed discrepancy between simulation and experiment suggests that subtle, previously unaccounted-for drifts in qubit frequencies may be transforming injected coherent errors into stochastic ones, effectively masking the expected performance difference. These frequency drifts introduce unwanted phase fluctuations, effectively randomising the coherent errors and diminishing their correctability. The team meticulously calibrated the device to minimise the initial detection probability of errors before injecting controlled noise, ensuring a fair comparison between the different error types. Analysis of the bitflip repetition code, implemented on a transmon quantum processor, a widely used type of superconducting qubit, revealed close alignment between observed performance and simulations when accounting for baseline qubit noise inherent in the system. Further investigation focused on the implications of these findings for building stable and reliable quantum computers, highlighting the critical need to comprehensively understand all forms of noise affecting quantum error correction and to develop strategies for mitigating their impact.

Distinguishing coherent and stochastic error impacts informs strong quantum correction strategies

Maintaining the delicate quantum states of qubits for a duration sufficient to perform useful calculations remains a formidable challenge despite significant advances in the fabrication and control of quantum processors. Protecting these fragile states from the pervasive influence of environmental noise is absolutely vital, and quantum error correction offers a promising, albeit complex, solution through the encoding of redundant information. This approach allows for the detection and correction of errors without directly measuring the quantum state, thereby preserving the superposition and entanglement necessary for quantum computation. Accurately predicting how different noise types will impact these correction schemes is proving surprisingly difficult, as this experimental investigation of quantum error correction reveals a divergence between simulated and observed performance with bitflip repetition codes. The bitflip repetition code, chosen for its simplicity and scalability, encodes a logical qubit across multiple physical qubits, allowing for the detection and correction of bitflip errors, where a 0 state is flipped to a 1, or vice versa.

This suggests that current theoretical models may underestimate the influence of real-world factors impacting qubit stability and coherence. While the experiments did not fully mirror simulations predicting differing impacts from coherent and stochastic errors, continued investigation into quantum error correction remains paramount. Subtle drifts in qubit frequencies, as previously identified, likely transform injected coherent errors, obscuring the anticipated advantage and opening important questions about accurately modelling noise in quantum processors. These drifts introduce phase-coherent noise, blurring the distinction between the two error types during testing and complicating the process of designing effective error correction strategies. Understanding the interplay between coherent and stochastic noise is crucial for developing robust quantum error correction codes that can effectively protect quantum information from decoherence and other sources of error. Future research will focus on refining noise models to incorporate these frequency drifts and exploring alternative error correction codes that are less susceptible to their effects, ultimately paving the way for the realisation of fault-tolerant quantum computers capable of solving complex problems beyond the reach of classical computers.

The research demonstrated that experimentally injecting coherent versus stochastic errors did not produce the differing logical fidelity predicted by simulation for distance-3 and distance-5 bitflip repetition codes. This matters because accurately modelling noise is vital for building effective quantum error correction, and the discrepancy between simulation and experiment highlights the complexity of real-world quantum systems. Researchers hypothesise that small drifts in qubit frequencies may be responsible, effectively converting coherent errors into stochastic ones. The authors intend to refine noise models to account for these frequency drifts and explore alternative error correction codes.