Predicting the Temporal Variability of Error Rates in Superconducting Quantum Processors: Mitigating Error Accuracy Loss in Queue-Based Quantum Computing Platforms [Focus: Quantum Software and its Engineering]
Laura Rodríguez-Soriano, Francisco García-Herrero, Carmen G. Almudéver
Multiple layers of the quantum system full-stack require data on temporal variations in error rates of physical elements. The absence of this information, especially when users access cloud-based quantum systems with waiting queues between compilation and execution, can lead to unreliable computations due to outdated calibration data. One example is the compilation process that impacts the success rate of quantum circuits. This paper proposes using five machine learning and time-series models to predict error rate variability for two-qubit gates based on calibration data. We compare model accuracy and complexity in terms of execution time. The predicted information is applied to a noise-aware compiler to evaluate the impact on success rate. Experimental results on a real superconducting quantum processor demonstrate accurate predictions of temporal error rate variability, achieving performance comparable to that of real calibration data, reducing the impact of queue waiting times that can reach up to two days.
Multiple layers of the quantum system full-stack require data on temporal variations in error rates of physical elements. The absence of this information, especially when users access cloud-based quantum systems with waiting queues between compilation and execution, can lead to unreliable computations due to outdated calibration data. One example is the compilation process that impacts the success rate of quantum circuits. This paper proposes using five machine learning and time-series models to predict error rate variability for two-qubit gates based on calibration data. We compare model accuracy and complexity in terms of execution time. The predicted information is applied to a noise-aware compiler to evaluate the impact on success rate. Experimental results on a real superconducting quantum processor demonstrate accurate predictions of temporal error rate variability, achieving performance comparable to that of real calibration data, reducing the impact of queue waiting times that can reach up to two days.
Reliably executing quantum algorithms on noisy intermediate-scale quantum (NISQ) devices is challenging, as they are severely constrained and prone to errors. Efficient quantum circuit compilation techniques are therefore crucial for overcoming their limitations and dealing with their high error rates. These techniques consider the quantum hardware restrictions, such as the limited qubit connectivity, and perform some transformations to the original circuit that can be executed on a given quantum processor. Certain compilation methods use error information based on calibration data to further improve the success probability or the fidelity of the circuit to be run. However, it is uncertain to what extent incorporating calibration information in the compilation process can enhance the circuit performance. For instance, considering the most recent error data provided by vendors after calibrating the processor might not be functional enough as quantum systems are subject to drift, making the latest calibration data obsolete within minutes. In this paper, we explore how different usage of calibration data impacts the circuit fidelity, by using several compilation techniques and quantum processors (IBM Perth and Brisbane). To this aim, we implemented a framework that incorporates some of the state-of-the-art noise-aware and non-noise-aware compilation techniques and allows the user to perform fair comparisons under similar processor conditions. Our experiments yield valuable insights into the effects of noise-aware methodologies and the employment of calibration data. The main finding is that pre-processing historical calibration data can improve fidelity when real-time calibration data is not available due to factors such as cloud service latency and waiting queues between compilation and execution on the quantum backend.
Transpilation, particularly noise-aware optimization, is widely regarded as essential for maximizing the performance of quantum circuits on superconducting quantum computers. The common wisdom is that each circuit should be transpiled using up-to-date noise calibration data to optimize fidelity. In this work, we revisit the necessity of frequent noise-adaptive transpilation, conducting an in-depth empirical study across five IBM 127-qubit quantum computers and 16 diverse quantum algorithms. Our findings reveal novel and interesting insights: (1) noise-aware transpilation leads to a heavy concentration of workloads on a small subset of qubits, which increases output error variability; (2) using random mapping can mitigate this effect while maintaining comparable average fidelity; and (3) circuits compiled once with calibration data can be reliably reused across multiple calibration cycles and time periods without significant loss in fidelity. These results suggest that the classical overhead associated with daily, per-circuit noise-aware transpilation may not be justified. We propose lightweight alternatives that reduce this overhead without sacrificing fidelity – offering a path to more efficient and scalable quantum workflows.
The successful implementation of algorithms on quantum processors relies on the accurate control of quantum bits (qubits) to perform logic gate operations. In this era of noisy intermediate-scale quantum (NISQ) computing, systematic miscalibrations, drift, and crosstalk in the control of qubits can lead to a coherent form of error that has no classical analog. Coherent errors severely limit the performance of quantum algorithms in an unpredictable manner, and mitigating their impact is necessary for realizing reliable quantum computations. Moreover, the average error rates measured by randomized benchmarking and related protocols are not sensitive to the full impact of coherent errors and therefore do not reliably predict the global performance of quantum algorithms, leaving us unprepared to validate the accuracy of future large-scale quantum computations. Randomized compiling is a protocol designed to overcome these performance limitations by converting coherent errors into stochastic noise, dramatically reducing unpredictable errors in quantum algorithms and enabling accurate predictions of algorithmic performance from error rates measured via cycle benchmarking. In this work, we demonstrate significant performance gains under randomized compiling for the four-qubit quantum Fourier transform algorithm and for random circuits of variable depth on a superconducting quantum processor. Additionally, we accurately predict algorithm performance using experimentally measured error rates. Our results demonstrate that randomized compiling can be utilized to leverage and predict the capabilities of modern-day noisy quantum processors, paving the way forward for scalable quantum computing.
A detailed understanding of the noise processes and errors in near-term quantum computers is essential for accurate calibration of gates, and for achieving optimal performance when executing near-term applications such as quantum chemistry, optimization, or machine learning problems. This paper introduces several sources of noise in superconducting qubit based quantum computers such as those deployed by IBM, describes some of the main techniques for characterizing the resulting errors, and for mitigating the effects of certain errors without access to quantum error correction. We demonstrate experimental application of these techniques on an IBM Quantum system using the open source Qiskit software library.
Quantum computing promises a paradigm shift in computational power. However, a major challenge is mitigating the inherent noise and errors in quantum circuits. As quantum computers operate, their fragile qubits are highly susceptible to environmental disturbances, leading to errors that affect the outcomes of repeated circuit executions. Understanding and predicting these errors is crucial for improving the accuracy and reliability of quantum computing systems. In this work, we analyze and predict the output patterns of a single qubit quantum circuit by treating the results of repeated executions as a time series. Specifically, we collect measurement data from multiple runs of a quantum circuit and construct a time series from these observations. By training a predictive model, we aim to forecast future outcomes, providing insights into the error behavior of the quantum circuit. Additionally, we analyze time series data from two different circuits executed on the same qubit to investigate potential relationships and assess whether one dataset can be used to predict the other. Our findings reveal key characteristics of quantum circuit outputs, including stationarity, autocorrelation, seasonality, trends, linearity, and causality. The analysis highlights intriguing behaviors within the dataset. Furthermore, we evaluate multiple time series prediction methods and determine that XGBoost (Extreme Gradient Boosting) outperforms other approaches, demonstrating its effectiveness in accurately predicting quantum computing outputs in subsequent runs.
While current quantum computers, referred to as Noisy Intermediate-Scale Quantum (NISQ) computers, are expected to be beneficial for different applications, they are prone to different types of errors. In order to enhance the reliability of quantum systems, noise-aware quantum compilers are used to generate physical quantum circuits to be executed on NISQ computers. The quantum hardware is calibrated very frequently and its error rates are computed accordingly. Based on the hardware error rates, a quantum compiler allocates physical qubits and schedules quantum operations. However, error rates may change post-calibration. To incorporate dynamic error rates into quantum circuit compilation with minimum cost, we propose a Machine Learning (ML)-based scheme to detect the incorrect output of the quantum circuit and predict the Probability of Successful Trials (PST) with high accuracy. Our approach can verify the error rates of the quantum hardware and validate the correctness of the extracted quantum circuit output. We provide a case study of our ML-based reliability models using IBM Q16 Melbourne quantum computer. Our results show that the proposed scheme achieves a very high prediction accuracy.