Anchor Technique Reduces Quantum Computer Performance Variability by 73%, Improving Output Consistency

Quantum computing promises to revolutionise many fields, but current noisy intermediate-scale (NISQ) devices suffer from hardware limitations that introduce errors and, crucially, significant variability in performance. Yuqian Huo, Daniel Leeds, and Jason Ludmir, all from Rice University, along with colleagues including Nicholas S. DiBrita and Tirthak Patel, tackled this challenge by developing Anchor, a novel technique that dramatically reduces performance fluctuations. Unlike existing methods which focus solely on minimising errors, Anchor uses linear programming to stabilise results, achieving an average 73% reduction in performance variability compared to current state-of-the-art error reduction implementations. This breakthrough represents a crucial step towards reliable quantum computation, paving the way for consistent and repeatable results even on imperfect hardware.

Circuit Variability and Prediction with Anchor

The research team developed Anchor, a technique to improve the reliability of quantum computations, particularly when using multiple quantum computers. A key challenge in quantum computing is variability, where different computers produce slightly different results for the same calculation due to inherent noise and imperfections. Anchor addresses this by generating multiple ways to compile a quantum circuit for each computer, then predicting how different these compilations will be in terms of their output probabilities. Lower predicted differences indicate more consistent results. The technique employs linear programming, an optimization algorithm, to determine the optimal number of repetitions, or “shots”, to run for each compilation on each computer, minimizing overall error and ensuring equalized performance.

Importantly, Anchor runs the shots only on the scheduled computer, using others to inform the optimization process. To illustrate how Anchor works, scientists tested it with a simple two-qubit circuit, generating two different compilations for each of two computers. By comparing the output distributions, they demonstrated how hardware noise affects program accuracy, quantifying the difference using Total Variation Distance. The linear programming algorithm then determined the optimal fraction of shots to allocate to each map on each computer, aiming for minimal overall error and equalized performance.

This approach delivers more reliable and consistent results, even with noisy hardware. Anchor reduces variability by diversifying circuit maps and equalizing performance, allowing researchers to focus execution on a single computer and simplifying resource management. The technique offers adaptability, allowing a trade-off between perfect equalization and minimizing overall error. In essence, Anchor is a sophisticated scheduling and optimization technique that leverages multiple circuit maps and linear programming to improve the reliability and consistency of quantum computations in a distributed environment.

Anchor Technique Minimizes Quantum Performance Variability

Scientists have developed Anchor, a novel technique to address performance variability in noisy intermediate-scale quantum (NISQ) computers. Recognizing that existing error reduction methods don’t fully resolve inconsistent results, the team focused on minimizing fluctuations in program performance. Researchers constructed and executed four-qubit circuits, collecting data from numerous “shots” to estimate the output probability distribution. They meticulously mapped logical qubits onto physical qubits, acknowledging that variations in physical qubit error rates significantly impact overall performance.

Crucially, they ran the same quantum circuit on two different circuit maps, each exhibiting different error characteristics. By comparing the output distributions, they demonstrated how hardware noise directly affects program fidelity, quantifying the difference using Total Variation Distance. They then implemented Anchor, a linear programming technique, to optimize qubit mapping and minimize performance variability, achieving an average reduction of 73% compared to state-of-the-art error reduction methods. The team openly shared Anchor’s framework, codebase, and experimental data, ensuring reproducibility and facilitating further research.

Anchor Technique Stabilises Quantum Computation Performance

Scientists have developed Anchor, a groundbreaking technique to address a critical challenge in quantum computing: performance variability. Current noisy intermediate-scale quantum (NISQ) computers are susceptible to hardware noise, causing fluctuations in the output of quantum programs, even when running the same program repeatedly. This variability stems from both temporal and spatial factors. Researchers identified and quantified this variability within the IBM Quantum cloud-based service, demonstrating its significant impact on program performance. The team’s innovative approach departs from conventional error-reduction methods by focusing directly on minimizing fluctuations in output fidelity, rather than simply improving the average result.

Anchor achieves this by intelligently distributing the execution of quantum circuit “shots” across multiple different circuit maps, leveraging a novel linear programming framework. A learning-based predictor estimates output fidelity values, providing crucial input for optimizing this shot assignment and accounting for stochastic noise effects. Through simulation and experiments on real quantum hardware, scientists demonstrated that Anchor reduces performance variability by an average of 73% compared to state-of-the-art error reduction implementations. This breakthrough delivers substantially more consistent and reliable results for users, regardless of the specific quantum hardware used or the time of execution. The team conducted comprehensive overhead and ablation analyses, confirming the efficacy of Anchor’s implementation and made the framework, codebase, and experimental data openly available.

Reducing Performance Variability in Quantum Circuits

Anchor represents a significant advance in addressing performance variability on noisy intermediate-scale quantum (NISQ) computers. Researchers have developed a novel technique that leverages linear programming and a performance predictor to deliver more consistent results from quantum circuits, irrespective of the specific computer used or the time of execution. This work departs from existing approaches which primarily focus on reducing errors, by directly tackling the issue of fluctuating performance that hinders reliable quantum computation. Evaluations demonstrate that Anchor reduces performance variability by an average of 73% compared to current state-of-the-art implementations, enhancing the reliability of quantum algorithms and facilitating more predictable outcomes. The team acknowledges that performance variability remains a challenge and made the framework, codebase, and experimental data openly available to the research community.