arXiv:2605.02966
QBalance: A Reproducible Multi-Objective Workflow for Quantum Compilation, Noise Suppression, and Error-Mitigation Strategy Selection
Soumyadip Sarkar
Abstract
Near-term quantum workloads are shaped by coupled compilation and execution choices: qubit layout, routing, basis translation, gate suppression, measurement mitigation, shot budget, and artifact reproducibility. This paper analyzes QBalance, a Python workflow library for dataset-level selection among quantum compilation, noise-suppression, and error-mitigation strategies built on the Qiskit ecosystem. The contribution is formulated as a finite multi-objective strategy-selection problem over circuits, backends, and transformation policies. The manuscript derives the implemented weighted objective, non-dominated selection rule, survival-product error proxy, Bayesian linear candidate-ordering surrogate, and distributional diagnostics. It also positions the system relative to established work on Qiskit pass-manager compilation, SABRE-style routing, randomized compiling, dynamical decoupling, zero-noise extrapolation, matrix-free measurement mitigation, circuit cutting, and Thompson sampling. The analysis shows that QBalance provides a reproducible orchestration and artifact model for quantum workflow studies. It also establishes precise limitations: the current bandit mechanism orders candidates but does not reduce the number of candidate evaluations, the custom layout heuristic is greedy and only partially topology-aware, the implemented ZNE helper is parity-centered, and the cutting integration is a hook rather than a full reconstruction pipeline.
AI Summary
QBalance addresses how to compare coupled compilation, noise-suppression, and error-mitigation choices across circuit datasets. It presents a Python workflow built on Qiskit that evaluates finite strategy sets, selects a strategy per circuit, and saves circuits, metrics, metadata, and reports. The paper analyzes version 0.1.0 through source-code inspection and mathematical derivations; it reports no executed benchmark matrix or hardware results. The default candidate generator produces at most 23 unique strategies spanning compilation settings and optional suppression, mitigation, and cutting paths. Selection minimizes $J=\mathrm{depth}+2n_{2q}+10\widehat E+0.1t_{\mathrm{compile}}$, optionally after Pareto filtering on depth, two-qubit count, and estimated error. The error proxy is $\widehat E=1-\prod_j(1-e_j)$, using calibration-derived or fallback operation errors. The analysis derives the Bayesian linear posterior used for Thompson-style candidate ordering and establishes that both grid and bandit modes attempt every candidate: candidate-evaluation count remains $O(NC)$ for $N$ circuits and $C$ strategies. Pareto filtering has quadratic worst-case cost in the number of unique metric vectors. The stated limitations separate workflow functionality from physical-performance evidence. The built-in dataset contains only Bell, three-qubit GHZ, and four-qubit QFT-like examples. The ordering model lacks circuit features; layout is greedy and only partially topology-aware; and the error proxy omits coherent and correlated noise. ZNE extrapolates parity rather than arbitrary observables, cutting lacks end-to-end reconstruction, and Runtime primitives are not used by the main workflow. Cache keys also omit software-version and calibration provenance.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.