arXiv:2609.31934
Quantum error mitigation from information dynamics
Philippe Suchsland, Thomas Schuster, Manuel S. Rudolph, Nicholas Noll, Thomas E. O'Brien, Zlatko K. Minev
Abstract
Overcoming experimental errors is a central challenge in quantum science. This challenge is particularly acute in large and complex quantum circuits, where quantum information spreads into exponentially many paths, foiling any direct attempt to understand errors' impact. In this work, we present a novel approach to understanding and mitigating experimental errors, based on a surprisingly compact representation of a quantum circuit's information dynamics, the reactivity function. The shape of a circuit's reactivity function governs its noise response, and its amenability to different error mitigation methods. From this viewpoint, we introduce two complementary such methods. The first, Pauli-path zero-noise extrapolation, fits a parameterized model to a circuit's reactivity function and uses it to accurately extrapolate to zero noise. The second, tunable error cancellation, filters a circuit's reactivity function to cancel noise on all information up to a threshold locality. Crucially, both methods feature a tuning parameter that allows one to systematically improve their accuracy until convergence. Across three circuit families from published experiments, our methods overcome conventional zero-noise extrapolation biases as large as 30%; at $\sim$$10^{-2}$ root-mean-square error, they reduce sampling cost relative to probabilistic error cancellation by up to a factor of $10^3$.
AI Summary
The paper asks which circuit properties govern the accuracy and sampling cost of quantum error mitigation, and how approximate mitigation can be systematically refined. It uses the reactivity function $R(w)$, the signed contribution of operator paths with accumulated noise weight $w$, to express the noisy expectation value as $C(\gamma)=\sum_w e^{-\gamma w}R(w)$. A smoothing construction motivates compact effective descriptions at finite measurement precision; spectral decompositions extend the representation beyond Pauli noise. Two mitigation approaches follow. Pauli-path zero-noise extrapolation fits cumulant or multi-exponential models, increasing model order to improve expressivity. Tunable error cancellation constructs linear-combination filters that approximately invert noise below a chosen path-weight threshold $w_*$. Its analytic noise-amplification construction has sampling overhead $e^{4\gamma_0w_*+O(\sqrt{\gamma_0w_*})}$ at fixed accuracy, rather than PEC’s $e^{3\gamma_0V}$ volume-dependent overhead. Discrete Pauli-error insertions can reduce overhead further, with PEC recovered as a limiting case. Numerical benchmarks use nuclear-spin OTOCs, Stark-localized XY dynamics and the two-dimensional transverse-field Ising model, with simulated shot noise and native rate $\gamma_0=0.0025$. Reported improvements include overcoming single-exponential extrapolation biases up to approximately 30% and sampling savings up to $10^3$ relative to PEC near $10^{-2}$ RMSE. Validation is numerical, primarily for systems up to 16 qubits under depolarizing noise. Fitting methods can retain bias plateaus or fail for sign-changing signals; threshold selection relies on empirical convergence. Discrete insertions require Pauli-noise characterization and twirling. Hardware validation, including coherent and non-unital noise, remains future work.
Error found: The displayed Gaussian-smoothing identity omits a boundary correction on the stated nonnegative weight domain and requires an explicit extension of the final integration to the whole real axis to hold exactly. The main-text adaptive cumulant estimator uses min-with-zero clipping, which suppresses significant corrections and can make insignificant corrections unbounded, contrary to the nonnegative stable prescription in the appendix.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.