AI-Powered Noisy Quantum Emulation - Generalized Gate-Based Protocols for Hardware-Agnostic Simulation

Sep 17, 2026

Uploaded by Sage

AI-powered noisy quantum emulation enables device-specific error modeling to improve the classical validation loop for NISQ programs. The work introduces a general protocol that approximates target-device emulators without pulse-level control, using machine learning on gate set tomography data to infer noise model input parameters. Effectiveness is shown by estimating unitary coupled cluster energy for the H2 molecule and benchmarking against real hardware, achieving high-accuracy noise capture with only 0.3% mean absolute deviation in expectation values versus state-vector simulation.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Introduction

Motivation in the NISQ era

Need for device-specific emulators

Limitations of calibration data and current approaches

arXiv :2502 . 19872v1 [ quant-ph] 27 Feb 2025 AI-Powered Noisy Quantum Emulation: Generalized Gate-Based Protocols for Hardware-Agnostic Simulation Matthew Ho, 1, ∗ Jun Yong Khoo, 1,† Adrian M. Mak, 1 and Stefano Carrazza2, 3, 4, 5 1 Institute of High Performance Computing (IHPC), Agency for Science, Technology and Research (A*STAR), 1 Fusionopolis Way, \#16-16 Connexis, Singapore 138632, Republic of Singapore‡ 2 TIF Lab, Dipartimento di Fisica, Universit`a degli Studi di Milano, Milan, Italy 3 INFN, Sezione di Milano, I-20133 Milan, Italy 4 Theoretical Physics Department, CERN, CH-1211 Geneva 23, Switzerland 5 Quantum Research Center, Technology Innovation Institute, Abu Dhabi, United Arab Emirates (Dated: February 28, 2025) Quantum computer emulators model the behavior and error rates of specific quantum processors. Without accurate noise models in these emulators, it is challenging for users to optimize and debug executable quantum programs prior to running them on the quantum device, as device-specific noise is not properly accounted for. To overcome this challenge, we introduce a general protocol to approximate device-specific emulators without requiring pulse-level control. By applying machine learning to data obtained from gate set tomography, we construct a device-specific emulator by predicting the noise model input parameters that best match the target device. We demonstrate the effectiveness of our protocol’s emulator in estimating the unitary coupled cluster energy of the H2 molecule and compare the results with those from actual quantum hardware. Remarkably, our noise model captures device noise with high accuracy, achieving a mean absolute difference of just 0.3% in expectation value relative to the state-vector simulation. I. INTRODUCTION In the current noisy intermediate-scale quantum [1](NISQ) computing era, access to quantum computers are predominantly cloud-based [2–7] . Despite the host of available NISQ computers from different service providers, their accessibility is often limited or costly; users can be subjected to long queuing times or have to pay for circuit executions. Costs will increase when compute-intensive problems require many iterations, especially when using quantum algorithms that are variational in nature [8–10] . As a result, despite variational quantum algorithms being the most prevalent type of quantum algorithm for NISQ computers [10–12], with the exception of simple toy problems, it is unfeasible for most users to perform the definitive routine of quantum circuit parameter optimization directly on NISQ computers. Instead, parameter optimization is often done via classical simulations, and the optimized variational quantum circuit is then run on the NISQ device for validation or comparison [13, 14] . To maximize the efficiency of the validation run on NISQ computers, it is crucial to have classical simulators that can faithfully replicate the behavior of individual NISQ devices through device-specific emulators. Users can construct an emulator for their chosen quantum computer, but not all input parameters required for a comprehensive noise model are regularly updated or readily available from the hardware provider. These parameters ∗ [matthew](matthew ho@ihpc.a-star.edu.sg)[ ](matthew ho@ihpc.a-star.edu.sg)[ho@ihpc.a-star.edu.sg](matthew ho@ihpc.a-star.edu.sg)[ ](matthew ho@ihpc.a-star.edu.sg)† [khoojy@ihpc.a-star.edu.sg](khoojy@ihpc.a-star.edu.sg)[ ](khoojy@ihpc.a-star.edu.sg)‡ MH and JYK contributed equally are derived from calibration protocols that require access to pulse-level control. In the absence of device-specific emulators, users typically rely on noiseless state-vector simulations of their quantum algorithms, the results of which would substantially deviate from those obtained on currently available NISQ devices. The lack of such device-specific emulators makes job submissions to NISQ devices costly and time-consuming, with limited conclusions

FAQs

Why are accurate noise models important for validating NISQ quantum programs?

Accurate noise models are required to reproduce device behavior and error rates in classical emulators. Without them, results from noiseless or mismatched simulations can substantially deviate from those on real hardware, making optimization and debugging unreliable.

What problem does the proposed protocol address?

The protocol targets the lack of pulse-level control and missing device-specific parameters for comprehensive noise models. It aims to approximate device-specific emulators even when detailed calibration data is not available.

How does the protocol build a device-specific emulator?

It applies supervised machine learning to data obtained from gate set tomography. The model predicts noise model input parameters that best match the target device’s noise behavior.

Sep 17, 2026

Uploaded by Sage