Quantum Error Mitigation by Global Randomized Error Cancellation for Adiabatic Evolution in the Schwinger Model

Oleg Kaikov, Theo Saporiti, Vasily Sazonov, Mohamed Tamaazousti · 2025

Read Quantum Error Mitigation by Global Randomized Error Cancellation for Adiabatic Evolution in the Schwinger Model by Oleg Kaikov, Theo Saporiti, Vasily Sazonov, Mohamed Tamaazousti (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-06601.txt.

Introduction

Oleg Kaikov and three co-authors wrote this 2025 arXiv paper on cleaning up errors in quantum computers. Today's quantum machines are noisy, so results drift from the truth. One fix, called global randomized error cancellation, learns the noise pattern and undoes it. The team extended that fix to a slow, step-by-step style of quantum evolution called adiabatic evolution.

They tested the method on a lattice version of the Schwinger model, a classic model of electric forces, run on a simulated noisy device. Training gathered in one setting of the model carried over to a different setting, even across a change of phase. The method also beat a rival fix called zero noise extrapolation while using fewer gates. The authors suggest ways to cut the cost further and to widen the method's reach.

Catalog record

CREATOROleg Kaikov, Theo Saporiti, Vasily Sazonov, Mohamed Tamaazousti
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE42 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-06601

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Quantum Error Mitigation by Global Randomized Error Cancellation for Adiabatic Evolution in the Schwinger Model Oleg Kaikov, Theo Saporiti, Vasily Sazonov, Mohamed Tamaazousti Abstract We extend the global randomized error cancellation (GREC) method for quantum error mitigation (QEM) in an application to adiabatic evolution of states on a noisy quantum device. We apply the adiabatic GREC method to the evolution of eigenstates in the lattice Schwinger model on a simulated quantum device with custom noise. Our results suggest that the corresponding QEM learned in one parameter regime of the model successfully transfers to a different parameter regime. In particular, our findings indicate that it transfers between different phases of the model. We observe that adiabatic GREC produces a smaller error than zero noise extrapolation (ZNE). Furthermore, in general, adiabatic GREC can be more cost-efficient in terms of the total number of gates used for the simulations. We comment on approaches to further reduce the necessary quantum computational resources. We also outline extensions of the introduced adiabatic GREC QEM method. I Introduction Quantum computing allows us to perform calculations in regimes inaccessible by classical numerical approaches e2020-100571-8 ; ab6311 ; PRXQuantum.5.037001 ; 1.430.0228 . For instance, standard Monte Carlo methods are inapplicable to lattice Quantum Chromodynamics (LQCD) in the presence of a chemical potential or a topological \theta -term deForcrand:2010j3 ; NAGATA2022103991 : These are examples of the so-called sign problem. However, while fully fault-tolerant quantum computing may be accomplished in the near future PhysRevA.52.R2493 ; PhysRevLett.77.793 ; PhysRevA.54.1098 ; CBO9780511976667 , present-day quantum hardware is affected by quantum errors 0034-4885/61/2/002 ; DiVincenzo_2000 ; nature08812 ; RevModPhys.86.153 ; q-2018-08-06-79 . This will also hold for pioneering hardware in the fault-tolerant era PRXQuantum.3.010345 ; arXiv.2501.05694 ; arXiv.2501.09079 . Therefore, current noisy intermediate-scale quantum (NISQ) devices require quantum error mitigation (QEM) of the performed measurements s41534-022-00618-z ; PhysRevLett.131.210602 ; RevModPhys.95.045005 ; s41567-024-02536-7 . To this end, various QEM techniques were developed, including q-2021-11-26-592 ; PhysRevResearch.3.033098 ; q-2023-06-06-1034 ; PRXQuantum.2.040330 ; arXiv.2102.02120 ; PhysRevA.105.042408 ; arXiv.2502.21159 ; 9259940 ; PhysRevA.105.032620 ; s41534-019-0233-0 ; QCE57702.2023.00102 ; PhysRevA.58.2733 ; PhysRevLett.119.180509 ; PhysRevX.8.031027 ; ACCESS.2020.3031607 ; PhysRevApplied.20.064027 ; PhysRevA.94.052325 ; PhysRevA.104.052607 ; PhysRevX.7.021050 ; s41586-023-06096-3 . I.1 Problem-partitioning Due to the differences between the quantum and the classical computational methods outlined above, numerical tasks, such as those affected by the sign problem in LQCD, can be partitioned into two regimes: First, the “classical-and-quantum” (CL-QU) regime accessible to both standard classical and noisy quantum computations. Second the “only-quantum” (only-QU) regime accessible solely to noisy quantum computations. A subset of QEM techniques, see for example q-2021-11-26-592 ; PhysRevResearch.3.033098 ; q-2023-06-06-1034 ; PRXQuantum.2.040330 ; arXiv.2102.02120 ; PhysRevA.105.042408 , is based on the idea of problem-partitioning: The properties of QEM learned in the CL-QU regime are extrapolated to the only-QU regime. One such QEM method that is relevant to the present work is global randomized error cancellation (GREC) PhysRevA.105.042408 . The GREC method can be outlined as follows: Controllable noise is added to the original noisy quantum circuit in the form of random insertions of parametrized gates. An observable of interest is measured in the CL-QU regime using the randomized quantum circuits, as well as on a classical device. A parametrized combination of the noisy measurements is found that best approximates the classical measurement. A mitigated quantum measurement of the observable in the only-QU regime is found by applying the previously learned combination’s parameters to the measurements performed in the only-QU regime that are obtained with the same randomized realizations of the quantum circuits. We review GREC in more detail in Sec. II.1 . There are (at least) two different tasks that can be addressed regarding a QEM technique that utilizes the idea of problem-partitioning. First, the performance of the QEM technique. Second, a particular case of the first, where the CL-QU and the only-QU regions are in different phases of the model. We expect the second task to be a more challenging test of a QEM technique applied to a quantum computational method than the first, as some approaches may suffer from effects induced by the phase transition (PT). In the present work we study the Schwinger model PhysRev.128.2425 , where for demonstration purposes we consider a sub-region of the parameter space to focus on providing an example of a combination of a quantum computational method and a QEM technique that can be successfully applied to the second task. The specific quantum computational method arXiv.2407.09224 that we use is based on the adiabatic theorem of BF01343193 (see the overview in Sec. II.2 ). Throughout the work, we refer to the CL-QU regime as the “learning” region \mathcal{L} and to the only-QU regime as the “prediction” region \mathcal{P} . We assume that the considered parameter sub-space of the \mathcal{P} region is in sufficient proximity to the \mathcal{L} region. I.2 Applicability Studying phase diagrams is a substantial part of understanding physical theories. Regions of PTs are therefore of particular significance. The GREC QEM scheme was developed to be used for the mitigation of the effects of nonanalyticities in the proximity of PTs. An assumption of GREC is that the corresponding QEM scheme does not depend on nonanalyticities. However, this assumption has not yet been tested. In fact, for an order parameter with a nonanalyticity at a first-order PT we expect the properties of a GREC-like QEM scheme to differ in the regions separated by the PT. Within the adiabatic GREC studied in the present work, the adiabatic evolution of a single eigenstate follows the line of the corresponding energy level. All lines of energy levels are analytic even at PTs. The PTs are encoded in the crossings of these lines. This allows the adiabatic GREC QEM scheme to avoid problems of nonanalyticity. The example of the Schwinger model PhysRev.128.2425 (see Sec. II.3 for a brief overview) illustrates the applicability of our mitigation method to a similar adiabatic evolution task motivated by the sign problem. Namely, the problem where the CL-QU regime includes both vanishing and small non-zero values of the background electric field l_{0} , while the only-QU regime is represented by arbitrarily large non-zero values of l_{0} . See e.g. ab6311 ; deForcrand:2010j3 ; NAGATA2022103991 for overviews of solutions to tasks involving small non-zero values of parameters that lead to the sign problem. In the adiabatic evolution task motivated by the sign problem, the CL-QU regime must contain non-zero values of l_{0} so that the properties of the QEM learned in that regime can be smoothly extrapolated to the only-QU regime. I.3 Goals Our first goal is to develop a GREC-like QEM scheme adapted to the mitigation of measurements obtained from a noisy quantum device by the adiabatic method in arXiv.2407.09224 applied to the lattice Schwinger model ( 4 ). For the adiabatic evolution in arXiv.2407.09224 , the time-dependent adiabatic parameter is l_{0} while all other parameters are fixed. Our second goal is to investigate whether the QEM of the energy levels learned by the adiabatic GREC method in one sub-domain of l_{0} transfers to a different sub-domain of l_{0} for the lattice Schwinger model ( 4 ). In particular, within the (l_{0},m/g) -parameter region ( m/g denotes the lattice mass over coupling ratio) containing the first-order PT at a critical

Read the complete text