Bulletin of the American Physical Society
2024 APS March Meeting
Monday–Friday, March 4–8, 2024; Minneapolis & Virtual
Session D52: Quantum Error Mitigation and Machine LearningFocus
|
Hide Abstracts |
Sponsoring Units: DQI Chair: Sohair Abdullah, University of Wisconsin-Madison Room: 201AB |
Monday, March 4, 2024 3:00PM - 3:12PM |
D52.00001: Resource efficient transfer learning approach for error mitigation of quantum circuits at scale Yuval Baum, Gavin S Hartnett, Nuiok Dicair Large-scale fault-tolerant quantum computers are likely to enable new solutions for problems known to be hard for classical computers. This potential is tempered by the reality that hardware is exceptionally fragile and error-prone, forming a bottleneck in the development of novel applications. While error suppression techniques can dramatically boost algorithmic performance, inherent and irreversible errors, such as T1 processes, limit the ability of achieving quantum utility at scale. Error mitigation techniques provide a route to go beyond that limit at the price of additional resources overhead. Some of these approaches, such as probabilistic error cancellation (PEC), aim to collect a full description of the noise model to facilitate noise suppression. However, this requires over-sampling, which is costly to acquire and scales poorly as the number of qubits increases. In this work, we develop a new, scalable error mitigation approach based on training machine learning models on real device data. We focus on problems where the quantity of interest is a set of expectation values, such as quantum simulations or variational quantum eigensolvers (VQE). Using non-parametric ensemble models, we machine-learn the approximate inverse noise map for a subset of quantum observables. This is achieved by training models on a closely-related family of circuits which are efficiently simulable, and then using transfer learning to apply the learned inverse noise map to the circuits of interest. As a demonstration of our approach, we estimate Pauli observables and show a consistent ability to mitigate errors with a significantly reduced runtime compared to established mitigation techniques such as PEC and zero-noise extrapolation (ZNE). |
Monday, March 4, 2024 3:12PM - 3:24PM |
D52.00002: ABSTRACT WITHDRAWN
|
Monday, March 4, 2024 3:24PM - 3:36PM |
D52.00003: Machine Learning for Practical Quantum Error Mitigation Haoran Liao, Derek S Wang, Iskandar Sitdikov, Ciro Salcedo, Alireza Seif, Zlatko K Minev Quantum computers are actively competing to surpass classical supercomputers, but quantum errors remain their chief obstacle. The key to overcoming these on near-term devices has emerged through the field of quantum error mitigation, enabling improved accuracy at the cost of additional runtime. In practice, however, the success of mitigation is limited by a generally exponential overhead. Can classical machine learning address this challenge on today's quantum computers? Here, through both simulations and experiments on state-of-the-art quantum computers using up to 100 qubits, we demonstrate that machine learning for quantum error mitigation (ML-QEM) can drastically reduce overheads, maintain or even surpass the accuracy of conventional methods, and yield near noise-free results for quantum algorithms. We benchmark a variety of machine learning models---linear regression, random forests, multi-layer perceptrons, and graph neural networks---on diverse classes of quantum circuits, over increasingly complex device-noise profiles, under interpolation and extrapolation, and for small and large quantum circuits. These tests employ the popular digital zero-noise extrapolation method as an added reference. We further show how to scale ML-QEM to classically intractable quantum circuits by mimicking the results of traditional mitigation results, while significantly reducing overhead. Our results highlight the potential of classical machine learning for practical quantum computation. |
Monday, March 4, 2024 3:36PM - 3:48PM |
D52.00004: Virtual distillation with neural quantum states Ryo Maekura, Yasunari Suzuki, Nobuyuki Yoshioka, Yuuki Tokunaga Virtual Distillation (VD) is one of the promising Quantum Error Mitigation (QEM) protocols that can exponentially attenuate computational bias by increasing the number of replicated quantum states. On the other hand, VD faces several challenges such as the difficulties in generating quantum state replicas and in globally entangled measurement. In this work, we address these two issues and propose a novel method to mitigate errors in computational outcomes based on Neural Quantum State (NQS).NQS enables efficient representation of large quantum states of interests with neural networks. Once NQS learns quantum states, we can increase the samples in Monte-Carlo sampling without re-executing quantum experiments, and the globally entangled measurement can be achieved without quantum errors. We numerically evaluated the performance of the technique. The calculated expectation value for the Bell state under stochastic and unitary errors indicate our methods can suppress the bias more efficiently than the existing methods. Thus, our method can be used as a QEM methods for unknown errors with a mechanism similar to VD with resolving its drawbacks. |
Monday, March 4, 2024 3:48PM - 4:00PM |
D52.00005: Quantum Computing of a Frenkel Hamiltonian with Deep Learning-Based Error Mitigation Yi-Ting Lee, Vijaya Begum-Hudde, Barbara Jones, Andre Schleife In recent decades, the development of quantum computers and quantum algorithms has made significantly progress in the physics and chemistry communities. Among various studied systems, one notable system that has not been simulated by quantum computers is the Frenkel-Davydov (FD) Hamiltonian, typically used to model the excitonic effects in organic solids. In this study, we used a single layer anthracene as a toy model and employed variational quantum deflation (VQD) to compute all the eigenvalues and associated observables. To address errors introduced by noisy qubits, we developed a novel error mitigation framework that incorporates post-selection (PS) gates and a deep learning (DL) technique to obtain the mitigated expectation value. We first demonstrate excellent agreement between the eigenvalues calculated on a noiseless simulator and those obtained through exact diagonalization, confirming the validity of our VQD algorithm. Furthermore, we have observed the effective mitigation of noisy results using our PS+DL technique for systems with more than 8 qubits on a noisy simulator. Remarkably, this technique remains effective for real hardware, even when calibrations occur during the simulations. Our results promise to contribute to the development of excited state discovery and new mitigation techniques within the quantum computing community. |
Monday, March 4, 2024 4:00PM - 4:12PM |
D52.00006: Automating physical experiments via Large Language Models: an attempt on superconducting quantum processors Zijian Zhang, Shuxiang Cao, Mohammed Alghadeer, Simone D Fasciati, Michele Piscitelli, Mustafa S Bakr, Peter J Leek, Alán Aspuru-Guzik Large Language Models has been recorganized a substantial advancement in the AI domain, driving new applications in content creation and code development fields. In this study, we introduce our attempt to employ Large Language Models in establishing an autonomous environment for the calibration processes of superconducting qubits, based on retrival argumented generation. We compare the performance of different large language models on analysing the experimental data and making decisions on the subsequent steps in the experiment. |
Monday, March 4, 2024 4:12PM - 4:48PM |
D52.00007: What is optimal way to mitigate/suppress quantum errors? Invited Speaker: Nobuyuki Yoshioka Quantum error mitigation methods are designed to eliminate the effect of noise in quantum computation by introducing a trade-off between bias and variance, using modified quantum circuits and classical postprocessing. While various techniques have been proposed with their own advantages and disadvantages, there is still no universal criterion to choose select the best method for a given application. In this talk, we focus on the sample complexity, or the measurement overhead, to perform unbiased quantum error mitigation, and discuss the performance bounds and their achievability. We show that, based on quantum estimation theory, the overhead generally grows exponentially with the circuit depth and also with the number of qubits in scrambling quantum circuits. We then show that the bounds on the overhead are provably tight under white noise, and that a simple rescaling technique achieves cost-optimality. Based on numerical simulations, we argue that a wide class of unital and nonunital noise are converted into white noise under sufficiently deep scrambling quantum circuits. This implies that, our findings become increasingly important when the error rate is reduced by hardware advancement or implementation of error correction. In this context, we also also discuss how to suppress algorithmic errors in a cost-optimal way under the framework of fault-tolerant quantum computing. If time permits, we also discuss the impact of realizing cost-optimality on resource estimation required for practical quantum advantage. |
Monday, March 4, 2024 4:48PM - 5:00PM |
D52.00008: Confirmation of Spatiotemporally-correlated Qubit Errors from Cosmic Rays Patrick M Harrington, Mingyu Li, Max Hays, Wouter Van De Pontseele, Daniel Mayer, Michael A Gingras, Bethany M Niedzielski, Hannah M Stickler, Jonilyn L Yoder, Mollie E Schwartz, Jeffrey A Grover, Kyle Serniak, Joseph A Formaggio, William D Oliver Quantum computation at scale requires sufficiently infrequent and sparse errors in time and space. In superconducting qubit processors, moments of spatiotemporally-correlated errors are believed to result from ionizing radiation interacting with the device. Here, we demonstrate unambiguous detection of qubit relaxation from cosmic rays, enabled by concurrent monitoring of qubit errors with scintillating radiation detectors and by adopting coincidence-timing techniques from nuclear and high energy physics. We identify the correlation length and duration of spatiotemporally-correlated qubit relaxation events from cosmic rays and estimate the occurrence rate of these events. While most events are from non-cosmogenic sources, such as ionizing radiation from the laboratory environment, we observe that our device is sensitive to nearly all cosmic ray impacts, which marks a possible challenge to future implementations of quantum error correction. These observations underscore a necessity to create quantum devices that are insensitive to all sources of ionizing radiation and highlights a benefit of operating superconducting qubit arrays in a low-background radiation environment. |
Monday, March 4, 2024 5:00PM - 5:12PM |
D52.00009: Modeling and testing novel quasiparticle traps on superconducting qubits via geometric gap engineering Ugur Alyanak, Ziwen Huang, Wei-Ting Lin, Xinyuan You, Alexander Romanenko, Anna Grassellino, Young-Kee Kim, Shaojiang Zhu Non-equilibrium quasiparticles have been identified as a major decoherence source in su- |
Monday, March 4, 2024 5:12PM - 5:24PM |
D52.00010: Modeling Particle Impacts and Quasiparticle Poisoning in Superconducting Qubit Arrays Eric Yelton, Clayton Larson, Kenneth R Dodge, Vito M Iaia, Gianluigi Catelani, Noah Kurinsky, Paul G Baity, Robert McDermott, B.L.T. Plourde Correlated errors in qubit arrays are catastrophic for preserving quantum information. Background radiation incident on superconducting qubit devices can create large bursts of pair-breaking phonons that generate quasiparticles (QPs) in the device layer. These energetic events lead to correlated errors in superconducting qubit arrays. To better understand this process and potential mitigation strategies, we model these events in a Monte-Carlo based simulation package called Geant4 Condensed Matter Physics (G4CMP). We model various device configurations of ground plane and back-side films. From this modeling we compute the normalized QP density as a function of position after an impact in dense superconducting qubit arrays. Using these results, we characterize the spatial and temporal dynamics of the QP poisoning footprint following a particle impact for various device configurations. These results help to inform future superconducting qubit device design. |
Monday, March 4, 2024 5:24PM - 5:36PM |
D52.00011: Optimal Experimental Design for Quantum Circuits:A case study for the mitigation of quasiparticle poisoning Paul G Baity, Natalie Isenberg, Kristofer Reyes, Byung-Jun Yoon, Gilchan Park, Eric Yelton, Clayton Larson, Spencer Weeden, Gabriel Spahn, Robert McDermott, Britton L Plourde, Nathan Urban, Adolfy Hoisie Superconducting circuits are the most widely used system for quantum computing, and many of the key accomplishments in this field have been performed using superconducting-based quantum processors. The improvement of device performance through the mitigation and control of qubit errors is currently a critical step for executing complex quantum algorithms. However, the experimental routines used to test error mitigation strategies are often pointwise, time-consuming, and costly. Instead of an unguided research strategy, Optimal Experimental Design (OED) is a methodological approach used to maximize the information gained from a limited number of iteratively designed, computationally optimized experiments. We are developing such a methodology for the design of quantum circuits. To demonstrate the benefits of our OED methodology, we apply it to the problem of quasiparticle poisoning (QP) in superconducting quantum devices, wherein environmental high-energy radiation creates qubit errors. Our OED methodology uses key advances in the computational simulation of QP physics and high-performance-computing optimization techniques to design quantum chips that are more resilient against QP-induced errors. Present results and ongoing work will be discussed. |
Monday, March 4, 2024 5:36PM - 5:48PM |
D52.00012: Impact of Ionizing Radiation on Energy Relaxation in Superconducting Fluxonium Qubits Nicolas Gosling, Francesco DeDominicis, Martin Spiecker, Patrick Paluch, Ambra Mariani, Nicolas Zapata, Ivan Colantoni, Angelo Cruciani, Nicolas Casali, Stefano Pirro, Laura Cardani, Ioan-Mihai Pop Measuring and quantifying the influence of environmental factors on superconducting qubit performance is a powerful tool to shed light on their decoherence mechanisms. We explore the effect of ionizing radiation on a superconducting fluxonium qubit, focusing on the statistics of quantum jumps. We operate the qubit in a deep underground facility, where it is protected from cosmic radiation. Moreover, we partially shield the cryostat from local ionizing sources by using a led brick shield. This allows us to start from a low background and to controllably increase the level of radioactivity experienced by the qubit. We show that the statistics of quantum jumps changes with radiation exposure. Furthermore, the long-lived two level system bath, which in our case forms the dominant loss mechanism of the qubit, is unaffected by exposure to background radiation. |
Monday, March 4, 2024 5:48PM - 6:00PM |
D52.00013: Characterizing frequency fluctuations induced non-Markovian noise in superconducting qubits Abhishek Agarwal, Lachlan Lindoy, Yannic Rath, Deep M Lall, Ivan Rungger Non-Markovian noise is an important source of nosie in superconducting qubits available today. In this presentation we show that including the effects of non-Markovian noise allows us to have a model that can accurately capture the device physics. We further develop a method to perform experiments on a superconducting qubit quantum computer to resolve qubit frequency fluctuations at different time scales, and show that the frequency fluctuations are the dominant source of observed non-Markovian noise in the device. The methods allow us to see the effects of quasiparticle induced charge parity fluctuations as well as frequency fluctuations due to two level fluctuators. We analyse the magnitude, rate, and symmetry of the charge parity fluctuations, as well as fluctuations in the charge parity frequency splitting. The understanding of these non-Markovian noise sources provided by our model and by our experiments allow to optimize the calibration of the devices and to further mitigate the effects of noise. |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700