Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2312.13473 · Machine Learning Science and Technology

Accuracy vs memory advantage in the quantum simulation of stochastic processes

Leonardo Banchi

Published Apr 26, 2024• 8 citations
arXivHTMLPDF

Abstract

Abstract Many inference scenarios rely on extracting relevant information from known data in order to make future predictions. When the underlying stochastic process satisfies certain assumptions, there is a direct mapping between its exact classical and quantum simulators, with the latter asymptotically using less memory. Here we focus on studying whether such quantum advantage persists when those assumptions are not satisfied, and the model is doomed to have imperfect accuracy. By studying the trade-off between accuracy and memory requirements, we show that quantum models can reach the same accuracy with less memory, or alternatively, better accuracy with the same memory. Finally, we discuss the implications of this result for learning tasks.

DOI
10.1088/2632-2153/ad444a

Related papers

Advantages and learning for quantum modelling

Liu, Qing

Simulating stochastic processes using less resources is a key pursuit in many sciences. This involves identifying and extracting the past information relevant to the process' future behavior and formulating `predictive models' for inferring the latter from the former. Significant efforts have been made towards finding the optimal predictive models that require the least amount of information. Quantum technologies offer a promising means to this end, allowing equally accurate future predictions whilst storing less past information. In this thesis, I explore three aspects of research related to this idea. First, we demonstrate quantum models can have unbounded memory advantage by studying a family of stochastic processes where a random walk on real numbers is modelled with progressively greater precision. Next, we document the optimal quantum models that generate predictions using unitary circuits. We use these to document `ambiguity of optimality', a uniquely quantum phenomena where the optimal model depends on whether it is used in i.i.d versus single-shot settings. Finally, we look at a related problem of learning how a desired unitary can be synthesized using unknown pulses. Together, these development help to further understand the properties and power of quantum models, as well as build towards tools for their synthesis.

•Jan 1, 2021•0 citations

Superior memory efficiency of quantum devices for the simulation of continuous-time stochastic processes

Thomas J. Elliott, Mile Gu

Abstract Continuous-time stochastic processes pervade everyday experience, and the simulation of models of these processes is of great utility. Classical models of systems operating in continuous-time must typically track an unbounded amount of information about past behaviour, even for relatively simple models, enforcing limits on precision due to the finite memory of the machine. However, quantum machines can require less information about the past than even their optimal classical counterparts to simulate the future of discrete-time processes, and we demonstrate that this advantage extends to the continuous-time regime. Moreover, we show that this reduction in the memory requirement can be unboundedly large, allowing for arbitrary precision even with a finite quantum memory. We provide a systematic method for finding superior quantum constructions, and a protocol for analogue simulation of continuous-time renewal processes with a quantum machine.

arXiv:1704.04231•npj Quantum Information•Feb 16, 2018•40 citations

Robust inference of memory structure for efficient quantum modeling of stochastic processes

Matthew Ho, Mile Gu, Thomas J. Elliott

A growing body of work has established the modeling of stochastic processes as a promising area of application for quantum technologies; it has been shown that quantum models are able to replicate the future statistics of a stochastic process while retaining less information about the past than any classical model must, even for a purely classical process. Such memory-efficient models open a potential future route to study complex systems in greater detail than ever before and suggest profound consequences for our notions of structure in their dynamics. Yet, to date methods for constructing these quantum models are based on having a prior knowledge of the optimal classical model. Here, we introduce a protocol for blind inference of the memory structure of quantum models---tailored to take advantage of quantum features---direct from time-series data, in the process highlighting the robustness of their structure to noise. This in turn provides a way to construct memory-efficient quantum models of stochastic processes while circumventing certain drawbacks that manifest solely as a result of classical information processing in classical inference protocols.

arXiv:1911.02767•Physical Review A•Mar 19, 2020•14 citations

Provably superior accuracy in quantum stochastic modeling

Chengran Yang, Andrew J. P. Garner, Feiyang Liu +5 more

In the design of stochastic models, there is a constant trade-off between model complexity and accuracy. Here we prove that quantum models enable a more favorable trade-off. We present a technique for identifying fundamental upper bounds on the predictive accuracy of dimensionality-constrained classical models. We identify quantum models that surpass this bound by creating an algorithm that learns quantum models given time-series data. We demonstrate that this quantum accuracy advantage is attainable in a present-day noisy quantum device. These results illustrate the immediate relevance of quantum technologies to time-series analysis and offer an instance where their resulting accuracy advantage can be provably established.

arXiv:2105.14434•Physical Review A•Aug 10, 2023•11 citations

Provably unbounded memory advantage in stochastic simulation using quantum mechanics

Andrew J. P. Garner, Qing Liu, Jayne Thompson +2 more

Simulating the stochastic evolution of real quantities on a digital computer requires a trade-off between the precision to which these quantities are approximated, and the memory required to store them. The statistical accuracy of the simulation is thus generally limited by the internal memory available to the simulator. Here, using tools from computational mechanics, we show that quantum processors with a fixed finite memory can simulate stochastic processes of real variables to arbitrarily high precision. This demonstrates a provable, unbounded memory advantage that a quantum simulator can exhibit over its best possible classical counterpart.

arXiv:1609.04408•New Journal of Physics•Oct 12, 2017•32 citations

Dimension reduction in quantum sampling of stochastic processes

Chengran Yang, M. Florido-Llinas, Mile Gu +1 more

Abstract Quantum technologies offer a promising route to the efficient sampling and analysis of stochastic processes, with potential applications across the sciences. Such quantum advantages rely on the preparation of a quantum sample state of the stochastic process, which requires a memory system to propagate correlations between the past and future of the process. Here, we introduce a method of lossy quantum dimension reduction that allows this memory to be compressed, not just beyond classical limits, but also beyond current state-of-the-art quantum stochastic sampling approaches. We investigate the trade-off between the saving in memory resources from this compression, and the distortion it introduces. We show that our approach can be highly effective in low distortion compression of both Markovian and strongly non-Markovian processes alike. We further discuss the application of our results to quantum stochastic modelling more broadly.

arXiv:2404.10338•npj Quantum Information•Feb 28, 2025•5 citations
Abstract Many inference scenarios rely on extracting relevant information from known data in order to make future predictions. When the underlying stochastic process satisfies certain assumptions, there is a direct mapping between its exact classical and quantum simulators, with the latter asymptotically using less memory. Here we focus on studying whether such quantum advantage persists when those assumptions are not satisfied, and the model is doomed to have imperfect accuracy. By studying the trade-off between accuracy and memory requirements, we show that quantum models can reach the same accuracy with less memory, or alternatively, better accuracy with the same memory. Finally, we discuss the implications of this result for learning tasks.
Simulating stochastic processes using less resources is a key pursuit in many sciences. This involves identifying and extracting the past information relevant to the process' future behavior and formulating `predictive models' for inferring the latter from the former. Significant efforts have been made towards finding the optimal predictive models that require the least amount of information. Quantum technologies offer a promising means to this end, allowing equally accurate future predictions whilst storing less past information. In this thesis, I explore three aspects of research related to this idea. First, we demonstrate quantum models can have unbounded memory advantage by studying a family of stochastic processes where a random walk on real numbers is modelled with progressively greater precision. Next, we document the optimal quantum models that generate predictions using unitary circuits. We use these to document `ambiguity of optimality', a uniquely quantum phenomena where the optimal model depends on whether it is used in i.i.d versus single-shot settings. Finally, we look at a related problem of learning how a desired unitary can be synthesized using unknown pulses. Together, these development help to further understand the properties and power of quantum models, as well as build towards tools for their synthesis.
Abstract Continuous-time stochastic processes pervade everyday experience, and the simulation of models of these processes is of great utility. Classical models of systems operating in continuous-time must typically track an unbounded amount of information about past behaviour, even for relatively simple models, enforcing limits on precision due to the finite memory of the machine. However, quantum machines can require less information about the past than even their optimal classical counterparts to simulate the future of discrete-time processes, and we demonstrate that this advantage extends to the continuous-time regime. Moreover, we show that this reduction in the memory requirement can be unboundedly large, allowing for arbitrary precision even with a finite quantum memory. We provide a systematic method for finding superior quantum constructions, and a protocol for analogue simulation of continuous-time renewal processes with a quantum machine.
A growing body of work has established the modeling of stochastic processes as a promising area of application for quantum technologies; it has been shown that quantum models are able to replicate the future statistics of a stochastic process while retaining less information about the past than any classical model must, even for a purely classical process. Such memory-efficient models open a potential future route to study complex systems in greater detail than ever before and suggest profound consequences for our notions of structure in their dynamics. Yet, to date methods for constructing these quantum models are based on having a prior knowledge of the optimal classical model. Here, we introduce a protocol for blind inference of the memory structure of quantum models---tailored to take advantage of quantum features---direct from time-series data, in the process highlighting the robustness of their structure to noise. This in turn provides a way to construct memory-efficient quantum models of stochastic processes while circumventing certain drawbacks that manifest solely as a result of classical information processing in classical inference protocols.
In the design of stochastic models, there is a constant trade-off between model complexity and accuracy. Here we prove that quantum models enable a more favorable trade-off. We present a technique for identifying fundamental upper bounds on the predictive accuracy of dimensionality-constrained classical models. We identify quantum models that surpass this bound by creating an algorithm that learns quantum models given time-series data. We demonstrate that this quantum accuracy advantage is attainable in a present-day noisy quantum device. These results illustrate the immediate relevance of quantum technologies to time-series analysis and offer an instance where their resulting accuracy advantage can be provably established.
Simulating the stochastic evolution of real quantities on a digital computer requires a trade-off between the precision to which these quantities are approximated, and the memory required to store them. The statistical accuracy of the simulation is thus generally limited by the internal memory available to the simulator. Here, using tools from computational mechanics, we show that quantum processors with a fixed finite memory can simulate stochastic processes of real variables to arbitrarily high precision. This demonstrates a provable, unbounded memory advantage that a quantum simulator can exhibit over its best possible classical counterpart.
Abstract Quantum technologies offer a promising route to the efficient sampling and analysis of stochastic processes, with potential applications across the sciences. Such quantum advantages rely on the preparation of a quantum sample state of the stochastic process, which requires a memory system to propagate correlations between the past and future of the process. Here, we introduce a method of lossy quantum dimension reduction that allows this memory to be compressed, not just beyond classical limits, but also beyond current state-of-the-art quantum stochastic sampling approaches. We investigate the trade-off between the saving in memory resources from this compression, and the distortion it introduces. We show that our approach can be highly effective in low distortion compression of both Markovian and strongly non-Markovian processes alike. We further discuss the application of our results to quantum stochastic modelling more broadly.