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:2611.01520

Near-Optimal Learning of Quantum Many-Body Systems from Gibbs and Thermal Metastable States via the Stationarity Test

Tai Hsuan Yang

quant-phmath-phSubmitted Nov 1, 2026• Announced Nov 3, 2026• 0 citations
arXivHTMLPDF

Abstract

We study the problem of learning a finite-range quantum Hamiltonian on an $n$-qubit $D$-dimensional lattice from copies of its Gibbs states at a known inverse temperature $β$ or its thermal metastable states. Existing efficient algorithms at arbitrary temperature have sample guarantees with an $e^{\poly(β)}$ dependence, leaving open whether the exponent can be made linear in $β$. We achieve this improvement: for $β\ge1$, our algorithm learns all Hamiltonian coefficients to $\ell_\infty$-error $ε$ with probability at least $1-δ$ using $\tO\bigl(e^{O(β)}ε^{-2}\log(n/δ)\bigr)$ copies and total time $\tO\bigl(n e^{O(β)}ε^{-2}\log(n/δ)\bigr)$. The measurements act on individual copies and are supported on balls of radius $O\bigl(β^2+β\log(1/ε)\bigr)$. Our sample complexity achieves the optimal exponential order in $β$, with optimal $ε^{-2}$ dependence up to logarithmic factors. Building on the stationarity-test framework \citep{Bergamaschi2026Stationarity}, we prove a unified Fisher-information coercivity bound showing that errors in the candidate Hamiltonian produce a detectable signal with only an $e^{-O(β)}$ loss in sensitivity. The key is a smooth frequency filter whose bandwidth remains independent of temperature, together with sharper estimates relating this signal to the Hamiltonian coefficients. For thermal metastable states, the guarantee holds under local drift $ζ\le e^{-Cβ^{2D+1}}$, for a geometry-dependent constant $C>0$, and above an accuracy floor $\tO(e^{O(β)}ζ)$. The argument applies to both Gibbs and thermal metastable states and, combined with local measurements and projected stochastic gradient descent (SGD), yields both learning guarantees.

AI Summary

The paper asks whether finite-range lattice Hamiltonians can be learned from thermal states with an $e^{O(\beta)}$ temperature dependence, rather than existing $e^{\mathrm{poly}(\beta)}$ guarantees. It considers $n$ qubits in fixed dimension $D$, known inverse temperature, and a known Pauli interaction dictionary with bounded range, weight, degree, and coefficients. For $\beta\ge1$, it gives an algorithm recovering every coefficient to additive error $\epsilon$, with probability $1-\delta$, using $\widetilde O(e^{O(\beta)}\epsilon^{-2}\log(n/\delta))$ copies and $\widetilde O(ne^{O(\beta)}\epsilon^{-2}\log(n/\delta))$ total runtime. Measurements use individual copies but require coherent operations within balls of radius $O(\beta^2+\beta\log(1/\epsilon))$. The bounds match the exponential order of known worst-case temperature lower bounds and the $\epsilon^{-2}$ dependence up to logarithmic factors. The method builds on stationarity testing: local observable drifts under a candidate detailed-balanced Gibbs sampler provide parameter-update directions. A smooth frequency filter with temperature-independent bandwidth yields a Fisher-information coercivity bound with sensitivity $e^{-O(\beta)}$. Local short-time sampler evolution estimates these drifts; projected stochastic updates and a spatial dependency-cone analysis avoid system-size-dependent sampling variance. No global mixing, spectral-gap, or correlation-decay assumption is required. The same guarantees extend to thermal metastable states, provided each site-summed sampler generator has trace-norm drift at most $\zeta\le e^{-C\beta^{2D+1}}$ and $\epsilon\gtrsim\widetilde O(e^{O(\beta)}\zeta)$. Stated open limitations include relaxing this restrictive metastability promise, reducing measurement radius, sharpening exponential constants, and learning unknown interaction structure.

Almost certainly largely AI-generatedAI use disclosedStrong assumptions
The proof was generated with OpenAI GPT 6 Astra Ultra. Gemini 4 Argon was used to assist with writing the manuscript.

Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.

Comments
43 pages, 1 table

Related papers

The stationarity test: a framework for learning quantum many-body systems from their thermal states

Thiago Bergamaschi

The task of learning the Hamiltonian interactions governing a quantum system, given samples of its thermal (or 'Gibbs') states, is a foundational question at the intersection of quantum learning theory and many-body physics. In this paper, we draw connections to the quantum Gibbs sampling literature to introduce a natural learning algorithm we call the stationarity test: which simply "guesses" the Hamiltonian, and measures the rate-of-change of local observables, under the associated detailed-balanced quantum Markov chain [CKG23]. We leverage the stationarity test to address the following applications: 1. To give the first learning algorithm for the underlying interaction graph, i.e. structure learning, of lattice Hamiltonians at all temperatures, given copies of their Gibbs states. 2. To give the first learning algorithm for the coefficients of a lattice Hamiltonian, given only copies of its thermal metastable states, modeled as the "local minima" of the free energy. 3. In addition, we present a refinement to the recent area law for thermal metastable states [BCV25], which holds in the thermodynamic limit. Our learning algorithms are rigorous, time-efficient, and nearly sample-optimal in system size and accuracy. At a technical level, our arguments are based on new approximate locality and convexity properties for the quantum Fisher information of these Gibbs sampling algorithms.

quant-pharXiv:2610.01074•Oct 2, 2026•1 citation

Efficient learning of quantum interactions from thermal metastable states

Bingrun Wang, Qi Ye, Chi-Fang Chen

Learning quantum interactions from finite-temperature many-body systems is a central task in emerging quantum platforms. Recently, the problem of learning from lattice quantum Gibbs states has found rigorous, efficient protocols. Nevertheless, exact Gibbs states, as the input premise, are in fact computationally intractable to prepare and may not faithfully represent generic finite-temperature quantum systems. In contrast, a system coupled to a heat bath can be stuck at an approximate stationary state (metastable state) long before it truly equilibrates. Here, we formulate a physically and algorithmically consistent alternative: learning from such metastable states of detailed-balanced master equations (Lindbladians) arising from system-bath interactions. We distill the algorithmic mechanism and structural condition underlying Gibbs-state learning and extend it in full to metastable states, attaining nearly optimal sample and computational complexity (in the system size and the precision). More broadly, we sharpen notions of metastability and develop a unified framework for finite-temperature learning.

quant-pharXiv:2610.01538•Oct 2, 2026•2 citations

Learning quantum Gibbs states locally and efficiently

Chi-Fang Chen, Anurag Anshu, Quynh T. Nguyen

Learning the Hamiltonian underlying a quantum many-body system in thermal equilibrium is a fundamental task in quantum learning theory and experimental sciences. To learn the Gibbs state of local Hamiltonians at any constant inverse temperature, the state-of-the-art provable algorithms fall short of the optimal sample and computational complexity, in sharp contrast with the locality and simplicity in the classical cases. In this work, we present a learning algorithm that learns each local term of a n-qubit Hamiltonian on any bounded-degree graph to a constant additive error with the optimal sample complexity $\mathcal{O}(\log n)$. The protocol uses parallelizable local quantum measurements that act within bounded neighborhoods of the graph and near-linear-time classical post-processing. We also give a learning algorithm for lattice Hamiltonians with near-optimal scaling on the learning precision and the inverse temperature. At the heart of our algorithm is the interplay between locality, the Kubo-MartinSchwinger condition, and the operator Fourier transform at arbitrary temperatures.

arXiv:2504.02706•Annual Symposium on Foundations of Computer Science•Dec 14, 2025•26 citations

Learning the structure of open quantum systems

Laura Lewis, Ewin Tang, John Wright

We design an algorithm for learning the coefficients of an $n$-qubit constant-local Lindbladian to $\varepsilon$ error with $O(g d^2 \log(n) / \varepsilon^2)$ total evolution time, where $g$ is the single-site energy and $d$ is the (approximate) degree of the interaction graph. Though Lindbladians present new challenges not present in the special case of Hamiltonians, our algorithm achieves the suite of desiderata attained by state-of-the-art Hamiltonian learning algorithms: (1) it uses non-adaptive, ancilla-free randomized Pauli measurement circuits with a time resolution of only $Θ(1/g)$; (2) it works without knowledge of the structure of the unknown Lindbladian; (3) it depends on a smooth form of degree, thereby supporting the learning of quasi-local and power-law Lindbladians. Moreover, we prove a lower bound showing that our algorithm is optimal in each parameter up to logarithmic factors. Our algorithm is a simple iterative method, where the objective function consists of Fourier coefficients of the Lindbladian restricted to few-site regions. Its analysis identifies the difficulty unique to open systems, which we call "confusing" terms. For settings where the "confusion" is limited, the performance of the algorithm improves. We demonstrate this for the case of structure learning of Hamiltonians from access to real-time evolution, where we obtain a new algorithm that is significantly simpler than previous work. In addition, using the same iterative method, we design the first efficient algorithm for structure learning Hamiltonians from high-temperature Gibbs states.

quant-pharXiv:2606.30358•Jun 30, 2026•13 citations

Optimal learning of quantum Hamiltonians from high-temperature Gibbs states

Jeongwan Haah, Robin Kothari, Ewin Tang

We study the problem of learning a Hamiltonian H to precision $\varepsilon$, supposing we are given copies of its Gibbs state $\rho =\exp(-\beta H)/\mathrm{Tr}(\exp(-\beta H))$ at a known inverse temperature $\beta$. Anshu, Arunachalam, Kuwahara, and Soleimanifar [AAKS21] recently studied the sample complexity (number of copies of $\rho$ needed) of this problem for geometrically local N-qubit Hamiltonians. In the high-temperature (low $\beta$) regime, their algorithm has sample complexity poly (N, 1/$\beta$, 1/$\varepsilon$) and can be implemented with polynomial, but suboptimal, time complexity. In this paper, we study the same question for a more general class of Hamiltonians. We show how to learn the coefficients of a Hamiltonian to error $\varepsilon$ with sample complexity $S=O(\log N/(\beta\varepsilon)^{2}$) and time complexity linear in the sample size, O(SN). Furthermore, we prove a matching lower bound showing that our algorithm’s sample complexity is optimal, and hence our time complexity is also optimal. In the appendix, we show that virtually the same algorithm can be used to learn H from a real-time evolution unitary $e^{-i t H}$ in a small t regime with similar sample and time complexity.

Annual Symposium on Foundations of Computer Science•Oct 1, 2022•49 citations

Quantum score matching with applications to learning thermal states

Yulong Dong, Jiaqi Leng

Score matching has driven major advances in classical generative learning by enabling models to learn from data without evaluating intractable normalization constants, or partition functions. Yet, extending this principle to quantum learning requires rethinking its foundations, as quantum states are described by noncommuting density operators rather than scalar probabilities. The noncommutativity creates fundamental challenges not only in defining quantum scores, but also in developing a training framework with efficient circuit implementations and rigorous theoretical guarantees. In this work, we bridge this gap by establishing a general quantum score-matching framework with end-to-end theoretical guarantees. Applied to Gibbs-state learning, our approach avoids additional thermal-state preparation and achieves information-theoretically optimal sample complexity in the high-temperature regime for Hamiltonians with bounded locality and interaction degree. This positions score matching as a new route to state-of-the-art performance in learning quantum Gibbs states. Beyond these theoretical results, numerical simulations show that our method remains effective even when gradients are estimated inaccurately under limited measurement budgets. Experiments on IBM quantum hardware further demonstrate that quantum score matching is NISQ-friendly: without any error mitigation or correction, it reduces the relative Hamiltonian-parameter error from 64% to approximately 10%. Together, these results extend score matching into an experimentally realizable paradigm for quantum-state learning.

quant-pharXiv:2609.28391•Sep 24, 2026•2 citations
We study the problem of learning a finite-range quantum Hamiltonian on an nnn-qubit DDD-dimensional lattice from copies of its Gibbs states at a known inverse temperature βββ or its thermal metastable states. Existing efficient algorithms at arbitrary temperature have sample guarantees with an e\poly(β)e^{\poly(β)}e\poly(β) dependence, leaving open whether the exponent can be made linear in βββ. We achieve this improvement: for β≥1β\ge1β≥1, our algorithm learns all Hamiltonian coefficients to ℓ∞\ell_\inftyℓ∞​-error εεε with probability at least 1−δ1-δ1−δ using \tO(eO(β)ε−2log⁡(n/δ))\tO\bigl(e^{O(β)}ε^{-2}\log(n/δ)\bigr)\tO(eO(β)ε−2log(n/δ)) copies and total time \tO(neO(β)ε−2log⁡(n/δ))\tO\bigl(n e^{O(β)}ε^{-2}\log(n/δ)\bigr)\tO(neO(β)ε−2log(n/δ)). The measurements act on individual copies and are supported on balls of radius O(β2+βlog⁡(1/ε))O\bigl(β^2+β\log(1/ε)\bigr)O(β2+βlog(1/ε)). Our sample complexity achieves the optimal exponential order in βββ, with optimal ε−2ε^{-2}ε−2 dependence up to logarithmic factors.
Building on the stationarity-test framework \citep{Bergamaschi2026Stationarity}, we prove a unified Fisher-information coercivity bound showing that errors in the candidate Hamiltonian produce a detectable signal with only an e−O(β)e^{-O(β)}e−O(β) loss in sensitivity. The key is a smooth frequency filter whose bandwidth remains independent of temperature, together with sharper estimates relating this signal to the Hamiltonian coefficients. For thermal metastable states, the guarantee holds under local drift ζ≤e−Cβ2D+1ζ\le e^{-Cβ^{2D+1}}ζ≤e−Cβ2D+1, for a geometry-dependent constant C>0C>0C>0, and above an accuracy floor \tO(eO(β)ζ)\tO(e^{O(β)}ζ)\tO(eO(β)ζ). The argument applies to both Gibbs and thermal metastable states and, combined with local measurements and projected stochastic gradient descent (SGD), yields both learning guarantees.
The paper asks whether finite-range lattice Hamiltonians can be learned from thermal states with an eO(β)e^{O(\beta)}eO(β) temperature dependence, rather than existing epoly(β)e^{\mathrm{poly}(\beta)}epoly(β) guarantees. It considers nnn qubits in fixed dimension DDD, known inverse temperature, and a known Pauli interaction dictionary with bounded range, weight, degree, and coefficients.
For β≥1\beta\ge1β≥1, it gives an algorithm recovering every coefficient to additive error ϵ\epsilonϵ, with probability 1−δ1-\delta1−δ, using O~(eO(β)ϵ−2log⁡(n/δ))\widetilde O(e^{O(\beta)}\epsilon^{-2}\log(n/\delta))O(eO(β)ϵ−2log(n/δ)) copies and O~(neO(β)ϵ−2log⁡(n/δ))\widetilde O(ne^{O(\beta)}\epsilon^{-2}\log(n/\delta))O(neO(β)ϵ−2log(n/δ)) total runtime. Measurements use individual copies but require coherent operations within balls of radius O(β2+βlog⁡(1/ϵ))O(\beta^2+\beta\log(1/\epsilon))O(β2+βlog(1/ϵ)). The bounds match the exponential order of known worst-case temperature lower bounds and the ϵ−2\epsilon^{-2}ϵ−2 dependence up to logarithmic factors.
The method builds on stationarity testing: local observable drifts under a candidate detailed-balanced Gibbs sampler provide parameter-update directions. A smooth frequency filter with temperature-independent bandwidth yields a Fisher-information coercivity bound with sensitivity e−O(β)e^{-O(\beta)}e−O(β). Local short-time sampler evolution estimates these drifts; projected stochastic updates and a spatial dependency-cone analysis avoid system-size-dependent sampling variance. No global mixing, spectral-gap, or correlation-decay assumption is required.
The same guarantees extend to thermal metastable states, provided each site-summed sampler generator has trace-norm drift at most ζ≤e−Cβ2D+1\zeta\le e^{-C\beta^{2D+1}}ζ≤e−Cβ2D+1 and ϵ≳O~(eO(β)ζ)\epsilon\gtrsim\widetilde O(e^{O(\beta)}\zeta)ϵ≳O(eO(β)ζ). Stated open limitations include relaxing this restrictive metastability promise, reducing measurement radius, sharpening exponential constants, and learning unknown interaction structure.
The task of learning the Hamiltonian interactions governing a quantum system, given samples of its thermal (or 'Gibbs') states, is a foundational question at the intersection of quantum learning theory and many-body physics. In this paper, we draw connections to the quantum Gibbs sampling literature to introduce a natural learning algorithm we call the stationarity test: which simply "guesses" the Hamiltonian, and measures the rate-of-change of local observables, under the associated detailed-balanced quantum Markov chain [CKG23]. We leverage the stationarity test to address the following applications: 1. To give the first learning algorithm for the underlying interaction graph, i.e. structure learning, of lattice Hamiltonians at all temperatures, given copies of their Gibbs states. 2. To give the first learning algorithm for the coefficients of a lattice Hamiltonian, given only copies of its thermal metastable states, modeled as the "local minima" of the free energy. 3. In addition, we present a refinement to the recent area law for thermal metastable states [BCV25], which holds in the thermodynamic limit. Our learning algorithms are rigorous, time-efficient, and nearly sample-optimal in system size and accuracy. At a technical level, our arguments are based on new approximate locality and convexity properties for the quantum Fisher information of these Gibbs sampling algorithms.
Learning quantum interactions from finite-temperature many-body systems is a central task in emerging quantum platforms. Recently, the problem of learning from lattice quantum Gibbs states has found rigorous, efficient protocols. Nevertheless, exact Gibbs states, as the input premise, are in fact computationally intractable to prepare and may not faithfully represent generic finite-temperature quantum systems. In contrast, a system coupled to a heat bath can be stuck at an approximate stationary state (metastable state) long before it truly equilibrates. Here, we formulate a physically and algorithmically consistent alternative: learning from such metastable states of detailed-balanced master equations (Lindbladians) arising from system-bath interactions. We distill the algorithmic mechanism and structural condition underlying Gibbs-state learning and extend it in full to metastable states, attaining nearly optimal sample and computational complexity (in the system size and the precision). More broadly, we sharpen notions of metastability and develop a unified framework for finite-temperature learning.
Learning the Hamiltonian underlying a quantum many-body system in thermal equilibrium is a fundamental task in quantum learning theory and experimental sciences. To learn the Gibbs state of local Hamiltonians at any constant inverse temperature, the state-of-the-art provable algorithms fall short of the optimal sample and computational complexity, in sharp contrast with the locality and simplicity in the classical cases. In this work, we present a learning algorithm that learns each local term of a n-qubit Hamiltonian on any bounded-degree graph to a constant additive error with the optimal sample complexity O(log⁡n)\mathcal{O}(\log n)O(logn). The protocol uses parallelizable local quantum measurements that act within bounded neighborhoods of the graph and near-linear-time classical post-processing. We also give a learning algorithm for lattice Hamiltonians with near-optimal scaling on the learning precision and the inverse temperature. At the heart of our algorithm is the interplay between locality, the Kubo-MartinSchwinger condition, and the operator Fourier transform at arbitrary temperatures.
We design an algorithm for learning the coefficients of an nnn-qubit constant-local Lindbladian to ε\varepsilonε error with O(gd2log⁡(n)/ε2)O(g d^2 \log(n) / \varepsilon^2)O(gd2log(n)/ε2) total evolution time, where ggg is the single-site energy and ddd is the (approximate) degree of the interaction graph. Though Lindbladians present new challenges not present in the special case of Hamiltonians, our algorithm achieves the suite of desiderata attained by state-of-the-art Hamiltonian learning algorithms: (1) it uses non-adaptive, ancilla-free randomized Pauli measurement circuits with a time resolution of only Θ(1/g)Θ(1/g)Θ(1/g); (2) it works without knowledge of the structure of the unknown Lindbladian; (3) it depends on a smooth form of degree, thereby supporting the learning of quasi-local and power-law Lindbladians. Moreover, we prove a lower bound showing that our algorithm is optimal in each parameter up to logarithmic factors. Our algorithm is a simple iterative method, where the objective function consists of Fourier coefficients of the Lindbladian restricted to few-site regions. Its analysis identifies the difficulty unique to open systems, which we call "confusing" terms. For settings where the "confusion" is limited, the performance of the algorithm improves. We demonstrate this for the case of structure learning of Hamiltonians from access to real-time evolution, where we obtain a new algorithm that is significantly simpler than previous work. In addition, using the same iterative method, we design the first efficient algorithm for structure learning Hamiltonians from high-temperature Gibbs states.
We study the problem of learning a Hamiltonian H to precision ε\varepsilonε, supposing we are given copies of its Gibbs state ρ=exp⁡(−βH)/Tr(exp⁡(−βH))\rho =\exp(-\beta H)/\mathrm{Tr}(\exp(-\beta H))ρ=exp(−βH)/Tr(exp(−βH)) at a known inverse temperature β\betaβ. Anshu, Arunachalam, Kuwahara, and Soleimanifar [AAKS21] recently studied the sample complexity (number of copies of ρ\rhoρ needed) of this problem for geometrically local N-qubit Hamiltonians. In the high-temperature (low β\betaβ) regime, their algorithm has sample complexity poly (N, 1/β\betaβ, 1/ε\varepsilonε) and can be implemented with polynomial, but suboptimal, time complexity. In this paper, we study the same question for a more general class of Hamiltonians. We show how to learn the coefficients of a Hamiltonian to error ε\varepsilonε with sample complexity S=O(log⁡N/(βε)2S=O(\log N/(\beta\varepsilon)^{2}S=O(logN/(βε)2) and time complexity linear in the sample size, O(SN). Furthermore, we prove a matching lower bound showing that our algorithm’s sample complexity is optimal, and hence our time complexity is also optimal. In the appendix, we show that virtually the same algorithm can be used to learn H from a real-time evolution unitary e−itHe^{-i t H}e−itH in a small t regime with similar sample and time complexity.
Score matching has driven major advances in classical generative learning by enabling models to learn from data without evaluating intractable normalization constants, or partition functions. Yet, extending this principle to quantum learning requires rethinking its foundations, as quantum states are described by noncommuting density operators rather than scalar probabilities. The noncommutativity creates fundamental challenges not only in defining quantum scores, but also in developing a training framework with efficient circuit implementations and rigorous theoretical guarantees. In this work, we bridge this gap by establishing a general quantum score-matching framework with end-to-end theoretical guarantees. Applied to Gibbs-state learning, our approach avoids additional thermal-state preparation and achieves information-theoretically optimal sample complexity in the high-temperature regime for Hamiltonians with bounded locality and interaction degree. This positions score matching as a new route to state-of-the-art performance in learning quantum Gibbs states. Beyond these theoretical results, numerical simulations show that our method remains effective even when gradients are estimated inaccurately under limited measurement budgets. Experiments on IBM quantum hardware further demonstrate that quantum score matching is NISQ-friendly: without any error mitigation or correction, it reduces the relative Hamiltonian-parameter error from 64% to approximately 10%. Together, these results extend score matching into an experimentally realizable paradigm for quantum-state learning.