The local characterization of global tensor network eigenstates
José Garre Rubio, András Molnár, Norbert Schuch, Frank Verstraete
Abstract
We study the conditions under which Matrix Product States (MPS) or Matrix Product Operators are exact eigenvectors of an extensive local operator, such as a Hamiltonian. By suitably choosing the local operator, this covers a wide range of settings: Exact eigenstates of Hamiltonians, including scar states, exact MPS trajectories for driven quantum systems, steady states of local Lindbladians, generalized symmetries of either Hamiltonians or density matrices, and many more. Our key result is that that a local, fixed-size equation -- namely, how a single term in the operator acts on a block of tensors -- provides a necessary and sufficient condition for exact solutions. This allows to characterize the full space of solutions in all of the aforementioned problems, and to identify them both analytically and numerically. We elaborate on the concrete application of this characterization to all of the aforementioned settings, and in particular exemplify the power of our local characterization by using it to recover the quantum group symmetries of the XXZ model. We also discuss applications to numerical algorithms with MPS and the generalization of our results to 2D, i.e., projected entangled pair states (PEPS).
AI Impact Assessments
(3 models)Scientific Impact Assessment
Core Contribution
This paper proves a necessary and sufficient local condition for an injective MPS to be an exact eigenstate of an extensive local operator. The key result (Theorem 1) states that an injective MPS |ψ_N⟩ satisfies O_N|ψ_N⟩ = E_N|ψ_N⟩ if and only if a local "telescoping" equation holds: the action of a single k-local term on k adjacent MPS tensors decomposes as a boundary difference involving an auxiliary tensor B. While the sufficiency direction (the telescoping sum cancels) has been used as an ad hoc ansatz across many works, the necessity direction—that any exact MPS eigenstate must admit such a decomposition—was previously unproven folklore. This closes a foundational gap in the MPS formalism.
The result is remarkably general: the operator O need not be Hermitian, the eigenvalue need not be extremal, and the MPS need not be the unique eigenstate. The equation is independent of system size N, meaning that verification at a single (sufficiently large) N guarantees the eigenstate property for all system sizes.
Methodological Rigor
The proof strategy is elegant and well-executed. The simplified argument in the main text uses the MPO representation of the time evolution operator e^{iOt}·S and the fundamental theorem of MPS, expanding order-by-order in t. The rigorous proof in the supplemental material proceeds through a sequence of four carefully chosen contractions of the eigenvalue equation with combinations of the left-inverse tensor X and conjugate tensors Ā, progressively extracting the local condition. The argument that both c = 0 and d = 0 (from comparing contractions with one versus two X tensors) is particularly clean. The characterization of the redundancy in B (Theorem 2) is also complete: B is unique up to addition of λA^{i_1}···A^{i_k}.
The requirement N ≥ 2L + 2k - 1 (where L is the injectivity length) is natural and clearly stated. The restriction to injective MPS is fundamental to the proof technique but is not overly limiting, since general MPS decompose into sums of injective MPS after blocking.
Potential Impact
The paper's unifying framework has broad applicability across multiple subfields:
1. Quantum many-body scars: The local characterization provides a systematic method to search for and classify scar states, which are exact MPS eigenstates in non-integrable systems. The framework recovers known constructions (e.g., PXP model scars with 2-site unit cells) and provides tools for discovering new ones.
2. MPO symmetries and integrability: The application to finding MPO symmetries of Hamiltonians is demonstrated concretely by recovering the quantum group SL(2)_q symmetry of the XXZ model. This is a non-trivial validation that also showcases the method's analytical power. Recent works on XYZ integrability and conserved quantities (refs [14-16]) already use the sufficient direction; this paper validates those approaches.
3. Dissipative systems: The characterization of Lindbladian steady states as MPDOs through a local equation opens systematic approaches to finding exactly solvable open quantum systems.
4. Time-dependent MPS: Reformulating the Schrödinger equation as a local condition on time-dependent MPS tensors connects to the theory of time-dependent variational principles.
5. Numerical methods: The conceptual connection to VUMPS—where the fixed-point condition is a tangent-plane-projected version of the telescoping identity—provides theoretical grounding for one of the most important MPS algorithms.
6. 2D extensions: The generalization to PEPS on square and hexagonal lattices, while currently limited to injective PEPS (sufficient condition for hexagonal), opens important directions for 2D quantum systems.
Timeliness & Relevance
This paper arrives at a moment of intense activity in several directions: quantum many-body scars have attracted enormous attention since ~2018; MPO symmetries and generalized symmetries are a hot topic connecting condensed matter to high-energy physics; and the push toward rigorous 2D tensor network methods is a major frontier. The paper provides foundational infrastructure that these communities need. The fact that several recent papers (2025-2026, refs [14-16, 19, 22]) already use the sufficient direction underscores the timeliness.
Strengths
Limitations
Overall Assessment
This is a foundational contribution to the theory of tensor networks that rigorously establishes what has long been assumed. Its unifying perspective across multiple physical settings and its clean mathematical formulation make it likely to become a standard reference. The main limitation is that its immediate practical impact beyond formalization is not yet demonstrated through novel discoveries, though the framework clearly enables systematic searches that could yield such discoveries.
Generated Mar 31, 2026
Comparison History (70)
Paper 1 provides a broadly applicable theoretical framework for characterizing exact tensor network eigenstates across numerous settings (Hamiltonians, scar states, Lindbladians, symmetries, driven systems), with extensions to 2D PEPS. Its generality and unifying nature impact multiple active research areas in quantum many-body physics and tensor network methods. Paper 2, while elegant in establishing conditions for the imaginary-time Mpemba effect, addresses a more specialized phenomenon with narrower scope. Paper 1's methodological contributions to both analytical and numerical tensor network approaches give it broader and deeper impact potential.
Paper 1 addresses a fundamental theoretical question about tensor network states with broad applicability across multiple areas of quantum physics (Hamiltonians, scar states, Lindbladians, symmetries, PEPS). Its key result—a local characterization providing necessary and sufficient conditions—is a powerful general framework that unifies many problems and enables both analytical and numerical advances. Paper 2, while useful for high-dimensional quantum circuit optimization with improved gate counts, addresses a more specialized problem with narrower scope. Paper 1's breadth of impact across condensed matter, quantum information, and mathematical physics gives it higher potential impact.
Paper 2 offers a broadly applicable, conceptually unifying result: a necessary-and-sufficient local condition for global tensor-network eigenstates across Hamiltonians, Floquet/driven dynamics, Lindbladians, symmetries, and scars, with extensions toward PEPS. This kind of structural theorem can influence multiple subfields (many-body theory, quantum information, numerical methods) and enable new analytical/numerical tools. Paper 1 is valuable and timely for large-scale open-system quantum transport, but its impact is more specialized to energy-transport modeling and a particular variational-polaron framework.
Paper 2 provides a fundamental theoretical framework with broad applicability across multiple areas of quantum physics—Hamiltonian eigenstates, scar states, driven systems, Lindbladian steady states, and symmetry characterization—unified through a single local characterization theorem. Its generality (covering MPS, MPO, and PEPS) and connections to both analytical and numerical methods give it wider impact across condensed matter, quantum information, and mathematical physics. Paper 1, while experimentally impressive in demonstrating orthogonal temporal modes for microwave quantum networking, addresses a more specialized capability within superconducting circuit QED.
Paper 1 offers a foundational theoretical result for Matrix Product States and tensor networks, which are ubiquitous computational and analytical tools across quantum many-body physics, condensed matter, and quantum information. Its ability to characterize exact solutions for a wide range of settings (Hamiltonians, Lindbladians, symmetries) guarantees broad methodological applicability. Paper 2, while presenting an innovative application of quantum light to strong-field ionization, is more specialized and likely to impact a narrower subfield of quantum optics and attosecond physics.
Many-body localization is a major topic in condensed matter and quantum physics, and this review paper covers a broad, highly active research area with connections to ergodicity, thermalization, and quantum computing. Review papers in such hot fields tend to accumulate high citation counts and serve as essential references. While Paper 1 presents novel technical results on tensor network characterizations with rigorous methodology, its scope is more specialized. Paper 2's breadth of impact across multiple subfields, timeliness, and accessibility as an introductory review give it higher overall scientific impact potential.
Paper 2 likely has higher impact due to a broadly applicable, conceptually novel framework: a necessary-and-sufficient local condition to characterize when tensor-network states/operators are exact eigenvectors of extensive local operators. This unifies and advances multiple active areas (Hamiltonian eigenstates including scars, Lindbladian steady states, driven dynamics, symmetries) and suggests new analytical and algorithmic tools, with possible extensions to PEPS. Paper 1 is timely and rigorous for superconducting-qubit readout modeling, but its scope is more specialized to circuit QED and specific bath-engineering effects.
Paper 2 provides a fundamental theoretical result about tensor network states with broad implications across quantum many-body physics, including exact eigenstates, scar states, driven systems, Lindbladian steady states, and symmetry characterization. The local-to-global characterization is mathematically elegant, widely applicable, and extends to 2D (PEPS). Paper 1 identifies an interesting security concern in quantum circuit cutting but addresses a relatively niche problem in quantum cloud computing security with limited broader scientific impact beyond that specific domain.
Paper 2 addresses a critical bottleneck in near-term quantum computing by proposing a Floquet-engineering framework to protect quantum simulations of lattice gauge theories from constraint violations. Its passive error-correction approach is highly timely and practically applicable to current hardware, giving it stronger immediate real-world impact compared to the foundational, theoretical tensor-network results in Paper 1.
Paper 2 provides a comprehensive mathematical framework (local characterization of tensor network eigenstates) that unifies and addresses multiple important problems across quantum many-body physics: exact eigenstates, scar states, driven systems, Lindbladian steady states, and generalized symmetries. Its breadth of applicability across numerous subfields, potential to enable both analytical and numerical advances, and extension to 2D PEPS give it broader impact. Paper 1, while elegant in tightening Carnot's bound, addresses a more specific problem in quantum thermodynamics with comparatively narrower scope of influence.