When level repulsion fails: non-normality and chaos in open quantum systems
Caio B. Naves, Thomas Klein Kvorning, Jonas Larson
Abstract
For Hamiltonian systems, level statistics provide a faithful diagnostic of quantum chaos. By analogy, the statistics of the Lindbladian spectrum are often used in open quantum systems, and the Grobe-Haake-Sommers conjecture proposes that systems with chaotic classical counterparts should exhibit level repulsion in the Lindbladian spectrum. Here we point out an important flaw in this analogy: Hamiltonian and Lindbladian spectra behave differently and have distinct physical interpretations, and one should therefore not expect the latter to provide a reliable diagnostic. For Lindbladians, the late-time dynamics are not determined by the bulk of the eigenvalues but only by those eigenvalues -- and their corresponding eigenvectors -- with small real parts. Combined with the strong non-normality typical of Lindbladians, this allows situations in which the level statistics can be tuned almost arbitrarily without affecting the dynamics on either short or long time scales. We explicitly demonstrate this phenomenon and provide examples in which Ginibre level repulsion arises while the system dynamics at no time show signatures of chaos. We further relate this mechanism to the emergence of a non-Hermitian skin effect in Liouville space, linking boundary-induced eigenvector localization to the observed spectral instability. Our results show that level statistics cannot universally serve as a reliable diagnostic of quantum chaos in open quantum systems and highlight the need for alternative diagnostics that remain robust in strongly non-normal regimes.
AI Impact Assessments
(3 models)Scientific Impact Assessment
Core Contribution
This paper challenges a widely adopted diagnostic practice in open quantum systems: using Lindbladian spectral level statistics (level repulsion) as an indicator of quantum chaos, following the Grobe-Haake-Sommers (GHS) conjecture. The central claim is that non-normality of Lindbladian superoperators renders their spectral statistics unreliable as chaos diagnostics. The authors demonstrate this through two integrable models—a driven quantum harmonic oscillator coupled to a thermal bath and an open tight-binding chain—where numerical diagonalization produces Ginibre-like level repulsion despite completely regular dynamics at all time scales.
The key mechanism identified is that Lindbladians are generically highly non-normal operators, meaning their eigenvalues are exponentially sensitive to perturbations (including unavoidable numerical round-off errors). The condition number κ grows exponentially with system size, so even machine-precision perturbations can completely reorganize the spectrum. The authors connect this spectral instability to the non-Hermitian skin effect in Liouville space, where eigenvectors localize near boundaries, providing a physical picture for the mathematical phenomenon.
Methodological Rigor
The approach is sound and well-constructed. The authors present:
1. Analytical spectra for both models (via third quantization for the oscillator, Fourier analysis for the tight-binding chain), establishing ground truth for the integrable cases.
2. Numerical spectral statistics showing clear Ginibre-level repulsion, with complex spacing ratios matching random matrix theory predictions (⟨r⟩ ≈ 0.74, −⟨cos θ⟩ ≈ 0.24-0.30, close to GinUE values).
3. Uhlmann fidelity analysis demonstrating that the actual dynamics remain arbitrarily close to integrable dynamics for arbitrarily long times as system size increases.
4. Eigenvalue condition number analysis showing exponential growth with system size for bulk eigenvalues versus constant behavior for steady-state eigenvalues.
The argument is logically tight: the non-commuting limits (thermodynamic limit before vs. after diagonalization) are clearly identified. The connection to pseudospectra provides formal mathematical backing. However, some limitations exist: the two examples are relatively simple (quadratic/free models), and the paper does not provide a systematic characterization of which classes of Lindbladians are "sufficiently non-normal" for this breakdown to occur. The authors acknowledge that not all Lindbladians are highly non-normal but don't precisely delineate the boundary.
The analysis of the converse direction—whether absence of level repulsion reliably indicates integrability—is also valuable. The authors argue that for highly non-normal systems, Poisson statistics are equally fragile, making level statistics a fundamentally unreliable diagnostic in both directions.
Potential Impact
This paper has significant implications across multiple subfields:
Open quantum systems and quantum chaos: The finding that a standard diagnostic tool is unreliable invalidates or calls into question numerous prior studies that relied solely on Lindbladian level statistics to diagnose chaos (the paper cites roughly a dozen such works). This forces the community to develop alternative diagnostics robust to non-normality.
Numerical methods: The observation that spectral calculations of non-normal Lindbladians are inherently unreliable has practical consequences for Krylov/Arnoldi methods, matrix-product-operator approaches, and any spectral-based simulation technique. Apparent spectral gaps and decay rates may be artifacts.
Quantum information: Noise models based on Lindblad generators may inherit strong non-normality, potentially affecting error rate estimates and threshold calculations for quantum error correction.
Non-Hermitian physics: The connection to the skin effect in Liouville space enriches the understanding of non-Hermitian phenomena and adds to growing evidence that spectral properties alone are insufficient for characterizing non-Hermitian systems.
Timeliness & Relevance
This work arrives at an opportune moment. The use of Lindbladian level statistics has proliferated rapidly in recent years, becoming almost routine in studies of open quantum chaos. Simultaneously, the non-Hermitian skin effect has attracted enormous attention. This paper bridges these two active areas and provides a timely correction to a growing body of literature that may be drawing incorrect conclusions. The concurrent preprint by Villaseñor et al. (2024) questioning the GHS conjecture from a different angle (classical strange attractors) reinforces the timeliness.
Strengths
Limitations
Overall Assessment
This is a conceptually important paper that identifies a fundamental flaw in a widely used diagnostic tool. While the technical execution relies on simple models, the underlying mathematical argument (exponential growth of condition numbers, pseudospectral instability) is general and compelling. The paper will likely prompt significant reassessment of prior results and stimulate development of more robust chaos diagnostics for open quantum systems.
Generated Apr 2, 2026
Comparison History (41)
Paper 1 challenges a widely-used diagnostic (level statistics) for quantum chaos in open systems, revealing a fundamental flaw in the analogy between Hamiltonian and Lindbladian spectra. This has broad theoretical implications across quantum chaos, open quantum systems, and non-Hermitian physics, potentially redirecting research efforts. Paper 2 presents valuable engineering work on fluxonium processor design, but is more incremental and application-specific. Paper 1's conceptual insight about non-normality undermining spectral diagnostics is likely to influence multiple subfields and prompt development of new theoretical tools.
Paper 2 has higher likely impact: it identifies a fundamental limitation of a widely used chaos diagnostic (Lindbladian level statistics), explains the mechanism (non-normality, spectral instability, eigenvector effects), and connects it to timely themes in non-Hermitian physics (skin effect) with broad relevance to open quantum systems, quantum chaos, and dissipative platforms. This can redirect methodology across multiple subfields and affects how experiments/theory interpret “chaos” under dissipation. Paper 1 is rigorous and novel, but its scope is more specialized to thermalization/operator growth and typical low-complexity ensembles.
Paper 2 challenges a widely accepted paradigm (the Grobe-Haake-Sommers conjecture) and demonstrates a fundamental flaw in using level statistics to diagnose chaos in open quantum systems. By invalidating a commonly used metric and linking the phenomenon to the highly topical non-Hermitian skin effect, it forces a profound methodological shift in the field of quantum chaos. While Paper 1 offers a highly versatile and practical algorithmic framework for quantum simulation, Paper 2's potential to overturn existing theoretical assumptions gives it a broader and more disruptive fundamental scientific impact.
Paper 2 has higher potential impact due to its direct, timely relevance to quantum optimization: it proposes a concrete, ostensibly tuning-free algorithm, provides analytic support (Floquet-Magnus effective Hamiltonian and gap claims), and backs it with both simulations and real-hardware experiments. If robust, it could influence quantum algorithm design and near-term applications across combinatorial optimization and quantum control. Paper 1 is conceptually important for open-system quantum chaos diagnostics and non-normality/skin-effect links, but its impact is more specialized and largely corrective (limiting an existing diagnostic) rather than offering a broadly deployable method or technology.
Paper 2 demonstrates scalable, integrated quantum memory on a silicon chip, addressing a critical bottleneck in photonic quantum computing. Its use of standard semiconductor foundry processing guarantees high real-world applicability and scalability. While Paper 1 provides important theoretical corrections to quantum chaos diagnostics in open systems, Paper 2's experimental achievement paves the way for immediate technological advancements in quantum networking and computing hardware. This gives Paper 2 a significantly broader and more direct technological impact.
Paper 2 likely has higher scientific impact: it presents a timely, application-driven advance with major real-world implications (cryptography and fault-tolerant quantum computing), substantially lowering qubit resource estimates for Shor’s algorithm and aligning with rapid experimental progress in neutral-atom platforms. Its breadth spans quantum error correction, architecture, and security, and could influence both research roadmaps and policy/industry. Paper 1 is conceptually important for open-quantum-systems diagnostics and non-normality, but its impact is more specialized and primarily reframes interpretation rather than enabling a comparably broad technological capability.
Paper 2 makes a breakthrough contribution to quantum error correction by combining nearly optimal LDPC code parameters with fault-tolerant non-Clifford gates—a major open problem. The algebraic-topological framework ('cupcap gates') is novel and general, with immediate practical implications for scalable fault-tolerant quantum computing. Paper 1 provides valuable conceptual clarification about level statistics in open quantum systems, but is more diagnostic/cautionary in nature. Paper 2's impact spans quantum computing theory, topology, and practical fault-tolerant architectures, addressing a central bottleneck in the field.
Paper 1 challenges a widely-used diagnostic (level statistics) for quantum chaos in open systems, demonstrating fundamental flaws in the Grobe-Haake-Sommers conjecture analogy. It connects non-normality, spectral instability, and the non-Hermitian skin effect in Liouville space—bridging multiple active research areas. Its implications affect how the entire community diagnoses chaos in open quantum systems, necessitating new tools. Paper 2 provides a useful closed-form formula for critical timescales in non-Hermitian dynamics but addresses a more specialized problem with narrower community impact, despite its mathematical elegance.
Paper 1 offers a more broadly enabling and novel methodology: a bootstrap that reconstructs hidden symmetry group representation theory from dynamical spectral data (xSFF), with demonstrated recovery of full character tables, branching rules, and projective features across multiple many-body models. This could become a practical tool for identifying emergent symmetries in numerics/experiments, impacting condensed matter, quantum information, and statistical/chaos physics. Paper 2 provides an important cautionary result about Lindbladian level statistics and non-normality, but its primary contribution is largely diagnostic/negative and likely narrower in downstream methodological adoption.
Paper 2 likely has higher impact: it challenges a widely used diagnostic (Lindbladian level statistics) by identifying a fundamental flaw tied to non-normality, demonstrates tunability of level statistics without dynamical consequences, and connects to timely themes like non-Hermitian physics and the skin effect. This has broad implications across open quantum systems, quantum chaos, and non-Hermitian dynamics, potentially reshaping methodology. Paper 1 is rigorous and valuable for quantum error correction, but its impact is more specialized and incremental within coding theory/QEC compared to Paper 2’s cross-field conceptual correction.