Robust quantification of spectral transitions in perturbed quantum systems

Daniel Burgarth; Davide Lonigro; Kazuya Yuasa; Paolo Facchi; Stefan Gehr; Zsolt Szab\'o

Robust quantification of spectral transitions in perturbed quantum systems

T0 review · 0 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves a universal time-independent ceiling on transitions between gapped spectral bands of a perturbed quantum system.

desk verdict A genuine and rigorous coarse-grained eternal leakage bound; the central proof is sound and the paper should go to peer review. read the letter →

arxiv 2505.19904 v1 pith:QZRZHVPA submitted 2025-05-26 quant-ph math-phmath.MP

Zsolt Szabó , Stefan Gehr , Paolo Facchi , Kazuya Yuasa , Daniel Burgarth , Davide Lonigro This is my paper · ORCID

classification quant-phmath-phmath.MP MSC 81Q1547A5581Q10

keywords eternaladiabaticityspectralgapSchrieffer-WolfftransformationBlochequationseffectiveHamiltonianleakageboundsperturbationtheoryunboundedHamiltonians

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proves that when a quantum system with a gapped energy spectrum is weakly perturbed, the chance of making a transition between distinct coarse-grained spectral components stays small forever. For a Hamiltonian $H = \gamma H_0 + V$ whose unperturbed part has a spectral gap $\eta$ between components, it shows that if $\gamma > 4\pi\lVert V\rVert/\eta$, then the leakage out of any bounded spectral component obeys $L_k(t) \le (1 - 4\pi\lVert V\rVert/(\gamma\eta))^{-1/2} - 1 \le 9\pi\lVert V\rVert/(\gamma\eta)$ for all times $t$. The estimate depends on $H_0$ only through the gap, so it covers continuous spectra, infinitely many bands, and unbounded unperturbed Hamiltonians. The same machinery gives uniform-in-time bounds on how well the Schrieffer–Wolff and Bloch effective dynamics approximate the true evolution.

What carries the argument

The load-bearing construction is the perturbative solution of the Bloch equations. At each order the unknown operator solves a Sylvester equation $[H_0,X]=Y$, and Lemma 8 gives the solution $X=\int_{-\infty}^{\infty} e^{-itH_0}Y e^{itH_0} f(t)\,dt$ for any $f\in L^1(\mathbb{R})$ whose Fourier transform is $1/s$ on $|s|\ge \eta$; the best possible choice has $\inf_f \int |f| = \pi/(2\eta)$. The resulting recursive series for the Bloch wave operator $\Omega$ has term norms bounded by Catalan numbers, so the Catalan generating function yields the explicit closeness estimate $\lVert\Omega - 1\rVert \le \delta(\lVert V\rVert/(\gamma\eta)) < 1$, which is what turns into the eternal leakage bounds.

What would settle it

A direct numerical test would simulate a two-band model, say $H_0$ with eigenvalues $0$ and $\eta$, plus an off-diagonal bounded perturbation of norm $\lVert V\rVert$, with $\gamma$ just above $4\pi\lVert V\rVert/\eta$, and compute $L_1(t)$ at many long times: exceeding $9\pi\lVert V\rVert/(\gamma\eta)$ at any $t$ would refute Theorem 7. A sharper test targets the premise directly by trying to construct a pair $H_0,V$ for which the minimal $L^1$ norm of such an $f$ is strictly larger than $\pi/(2\eta)$, which would break Proposition 3.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is Theorem 7: for a self-adjoint $H_0$ (possibly unbounded), a coarse-grained spectral decomposition with gap $\eta = \inf_{k\neq l}\operatorname{dist}(\sigma_k,\sigma_l) > 0$, and a bounded self-adjoint perturbation $V$, whenever $\gamma > 4\pi\lVert V\rVert/\eta$ the leakage $L_k(t) = \lVert Q_k e^{-itH}P_k\rVert$ out of any bounded spectral component is bounded uniformly in time by $(1 - 4\pi\lVert V\rVert/(\gamma\eta))^{-1/2} - 1 \le 9\pi\lVert V\rVert/(\gamma\eta)$. The proof works by constructing a block-diagonal effective generator $H_{\mathrm{Bloch}}$ that is similar to $H$, showing the distance between the two evolutions is small for all $t$, and then passing to unbounded $H_0$ by truncating the spectrum and taking the limit. A Hermitian effective generator $H_{\mathrm{SW}}$ is also built through a Schrieffer–Wolff rotation, with its own uniform error bound.

Load-bearing premise

The whole chain of bounds depends on the existence of an integrable function $f$ whose Fourier transform is exactly $1/s$ for every frequency $|s|\ge\eta$, with a finite and optimally small $L^1$ norm; if that single analytic fact failed, the recursive Bloch solution and all subsequent estimates would not get off the ground.

Editorial extensions

If this is right

For any bounded spectral component, the leakage is forever bounded by $\sim 9\pi\lVert V\rVert/(\gamma\eta)$, so in the weak-perturbation limit transitions out of a band are suppressed linearly in the perturbation strength at all times.
The bound applies to unbounded and continuous spectra as long as the coarse-grained gap is positive and the perturbation is bounded, covering crystal band models, harmonic chains, and transmon-like cosine potentials.
The effective evolutions generated by the Bloch and Schrieffer–Wolff Hamiltonians stay uniformly close to the true evolution, putting quantitative error control under standard effective-Hamiltonian methods.
Because the estimate depends on $H_0$ only through $\eta$, it is independent of the number of bands, the dimension of each band, and the system size, including infinite chains.
The leakage decays at least as $O(1/\gamma)$ as the unperturbed part strengthens, matching the leading-order behaviour seen in the paper's numerical examples.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural next step is to allow unbounded perturbations, for which the authors expect state-dependent bounds with a slower-than-$1/\gamma$ decay or even persistent leakage for arbitrarily weak perturbations.
The same block-diagonal construction could be adapted to open quantum systems, where a Lindblad generator that is block-diagonal in the unperturbed decomposition might yield an eternal bound on population leakage under dissipation.
Because the threshold $\gamma > 4\pi\lVert V\rVert/\eta$ comes from the optimal $L^1$ constant, simple few-level models with $\lVert V\rVert/\eta$ close to $1/(4\pi)$ could probe how tight the bound is; the numerics in the paper only test the weak-coupling scaling.
One could test the bound's independence of system size directly in the tight-binding chain: the paper's numerics show size independence for 50–500 cells, and the theorem predicts the same bound for the infinite chain, which is outside numerical reach but within the theorem's scope.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

The short version: this is a rigorous, parameter-free extension of eternal adiabaticity to coarse-grained spectral decompositions, including unbounded unperturbed Hamiltonians. I read the proof carefully and the central bound holds up; the stress-test note is right that there is no load-bearing gap. The paper deserves serious refereeing.

What is actually new: prior eternal bounds [12,13] were fine-grained (discrete eigenspaces). Here the authors adapt the Bloch and Schrieffer–Wolff constructions to arbitrary coarse-grained decompositions with a finite spectral gap, and prove a time-independent leakage bound depending only on ||V|| and η. The extension to unbounded H0 via truncation (Theorem 7) is handled honestly, with strong dynamical convergence proved in Appendix G. The Catalan-series argument in Proposition 3 is clean, and the bound δ(x) is explicit. The paper also shows the Schrieffer–Wolff version and bounds its distance to the true evolution. That is a genuine advance over the cited fine-grained results.

Soft spots, in proportion: the numerical examples are illustrative and no code is provided, so they are not independently reproducible. That is acknowledged and not central. The restriction to bounded perturbations is stated up front and left as future work, so not a flaw. There is a small imprecision in Section V: they call '2' the maximal meaningful value of the leakage, but L_k(t) as an operator norm cannot exceed 1. It doesn't affect the result, since the linear bound is trivially valid for large scaling. The proof relies on the existence of an L1 function f with \hat f(s)=1/s for |s|≥η; this is standard from Bhatia and the paper invokes it explicitly, so I don't regard it as a soft spot.

Who should read this: people working on adiabatic quantum computing, effective Hamiltonians, and rigorous perturbation theory. It is a solid theoretical tool, not a paradigm shift.

Recommendation: send it to a serious referee. The proofs are detailed enough for careful checking, and the result is likely to be cited as the standard coarse-grained eternal bound.

Referee Report

0 major / 5 minor

Summary. The paper studies the leakage between coarse-grained spectral components of a Hamiltonian H = γH0 + V, where H0 has a spectral gap η and V is a bounded perturbation. It constructs two effective generators: a Bloch-type generator H_Bloch and a Hermitian Schrieffer–Wolff generator H_SW, and proves time-independent bounds on the distance between the true evolution and the effective evolutions. The main result, Theorem 7, states that for γ > 4π||V||/η the leakage out of any bounded spectral component obeys L_k(t) ≤ (1 − 4π||V||/(γη))^{-1/2} − 1 ≤ 9π||V||/(γη) for all t ∈ R, including for unbounded self-adjoint H0 via a truncation argument. The paper also applies these bounds to a tight-binding chain, a harmonic chain with infinite bands, and a transmon architecture.

Significance. If correct, Theorem 7 is a substantial and useful result: it gives an eternal, time-independent bound on band-to-band transitions that depends only on the perturbation norm and the coarse-grained spectral gap, with no dependence on the number, size, or dimensionality of the spectral components. The derivation is parameter-free and the constants are explicit; the extension to unbounded spectra via strong dynamical convergence is a genuine step beyond earlier discrete-spectrum results in [12,13]. The proofs are detailed and the technical premise from Bhatia's matrix analysis is standard. The numerical examples are illustrative rather than exhaustive, and no code is supplied, but the analytical claims are the paper's core contribution.

minor comments (5)

[Section V, proof of Theorem 7] In the truncation argument, the claim that "by suitably choosing E_n, P_k can be made to coincide with P_{H0^(n)}(σ_k)" needs clarification: this works when E_n is chosen in a different spectral component than σ_k, not for an arbitrary E_n. Please state this explicitly and justify that such a choice is always possible (for m=1 the leakage is trivial, and otherwise one can pick a point in another component and take n large enough).
[Section V, after Eq. (70)] The sentence stating that the first bound on the leakage "exceeds the maximal meaningful value of 2" is imprecise: for the leakage L_k(t) the trivial upper bound is 1, not 2. The final linear bound 9π||V||/(γη) is still valid, but the argument should be rephrased: for ||V||/(γη) ≥ 1/(9π) the linear bound is trivial, while for smaller values the inequality ε(x) ≤ 9πx follows from a short elementary estimate.
[Appendix A, Lemma 8] Lemma 8 states the Fourier condition for |s| > η, whereas Proposition 2 and Eq. (28) use |s| ≥ η. The two are effectively equivalent for admissible L1 functions because the Fourier transform is continuous, but the statements should be aligned to avoid confusion.
[Theorem 4 proof] The phrase "Following standard arguments (cf. Appendix B)" appears to cite the wrong appendix for the inequalities (38)–(40); these are proved in Appendix D. Please correct the cross-reference.
[Section VI] The numerical simulations in the examples are described without code or full parameter specifications (e.g., the realizations {r_i} in the tight-binding chain). Since the paper's main results are analytic, this is not blocking, but adding a short reproducibility note would improve the presentation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the leakage bound is derived from stated assumptions and standard external lemmas, with no fitted parameter or self-citation carrying the proof.

full rationale

Walking the derivation chain, the central bound in Theorem 4 follows from Proposition 3's convergent perturbative solution of the Bloch equations, and the estimate is obtained by Catalan-number bounds plus the explicit L1 Fourier lemma from Bhatia [15], which is also re-proven in Appendix A for the infinite-dimensional setting. No parameter is fitted to the leakage or to the dynamics being bounded; the only imported technical premise, existence of f in L1(R) with hat f(s)=1/s for |s|>=η and inf ∫|f|=π/(2η), is a standard external result, does not contain the target bound, and is used as a lemma rather than as an equivalent reformulation of the conclusion. The self-citations [12,13] are used for context and for a Schrieffer-Wolff construction that is independently proven in Appendix E; they do not carry the core calculation. The unbounded extension in Theorem 7 is an honest truncation argument: the spectral gap condition transfers via η<=η_n and strong dynamical convergence is supplied via Proposition 12 and Appendix G. The stated limitation that unbounded perturbations are not covered is explicit and does not mask any circular step. There are no fitted inputs renamed as predictions, no ansatz smuggled in through self-citation, and no uniqueness claim imported from the authors' prior work. The derivation is self-contained relative to standard operator theory and the admitted assumptions, so the circularity score is 0.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No free parameters are fitted; the bound's input data are the operator norm ∥V∥ and the spectral gap η. The paper assumes a coarse-grained decomposition with positive gap, a bounded perturbation, and a large-γ condition. It relies on standard operator-theory tools: a representation of Sylvester equations using an L1 test function, the spectral theorem, and strong dynamical convergence of truncated Hamiltonians. No new physical entities, forces, particles, or dimensions are introduced.

assumptions (5)

domain assumption Coarse-grained spectral decomposition of σ(H0) into components with positive gap η
Eqs. (2)-(3) and Theorem 7; all leakage bounds require η=inf dist(σ_k,σ_l)>0.
domain assumption Bounded perturbation V, with H0 possibly unbounded
The bounds use ∥V∥ and the paper explicitly leaves unbounded V to future work in Section VII.
domain assumption Large-γ condition γ>4π∥V∥/η or the stricter Schrieffer-Wolff condition
Required for convergence of the Bloch series and δ<1, Proposition 3 and Theorems 4, 6, 7.
standard math Existence of f∈L1(R) with Fourier transform 1/s on |s|≥η
Used in Eq. (28) and Lemma 8 to solve Sylvester equations; existence and infimum ∫|f|=π/(2η) are cited to Bhatia [15, Thm VII.2.5].
standard math Strong dynamical convergence of truncated Hamiltonians
Proposition 12 from de Oliveira [26] supplies Eq. (64), used in Theorem 7 to pass to unbounded H0.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Robust quantification of spectral transitions in perturbed quantum systems." pith.science (2026). https://pith.science/paper/QZRZHVPA

@misc{pith2026250519904,
  author       = {Pith},
  title        = {Pith review of: Robust quantification of spectral transitions in perturbed quantum systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QZRZHVPA}},
  note         = {Machine review of arXiv:2505.19904}
}

read the original abstract

A quantum system subject to an external perturbation can experience leakage between uncoupled regions of its energy spectrum separated by a gap. To quantify this phenomenon, we present two complementary results. First, we establish time-independent bounds on the distances between the true dynamics and the dynamics generated by block-diagonal effective evolutions constructed via the Schrieffer-Wolff and Bloch methods. Second, we prove that, under the right conditions, this leakage remains small eternally. That is, we derive a time-independent bound on the leakage itself, expressed in terms of the spectral gap of the unperturbed Hamiltonian and the norm of the perturbation, ensuring its validity for arbitrarily large times. Our approach only requires a finite spectral gap, thus accommodating continuous and unbounded spectra. Finally, we apply our bounds to specific systems of practical interest.

Figures

Figures reproduced from arXiv: 2505.19904 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 18 canonical work pages

[1]

Farhi, J

E. Farhi, J. Goldstone, S. Gutmann, and M. Sipser, Quantum Computation by Adiabatic Evolution, arXiv:quant-ph/0001106 (2000)

arXiv 2000
[2]

Albash and D

T. Albash and D. A. Lidar, Adiabatic quantum compu- tation, Rev. Mod. Phys.90, 015002 (2018)

2018
[3]

Emerging Tal- ent Initiative

The similarity transformation Ω(γ) should be close to the identity,∥Ω(γ)−11∥=O(γ −1), for largeγ. If such an operatorH eff (γ) is found, the leakage out of the kth component of the Hilbert space can be equivalently quantified as Lk(t;γ) =∥Q k(e−itH(γ) −e −itHeff (γ))Pk∥,(8) asH eff (γ), being block-diagonal, satisfies Qke−itHeff (γ)Pk = 0. Since the opera...

work page 2024
[4]

A. Y. Kitaev, Fault-tolerant quantum computation by anyons, Ann. Phys. (N.Y.)303, 2 (2003)

work page 2003
[5]

Chesi, D

S. Chesi, D. Loss, S. Bravyi, and B. M. Terhal, Thermo- dynamic stability criteria for a quantum memory based on stabilizer and subsystem codes, New J. Phys.12, 025013 (2010)

work page 2010
[6]

J. R. Schrieffer and P. A. Wolff, Relation between the Anderson and Kondo Hamiltonians, Phys. Rev.149, 491 (1966), publisher: American Physical Society

work page 1966
[7]

Bravyi, D

S. Bravyi, D. P. DiVincenzo, and D. Loss, Schrieffer- Wolff transformation for quantum many-body systems, Ann. Phys. (N.Y.)326, 2793 (2011)

work page 2011
[8]

Bloch, Sur la th´ eorie des perturbations des ´ etats li´ es, Nucl

C. Bloch, Sur la th´ eorie des perturbations des ´ etats li´ es, Nucl. Phys.6, 329 (1958)

1958

Show all 27 references

[9]

Lindgren, The Rayleigh-Schr¨ odinger perturbation and the linked-diagram theorem for a multi-configurational model space, J

I. Lindgren, The Rayleigh-Schr¨ odinger perturbation and the linked-diagram theorem for a multi-configurational model space, J. Phys. B: Atom. Mol. Phys.7, 2441 (1974)

1974
[10]

Durand, Direct determination of effective Hamilto- nians by wave-operator methods

P. Durand, Direct determination of effective Hamilto- nians by wave-operator methods. I. General formalism, Phys. Rev. A28, 3184 (1983)

1983
[11]

Jolicard, Effective Hamiltonian theory: an intermedi- ate representation method for the wave operator calcula- tion, Chem

G. Jolicard, Effective Hamiltonian theory: an intermedi- ate representation method for the wave operator calcula- tion, Chem. Phys.115, 57 (1987)

1987
[12]

J. P. Killingbeck and G. Jolicard, The Bloch wave op- erator: generalizations and applications: Part I. The time-independent case, J. Phys. A: Math. Gen.36, R105 (2003)

2003
[13]

Burgarth, P

D. Burgarth, P. Facchi, H. Nakazato, S. Pascazio, and K. Yuasa, Eternal adiabaticity in quantum evolution, Phys. Rev. A103, 032214 (2021)

2021
[14]

Facchi, M

P. Facchi, M. Ligab` o, and V. Viesti, Robustness of quan- tum symmetries against perturbations, J. Phys. A: Math. Theor.58, 125305 (2025)

2025
[15]

C. E. Soliverez, General theory of effective Hamiltonians, Phys. Rev. A24, 4 (1981)

1981
[16]

Bhatia,Matrix Analysis(Springer, New York, 1997)

R. Bhatia,Matrix Analysis(Springer, New York, 1997)

1997
[17]

Bloch, ¨Uber die Quantenmechanik der Elektronen in Kristallgittern, Z

F. Bloch, ¨Uber die Quantenmechanik der Elektronen in Kristallgittern, Z. Phys.52, 555 (1929)

1929
[18]

Endres, H

M. Endres, H. Bernien, A. Keesling, H. Levine, E. R. Anschuetz, A. Krajenbrink, C. Senko, V. Vuletic, M. Greiner, and M. D. Lukin, Atom-by-atom assembly of defect-free one-dimensional cold atom arrays, Science 354, 1024 (2016)

2016
[19]

J. Kang, R. Wei, Q. Zhang, and G. Dong, Topological Photonic States in Waveguide Arrays, Adv. Phys. Res. 2, 2200053 (2023)

2023
[20]

K. K. Likharev and A. B. Zorin, Theory of the Bloch- wave oscillations in small Josephson junctions, J. Low Temp. Phys.59, 347 (1985)

1985
[21]

J. Koch, T. M. Yu, J. Gambetta, A. A. Houck, D. I. Schuster, J. Majer, A. Blais, M. H. Devoret, S. M. Girvin, and R. J. Schoelkopf, Charge-insensitive qubit design de- rived from the Cooper pair box, Phys. Rev. A76, 042319 (2007)

2007
[22]

S. A. Wilkinson, N. Vogt, D. S. Golubev, and J. H. Cole, Approximate solutions to Mathieu’s equation, Physica E 100, 24 (2018)

2018
[23]

H. Paik, D. I. Schuster, L. S. Bishop, G. Kirchmair, G. Catelani, A. P. Sears, B. R. Johnson, M. J. Reagor, L. Frunzio, L. I. Glazman, S. M. Girvin, M. H. De- voret, and R. J. Schoelkopf, Observation of High Coher- ence in Josephson Junction Qubits Measured in a Three- Dimensi...

2011
[24]

Barends, J

R. Barends, J. Kelly, A. Megrant, D. Sank, E. Jeffrey, Y. Chen, Y. Yin, B. Chiaro, J. Mutus, C. Neill, P. O’Malley, P. Roushan, J. Wenner, T. C. White, A. N. Cleland, and J. M. Martinis, Coherent Josephson Qubit Suitable for Scalable Quantum Integrated Circuits, Phys. Rev. Let...

2013
[25]

A. A. Golubov, M. Y. Kupriyanov, and E. Il’ichev, The current-phase relation in Josephson junctions, Rev. Mod. Phys.76, 411 (2004)

2004
[26]

P. J. Miana and N. Romero, Catalan generating functions for bounded operators, Ann. Funct. Anal.14, 69 (2023)

2023
[27]

C. R. de Oliveira,Intermediate Spectral Theory and Quantum Dynamics(Birkh¨ auser, Basel, 2009)

2009

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Farhi, J

E. Farhi, J. Goldstone, S. Gutmann, and M. Sipser, Quantum Computation by Adiabatic Evolution, arXiv:quant-ph/0001106 (2000)

arXiv 2000

[2] [2]

Albash and D

T. Albash and D. A. Lidar, Adiabatic quantum compu- tation, Rev. Mod. Phys.90, 015002 (2018)

2018

[3] [3]

Emerging Tal- ent Initiative

The similarity transformation Ω(γ) should be close to the identity,∥Ω(γ)−11∥=O(γ −1), for largeγ. If such an operatorH eff (γ) is found, the leakage out of the kth component of the Hilbert space can be equivalently quantified as Lk(t;γ) =∥Q k(e−itH(γ) −e −itHeff (γ))Pk∥,(8) asH eff (γ), being block-diagonal, satisfies Qke−itHeff (γ)Pk = 0. Since the opera...

work page 2024

[4] [4]

A. Y. Kitaev, Fault-tolerant quantum computation by anyons, Ann. Phys. (N.Y.)303, 2 (2003)

work page 2003

[5] [5]

Chesi, D

S. Chesi, D. Loss, S. Bravyi, and B. M. Terhal, Thermo- dynamic stability criteria for a quantum memory based on stabilizer and subsystem codes, New J. Phys.12, 025013 (2010)

work page 2010

[6] [6]

J. R. Schrieffer and P. A. Wolff, Relation between the Anderson and Kondo Hamiltonians, Phys. Rev.149, 491 (1966), publisher: American Physical Society

work page 1966

[7] [7]

Bravyi, D

S. Bravyi, D. P. DiVincenzo, and D. Loss, Schrieffer- Wolff transformation for quantum many-body systems, Ann. Phys. (N.Y.)326, 2793 (2011)

work page 2011

[8] [8]

Bloch, Sur la th´ eorie des perturbations des ´ etats li´ es, Nucl

C. Bloch, Sur la th´ eorie des perturbations des ´ etats li´ es, Nucl. Phys.6, 329 (1958)

1958