Pith. sign in

REVIEW 4 major objections 5 minor 47 references

Orthogonal Quantum Krylov Diagonalisation

T0 review · 4 major / 5 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read The paper claims that the classical Lanczos recursion can be reconstructed on a quantum processor from diagonal expectation values alone, yielding an orthonormal Krylov basis without overlap-matrix regularization.

desk verdict The recursion is right and the idea is genuinely new, but the numerical-stability story skips the one thing that can break it: finite-shot noise in the measured moments. read the letter →

arxiv 2607.09476 v2 pith:DWEPLW3G submitted 2026-07-10 quant-ph

Hadi Rammal , Alexandre Perrin , Oumaya Ladhari , Clément Dutreix , Jérémie Messud , Matthieu Saubanere This is my paper · ORCID
classification quant-ph MSC 81P6865F15 PACS 03.67.Ac
keywords QuantumKrylovsubspaceLanczosalgorithmblockencodingsignalprocessingoverlapmatrixeigenvalueestimationstatepreparationHeisenbergmodel
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 proposes a quantum algorithm that reproduces the classical Lanczos procedure without ever forming the non-orthogonal Krylov basis that plagues earlier quantum Krylov methods. The key move is to rewrite each Lanczos vector as a polynomial of the rescaled Hamiltonian acting on a fixed initial state, and to close the Lanczos recursion from two diagonal expectation values per step. The authors argue this yields an orthonormal basis by construction, a tridiagonal projected Hamiltonian, and the same convergence as classical Lanczos, so no overlap-matrix regularization is needed. They support this with exact-arithmetic simulations on a 4x4 J1-J2 Heisenberg model, an asymptotic measurement-complexity analysis, and a restarted protocol for state preparation.

What carries the argument

The recursive coefficient identity β̃²_{i+1} = ⟨ψ_i|H̃²|ψ_i⟩ − ᾱ_i² − β̃_i² (Eq. 14), derived from the tridiagonal structure of the projected Hamiltonian, is the mechanism that lets the algorithm compute the next Lanczos coefficient from diagonal measurements alone. It feeds the polynomial recurrence P_{n+1}(x) = ((x − ᾱ_n)P_n(x) − β̃_n P_{n−1}(x))/β̃_{n+1}, whose Chebyshev coefficients are then implemented as a GQSP circuit acting on a block-encoded walk operator.

What would settle it

Run the OQKD recursion on a small, exactly solvable Hamiltonian using moments corrupted by controlled noise (e.g., rounding to k bits or adding Gaussian shot noise), and track the condition number of the implicitly generated overlap matrix or the ground-state energy error. If small moment errors cause β̃_i to drift and orthogonality to degrade at modest Krylov dimension, the practical claim collapses; the paper reports no such noise experiment.

Watch

Extended reading notes

Core claim

The central claim is that the classical Lanczos three-term recurrence can be lifted from vectors to operators: with H̃ = H/α, the nth Lanczos vector is |ψ⊥_n⟩ = P_n(H̃)|Φ0⟩, and the normalization coefficient in the recurrence is fixed by β̃²_{n+1} = ⟨ψ⊥_n|H̃²|ψ⊥_n⟩ − ᾱ²_n − β̃²_n (Eq. 14). This identity, a consequence of the tridiagonal form of H in the Lanczos basis, means the entire basis can be generated from diagonal moments ⟨ψ|H̃|ψ⟩ and ⟨ψ|H̃²|ψ⟩, with no overlap matrix, no Hadamard tests, and no quantum memory. The authors show the Lanczos polynomials can be realized with block encoding and Generalized Quantum Signal Processing, with the same query complexity as Chebyshev-based Krylov

Load-bearing premise

The recursion stays numerically stable when the expectation values ⟨ψ_i|H̃|ψ_i⟩ and ⟨ψ_i|H̃²|ψ_i⟩ are estimated from finitely many noisy measurements; the paper's simulations use exact arithmetic and never test how shot noise degrades the recursively built orthonormal basis.

Editorial extensions

If this is right

  • No overlap-matrix regularization: the overlap matrix is the identity by construction, so the generalized eigenvalue problem reduces to a standard tridiagonal eigenproblem.
  • Convergence matches classical Lanczos: in exact-arithmetic simulations on J1-J2 Heisenberg, ground-state energy error follows the classical Lanczos curve to machine precision, while QKD/CQKD stall due to conditioning.
  • Measurement complexity decouples from initial-state overlap: per matrix element it is order 1/ε², independent of |γ0|, whereas prior Chebyshev QKD inherits exponential dependence on system size through |γ0|.
  • Same asymptotic query complexity as Chebyshev-based QKD.
  • Restarted protocol gives a state-preparation route for Quantum Phase Estimation with nearly constant success probability per restart.

Reading between the lines

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

  • If noise in the measured moments propagates through the recursion, the orthogonality that OQKD gains 'by construction' could be lost in practice; a finite-shot error-propagation analysis is the natural next test, and the paper does not provide one.
  • Because the Λ_n growth is tied to the Lanczos polynomial coefficients, the exponential overhead may be inherent to orthogonalization in any polynomial encoding, not just GQSP; exploring alternative encodings or approximate orthogonal polynomials could reveal whether the trade-off can be shifted.
  • The restarted protocol can be viewed as iterative polynomial filtering; combining it with error-mitigation techniques might extend it to excited states or noisy devices.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper introduces Orthogonal Quantum Krylov Diagonalization (OQKD), a quantum subspace method that constructs an orthonormal Lanczos basis by representing Lanczos vectors as polynomial transformations of a rescaled Hamiltonian. The key recursion, Eq. (14), determines the Lanczos normalization coefficient β̃_{i+1} from diagonal moments ⟨ψ_i|H̃²|ψ_i⟩, α̃_i and β̃_i, avoiding off-diagonal Hamiltonian measurements and overlap-matrix regularization. The authors derive this recursion in Appendix A, provide a GQSP/block-encoding implementation in Section III.B and Appendix C, and benchmark the method on a 4×4 J1–J2 Heisenberg model, reporting convergence identical to classical Lanczos and κ(S)≈1 in exact-arithmetic simulations. They further introduce a restarted protocol (ROQKD) intended to keep the GQSP success probability nearly constant while iteratively improving a trial state, and they argue that OQKD removes the initial-state-overlap dependence of the measurement complexity. The central theoretical construction is arithmetically sound, but the practical claim that OQKD eliminates the numerical instability of QKD/CQKD is established only in exact arithmetic; finite-shot measurement noise is never analyzed or simulated.

Significance. If the central claim holds, OQKD would be a conceptually important bridge between classical Lanczos diagonalization and quantum Krylov methods, providing an orthonormal Krylov basis without overlap-matrix inversion or Hadamard tests. The derivations in Appendices A and C are clean: the β-recursion follows directly from the Lanczos three-term identity, and the GQSP realization of the polynomial P_n(H̃) is correctly described with an explicit normalization condition. The exact-arithmetic numerical demonstration on the J1–J2 model is a useful proof of principle, and the idea of restarting with a GQSP-implemented approximate ground state is a promising direction for state preparation. However, the load-bearing advantage of OQKD over existing methods is numerical stability in a realistic quantum setting, and the manuscript does not provide any finite-shot-noise analysis, error propagation for Eq. (14), or simulation with measurement noise. The exponential decrease of the GQSP success probability is acknowledged but is not reconciled with the claimed query-complexity equivalence to Chebyshev-based methods. The contribution is therefore significant in exact arithmetic but presently condit

major comments (4)
  1. [§III.A, Eq. (14); Algorithm 1 steps 3(b)–3(c)] The central practical claim — that OQKD preserves orthogonality and eliminates overlap-matrix regularization — is not established under finite-shot measurement noise. In a real run, α̃_i and ⟨ψ_i|H̃²|ψ_i⟩ are estimated from a finite number of shots, so β̃_{i+1} computed from Eq. (14) carries statistical error. Because P_{i+1} is built from all previous noisy coefficients, the actual prepared state |ψ_i⟩ differs from the exact Lanczos vector, orthogonality is lost, and the projected Hamiltonian is no longer tridiagonal. The RHS of Eq. (14) can also become negative, yielding an imaginary β and breaking the algorithm. Neither an error-propagation bound nor a noisy simulation is provided; Figs. 2 and 4 use exact arithmetic. This is load-bearing because the advantage over CQKD is precisely numerical stability.
  2. [§III.B and §III.D, Eqs. (22)–(25), (34)] The abstract claims 'the same asymptotic query complexity as Chebyshev-based QKD methods,' but this is only true for the circuit depth of a single polynomial application, not for the expected cost of successfully preparing a Lanczos vector. The GQSP success probability is p_succ = 1/Λ_n², which Fig. 3 shows decreases exponentially with the Lanczos dimension. With amplitude amplification, the expected number of walk-operator calls becomes O(Λ_n n), not O(n), whereas CQKD has p_succ ≈ constant. The effective cost in Eq. (34) contains 1/p_succ(D), so the asymptotic comparison in Eq. (33) vs. Eq. (34) is misleading unless the success-probability overhead is included in the stated query complexity. The manuscript should either qualify the claim or analyze the total query complexity including postselection/amplification.
  3. [§IV and Appendix D] The restarted protocol's main assertion is that the restart polynomial Q^(k) remains well-conditioned and that the accumulated success probability Λ^(R) is nearly constant. This is supported only by the 4×4 J1–J2 example in Fig. 4; Appendix D, which is described as a 'detailed analysis,' contains no bound on Λ^(R), no system-size scaling of the success probability, and no argument beyond the statement that Q^(k) is a refined ground state. Since the entire advantage of ROQKD over direct OQKD rests on this point, a concrete analysis or a set of larger-system simulations is needed. The promised detailed analysis in Appendix D is absent.
  4. [§III.D, Eqs. (28)–(34)] The measurement-complexity comparison assumes that, in OQKD, only diagonal moments need to be measured and that the projected Hamiltonian is exactly tridiagonal. This is true in exact arithmetic, but it is precisely the property that finite-shot noise destroys. The error bound in Eq. (28) for CQKD explicitly includes statistical sampling error and thresholding effects; no analogous bound is derived for OQKD that includes the statistical error of the moments entering Eq. (14) and its propagation into the Ritz values. Without this, the claimed Θ(1/ε²) per-matrix-element cost and the removal of the |γ0| dependence are not substantiated as a practical complexity statement.
minor comments (5)
  1. [General] There are several typos and inconsistent notations: 'OKQD' appears in §IV.A; 'straightworward' in §III.B; 'rationalization' should be 'orthogonalization' in §II; 'it is seems suited' in §IV. The tildes on α and β are sometimes dropped in the text surrounding Eq. (13) and Appendix A; please make the scaled/unscaled notation consistent.
  2. [§III.C, Fig. 3] The exponential decrease of the success probability is shown, but the horizontal axis is the Lanczos subspace dimension on a 4×4 model. Reporting the same quantity for a larger system or, better, the scaling of Λ_n with n for a simple Hamiltonian would make the limitation quantitative rather than illustrative.
  3. [Appendix C.1] The complementary-polynomial construction and recursive phase-factor extraction follow Ref. [41], but the text should note that the numerical optimization of the complementary polynomial has no provable runtime guarantee in general. This is relevant because the claimed O(n) classical overhead assumes successful convergence of this step.
  4. [§IV.B, Fig. 4] Panel (b) shows that the success probability is 'nearly constant,' but the curves appear to be for one small model and a small range of restart counts. Please specify the system parameters, the restart count range, and the numerical precision used; otherwise the claim of near-constancy is not robustly supported.
  5. [References] Ref. [26] is a preprint without a DOI; consider citing the published version if available. Also, the text mentions 'Q-SENSE' but the reference list does not include the original Q-SENSE paper; please add a proper citation.

Circularity Check

1 steps flagged · score 2.0 of 10

No substantive circularity; the beta-recursion is derived, not assumed. The only mild self-definitional aspect is that OQKD is built from the same Lanczos recurrence whose convergence it reports.

  1. self definitional [Abstract; Sec. III.A Eq. (10)]
    "By expressing Lanczos vectors as polynomial transformations of the Hamiltonian, OQKD reproduces the orthogonality, tridiagonal structure, and convergence behavior of the classical Lanczos algorithm ... Following the Lanczos algorithm, each vector is written as a polynomial of the rescaled Hamiltonian acting on the initial state ... obey the rescaled Lanczos recurrence"

    The OQKD basis is defined by the same three-term Lanczos recurrence (Eq. 10) that classical Lanczos uses; hence the numerical agreement in Fig. 2 is an identity-by-construction, not an independent confirmation of a prediction. This does not undermine the genuinely new part (GQSP implementation), but the 'reproduces Lanczos convergence' statement is definitional rather than empirical.

full rationale

The load-bearing recursion beta_{i+1}^2 = <psi_i|H^2|psi_i> - alpha_i^2 - beta_i^2 (Eq. 14) is derived in Appendix A from the Lanczos three-term identity and orthogonality; it is not an assumed fit. The projected eigenvalues are solved from measured moments, and no target energies are used to set alpha or beta, so there is no fitted-input-called-prediction circularity. The comparison with exact classical Lanczos is an external benchmark. There is one non-load-bearing self-citation (Ref. 21, cited for VQE difficulties) and no uniqueness theorem or ansatz smuggled in via self-citation; GQSP and block-encoding ingredients are independent prior results. The main caveats (no shot-noise/error-propagation analysis for Eq. 14; restart success-probability analysis deferred) are correctness/robustness gaps, not circularity. The only mild circularity is framing: OQKD is intentionally defined by the Lanczos recurrence, so 'reproduces Lanczos convergence' is true by construction. This warrants a low score, not a finding of substantive circularity.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

No new physical entities are postulated, and no free parameters are fitted to the target energies: the J1-J2 constants and the OQKD coefficients are fixed or computed. The ledger is dominated by domain assumptions about the availability and accuracy of block encodings and GQSP synthesis, plus the unproved well-conditioning of the restart polynomials, which is the main uncharged cost.

free parameters (2)
  • QKD/CQKD overlap threshold = 10^-9
    Regularization threshold used for the baseline comparisons in Fig. 2; chosen by hand. It affects where CQKD deviates from classical Lanczos, but is not part of the OQKD construction.
  • Restart hyperparameters = n_max ∈ {2,4,6}; n_k = largest numerically stable dimension
    Algorithm-2 control parameters for the numerical demonstrations; the RCQKD comparison depends on the (asymmetric) n_k selection rule.
assumptions (6)
  • domain assumption GQSP phase-factor existence and efficient classical synthesis (Motlagh-Wiebe, Ref. [41])
    The quantum implementation of the Lanczos polynomials relies on the GQSP theorem: for every normalized Laurent polynomial there exists an SU(2) phase-factor sequence realizing it, reliably computable numerically. If the phase-factor optimization fails for high-degree c_{nk}/Λ_n, the circuit construction breaks.
  • domain assumption Block encoding of H̃ with norm α ≥ ||H|| and LCU form with T terms
    Eqs. (11), (16). The algorithm starts from a block encoding of the rescaled Hamiltonian; the cost depends on T and α.
  • domain assumption Eigenvalue measurements of ⟨ψ|H̃|ψ⟩ and ⟨ψ|H̃²|ψ⟩ can be performed at required precision
    Algorithm 1 steps 3(b); the β recursion assumes these moments are known exactly. Finite-shot noise and its propagation are not analyzed in the paper.
  • domain assumption Physical scaling assumptions for the Kaniel-Paige bound (bandwidth O(N), gap O(1/N), squared overlap 2^{-N})
    Appendix D Eqs. (D4)-(D9): the D = O(N²) estimate and the claimed exponential shift rely on these thermodynamic-limit assumptions.
  • ad hoc to paper Restart polynomials Q^(k) remain well-conditioned with nearly constant Λ^(R)
    Sec. IV: claimed without proof; supported only by the 4×4 numerical example. The text announces a detailed analysis in Appendix D, but Appendix D provides no such analysis.
  • standard math Complementary polynomial Q(z) exists for the GQSP synthesis
    Appendix C1 uses spectral factorization/Fejér-Riesz; standard, but the numerical optimization route of [41] is invoked.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Orthogonal Quantum Krylov Diagonalisation." pith.science (2026). https://pith.science/paper/DWEPLW3G

@misc{pith2026260709476,
  author       = {Pith},
  title        = {Pith review of: Orthogonal Quantum Krylov Diagonalisation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DWEPLW3G}},
  note         = {Machine review of arXiv:2607.09476}
}
abstract

Quantum subspace-diagonalization methods, particularly Quantum Krylov Diagonalization (QKD), provide a promising route for computing low-energy spectra of quantum many-body Hamiltonians. However, existing quantum Krylov approaches rely on non-orthogonal Krylov bases, requiring overlap-matrix regularization that limits numerical stability and accuracy. In this work, we introduce an Orthogonal Quantum Krylov Diagonalization (OQKD) framework that reformulates the classical Lanczos recursion at the operator level, enabling an orthogonal quantum implementation of Krylov-subspace diagonalization. By expressing Lanczos vectors as polynomial transformations of the Hamiltonian, OQKD reproduces the orthogonality, tridiagonal structure, and convergence behavior of the classical Lanczos algorithm thus eliminating the need for overlap-matrix regularization. We further show that the required Lanczos polynomials can be implemented using block encoding and Generalized Quantum Signal Processing with the same asymptotic query complexity as Chebyshev-based QKD methods. Numerical simulations of the $J_1$--$J_2$ Heisenberg model confirm the classical Lanczos convergence and numerical stability of the proposed method, while the measurement-complexity scaling is established analytically. Building upon the OQKD framework, we then introduce a restarted state-preparation protocol that replaces a single high-degree polynomial transformation with a sequence of fixed low-degree transformations, maintaining an affordable block encoding success probability while retaining comparable convergence. These results establish OQKD as an orthogonal quantum analog of the classical Lanczos algorithm and identify the restarted protocol as a promising state-preparation strategy for Quantum Phase Estimation.

Figures

Figures reproduced from arXiv: 2607.09476 by the authors.

Figure 1
Figure 1. FIG. 1. Schematic comparison between conventional [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Results for the [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Probability of successfully implementing the normal [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Performance of the restarted OQKD state-preparation protocol. (a) Infidelity, [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. Success probability for preparing the Krylov ba [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 8 linked inside Pith

  1. [1]

    A. Y. Kitaev, Quantum measurements and the abelian stabilizer problem (1995), arXiv:quant-ph/9511026 [quant-ph]

    arXiv 1995

  2. [2]

    Yoshioka, M

    N. Yoshioka, M. Amico, W. Kirby, P. Jurcevic, A. Dutt, B. Fuller, S. Garion, H. Haas, I. Hamamura, A. Ivrii, et al., Nature Communications16, 5014 (2025)

    2025

  3. [3]

    Motta, W

    M. Motta, W. Kirby, I. Liepuoniute, K. J. Sung, J. Cohn, A. Mezzacapo, K. Klymko, N. Nguyen, N. Yoshioka, and J. E. Rice, Electronic Structure6, 013001 (2024)

    2024

  4. [4]

    Oumarou, P

    O. Oumarou, P. J. Ollitrault, S. Polla, and C. Gogolin, Optimizing and comparing quantum resources of statisti- cal phase estimation and krylov subspace diagonalization (2026), arXiv:2603.15552 [quant-ph]

    arXiv 2026

  5. [5]

    K. Wan, M. Berta, and E. Campbell, Physical Review Letters127, 160505 (2021)

    2021

  6. [6]

    Z. Ding, H. Li, L. Lin, H. Ni, L. Ying, and R. Zhang, Quantum8, 1190 (2024)

    2024

  7. [7]

    O’Brien, B

    T. O’Brien, B. Tarasinski, and B. Terhal, New Journal of Physics21, 023022 (2019)

    2019

  8. [8]

    H. Ni, H. Li, and L. Ying, Quantum7, 1055 (2023)

    2023

Show all 47 references
  1. [9]

    A. Y. Kitaev, A. H. Shen, and M. N. Vyalyi,Classical and Quantum Computation, Graduate Studies in Mathe- matics, Vol. 47 (American Mathematical Society, 2002)

    2002

  2. [10]

    Aharonov and M

    D. Aharonov and M. Ben-Or, SIAM Journal on Com- puting38, 1207 (2008), preliminary version presented at STOC 1997

    2008

  3. [11]

    S. Lee, J. Lee, H. Zhai, Y. Tong, A. M. Dalzell, A. Ku- mar, P. Helms, J. Gray, Z.-H. Cui, W. Liu, M. Kasto- ryano, R. Babbush, J. Preskill, D. R. Reichman, E. T. Campbell, E. F. Valeev, L. Lin, and G. K.-L. Chan, Na- ture communications14, 1952 (2023)

    1952

  4. [12]

    Louvet, T

    T. Louvet, T. Ayral, and X. Waintal, Phys. Rev. B113, 125112 (2026)

    2026

  5. [13]

    G. K.-L. Chan, Wiley Interdisciplinary Reviews: Com- putational Molecular Science2, 907 (2012)

    2012

  6. [14]

    J. R. McClean, R. Babbush, P. J. Love, and A. Aspuru- Guzik, Journal of Physical Chemistry Letters5, 4368 (2014)

    2014

  7. [15]

    Peruzzo, J

    A. Peruzzo, J. McClean, P. Shadbolt, M.-H. Yung, X.-Q. Zhou, P. J. Love, A. Aspuru-Guzik, and J. L. O’Brien, Nature Communications5, 4213 (2014)

    2014

  8. [16]

    Preskill, Quantum2, 79 (2018)

    J. Preskill, Quantum2, 79 (2018)

    2018

  9. [17]

    Cerezo, A

    M. Cerezo, A. Arrasmith, R. Babbush, S. C. Benjamin, S. Endo, K. Fujii, J. R. McClean, K. Mitarai, X. Yuan, 12 L. Cincio, and P. J. Coles, Nature Reviews Physics3, 625 (2021)

    2021

  10. [18]

    Lin and Y

    L. Lin and Y. Tong, PRX Quantum3, 010318 (2022)

    2022

  11. [19]

    Takagi, S

    R. Takagi, S. Endo, S. Minagawa, and M. Gu, Nature Communications14, 6439 (2023)

    2023

  12. [20]

    Bauer, S

    B. Bauer, S. Bravyi, M. Motta, and G. K.-L. Chan, Chemical Reviews120, 12685 (2020), pMID: 33090772, https://doi.org/10.1021/acs.chemrev.9b00829

    2020 doi

  13. [21]

    H. A. Akande, A. Perrin, B. Senjean, and M. Saubanère, Phys. Rev. A112, 032623 (2025)

    2025

  14. [22]

    J. R. McClean, S. Boixo, V. N. Smelyanskiy, R. Babbush, and H. Neven, Nature Communications9, 4812 (2018)

    2018

  15. [23]

    Uvarov and J

    A. Uvarov and J. Biamonte, Journal of Physics A: Math- ematical and Theoretical54, 245301 (2021)

    2021

  16. [24]

    S. Wang, E. Fontana, M. Cerezo, K. Sharma, A. Sone, L. Cincio, and P. J. Coles, Nature Communications12, 6961 (2021)

    2021

  17. [25]

    Cerezo, A

    M. Cerezo, A. Sone, T. Volkoff, L. Cincio, and P. J. Coles, Nature Communications12, 1791 (2021)

    2021

  18. [26]

    Zimborás, B

    Z. Zimborás, B. Koczor, Z. Holmes, E.-M. Borrelli, A. Gilyén, H.-Y. Huang, Z. Cai, A. Acín, L. Aolita, L. Banchi, F. G. S. L. Brandão, D. Cavalcanti, T. Cubitt, S. N. Filippov, G. García-Pérez, J. Goold, O. Kálmán, E. Kyoseva, M. A. C. Rossi, B. Sokolov, I. Taver- nelli, and S...

    2025 doi arXiv

  19. [27]

    A. N. Krylov, Izvestiya Akademii Nauk SSSR, Otdelenie Matematicheskikh i Estestvennykh Nauk7, 491 (1931), in Russian

    1931

  20. [28]

    Saad, SIAM Journal on Numerical Analysis17, 687 (1980)

    Y. Saad, SIAM Journal on Numerical Analysis17, 687 (1980)

    1980

  21. [29]

    Kirby, M

    W. Kirby, M. Motta, and A. Mezzacapo, Quantum7, 1018 (2023), arXiv:2208.00567 [quant-ph]

    2023 arXiv

  22. [31]

    McArdle, T

    S. McArdle, T. Jones, S. Endo, Y. Li, S. C. Benjamin, and X. Yuan, npj Quantum Information5, 75 (2019)

    2019

  23. [32]

    Motta, C

    M. Motta, C. Sun, A. T. K. Tan, M. J. O’Rourke, E. Ye, A. J. Minnich, F. G. S. L. Brandao, and G. K.-L. Chan, Nature Physics16, 205 (2020)

    2020

  24. [33]

    U. Baek, D. Hait, J. Shee, O. Leimkuhler, W. J. Huggins, T. F. Stetina, M. Head-Gordon, and K. B. Whaley, arXiv preprint arXiv:2205.09009 (2022), arXiv:2205.09009 [quant-ph]

    2022 arXiv

  25. [34]

    R. M. Parrish and P. L. McMahon, arXiv preprint arXiv:1909.08925 (2019)

    1909 arXiv

  26. [35]

    N. H. Stair, R. Huang, and F. A. Evangelista, Journal of Chemical Theory and Computation16, 2236 (2020)

    2020

  27. [36]

    J. Cohn, M. Motta, and R. M. Parrish, PRX Quantum 2, 040352 (2021)

    2021

  28. [37]

    Klymko, C

    K. Klymko, C. Mejuto-Zaera, S. J. Cotton, F. Wudarski, M. Urbanek, D. Hait, M. Head-Gordon, K. B. Whaley, J. Moussa, N. Wiebe, W. A. de Jong, and N. M. Tubman, PRX Quantum3, 020323 (2022)

    2022

  29. [38]

    Seki and S

    K. Seki and S. Yunoki, PRX Quantum2, 010333 (2021)

    2021

  30. [39]

    C. L. Cortes and S. K. Gray, Phys. Rev. A105, 022417 (2022)

    2022

  31. [40]

    G. H. Low and I. L. Chuang, Quantum3, 163 (2019)

    2019

  32. [41]

    Motlagh and N

    D. Motlagh and N. Wiebe, arXiv preprint arXiv:2308.01501 (2024), arXiv:2308.01501 [quant- ph]

    2024 arXiv

  33. [42]

    Sünderhauf, arXiv preprint arXiv:2312.00723 (2023), arXiv:2312.00723 [quant-ph]

    C. Sünderhauf, arXiv preprint arXiv:2312.00723 (2023), arXiv:2312.00723 [quant-ph]

    2023 arXiv

  34. [43]

    Patel, P

    S. Patel, P. Jayakumar, R. Huang, T. Zeng, and A. F. Izmaylov, Quantum seniority-based subspace expansion: Linear combinations of short-circuit unitary transfor- mations for the electronic structure problem (2025), arXiv:2509.01061 [quant-ph]

    2025

  35. [44]

    Saad,Numerical Methods for Large Eigenvalue Prob- lems, 2nd ed

    Y. Saad,Numerical Methods for Large Eigenvalue Prob- lems, 2nd ed. (SIAM, 2011)

    2011

  36. [45]

    A. M. Childs, R. Kothari, and R. D. Somma, SIAM Jour- nal on Computing46, 1920 (2017)

    1920

  37. [46]

    Brassard, P

    G. Brassard, P. Høyer, M. Mosca, and A. Tapp, Quantum amplitude amplification and estimation (2002)

    2002

  38. [47]

    N. M. Tubman, C. Mejuto-Zaera, J. M. Epstein, D. Hait, D. S. Levine, W. J. Huggins, Z. Jiang, J. R. McClean, R. Babbush, M. Head-Gordon, and K. B. Whaley, arXiv preprint arXiv:1809.05523 (2018), arXiv:1809.05523 [quant-ph]. Appendix A: Recursive determination of the Lanczos co...

    2018 arXiv

  39. [48]

    Calculation of the GQSP rotation angles Given the normalized polynomialep n(z), the corre- sponding GQSP circuit is synthesized by first construct- ing a complementary polynomialQ(z), Q(z) = nX k=0 qkzk,(C13) satisfying |epn(z)|2 +|Q(z)| 2 = 1,|z|= 1.(C14) The existence of the...

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.