Pith. sign in

REVIEW 3 major objections 3 minor 1 cited by

This paper shows that Grover's unstructured search can be solved by Riemannian gradient ascent with a Grover-compatible retraction, achieving O(√N log(1/ε)) queries.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-03 17:43 UTC pith:4HV3JLFR

load-bearing objection A careful, mostly sound optimization-theoretic re-derivation of Grover's speedup, with a real caveat about the known-M assumption that the paper underplays. the 3 major comments →

arxiv 2512.08432 v3 pith:4HV3JLFR submitted 2025-12-09 quant-ph math.OC

A Grover-compatible manifold optimization algorithm for quantum search

Zhijian Lai , Dong An , Jiang Hu , Zaiwen Wen This is my paper
classification quant-ph math.OC MSC 81P6890C2690C4865K05 PACS 03.67.Ac
keywords Grover's algorithmunstructured searchRiemannian gradient ascentunitary manifoldGrover-compatible retractionPolyak–Łojasiewicz inequalityquantum query complexityclassical simulability
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Grover's quadratic speedup is usually attributed to the structure of the oracle and diffusion operators. This paper argues that the same speedup emerges from a standard first-order optimization method: Riemannian gradient ascent on the unitary manifold, provided the update step is restricted to a 'Grover-compatible retraction' built from physically implementable oracle and diffusion gates. The authors prove a local Riemannian Polyak–Łojasiewicz inequality with μ = 1/2 and show the Riemannian Lipschitz constant is O(√(N/M)), which yields linear convergence to a uniform superposition of marked states in O(√(N/M) log(1/ε)) iterations. If correct, this recovers the optimal quantum query complexity from an optimization-theoretic perspective and suggests that adaptive gradient-based angle choices, rather than fixed π rotations, are the essential mechanism behind the speedup.

Core claim

The paper claims that unstructured quantum search can be formulated as maximizing f(U)=Tr(H U ψ0 U†) over U(N), where H is the projector onto the marked subspace and ψ0 the initial uniform state. The central discovery is the Grover-compatible retraction: a map from the tangent space back to the unitary manifold that is a product of gates e^{iθH} and e^{iθψ0}. Using it, every Riemannian gradient ascent update remains in a fixed two-dimensional invariant subspace (the Grover plane), so the entire iteration is classically simulable. The authors establish a local Riemannian μ-PL inequality with μ=1/2, i.e., ∥grad f(U)∥² ≥ 1−f(U) for f(U)≥1/2, and bound the Riemannian Lipschitz constant by L_Rie

What carries the argument

The carrying object is the Grover-compatible retraction, a family of retractions on the unitary manifold defined as finite products of the oracle gate e^{iθH} and the diffusion gate e^{iθψ0} (the paper gives explicit 5-, 6-, and 8-factor variants). Each retraction is a valid retraction on the two-dimensional subspace W = span{X0, Y0} of the tangent space, where X0=[H,ψ0] and Y0=i[H,X0]; this ensures the Riemannian gradient at every iterate lies in W, states stay in the Grover plane, and the update is physically implementable. The convergence proof hinges on two quantitative ingredients: the local Riemannian PL inequality (μ=1/2) and the Riemannian Lipschitz constant L_Rie = O(√(N/M)), which

Load-bearing premise

The step size and all gate angles require knowing the number of marked items M; if M is unknown, the proven O(√(N/M) log(1/ε)) bound does not apply.

What would settle it

Run the 2D classical simulation for the 5-factor retraction with fixed step size 1/L_Rie for a specific instance, say N=2^20, M=1, ε=10^{-4}, and count the number of e^{iθH} calls until the success probability reaches 1−ε. If the count exceeds ⌈6 L_Rie log(1/ε)⌉ (or the final probability is below 1−ε at that iteration), the central claim is false.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • If the central claim holds, Grover's quadratic speedup is a corollary of a general optimization principle—gradient ascent under a PL-type inequality—applied to the unitary manifold, not an isolated algorithmic trick.
  • The classical simulability theorem means the optimal circuit angles can be computed on a classical computer before execution; the quantum device only evaluates the resulting gate sequence, so the quantum advantage is confined to the oracle queries.
  • The bound O(√(N/M) log(1/ε)) is asymptotically optimal in N and linear in log(1/ε), so the procedure avoids the 'soufflé' overshoot problem when M is known.
  • Because retractions of different lengths give different pullback Lipschitz constants, the choice of retraction directly affects the provable iteration count; the numerics indicate that longer retractions can reduce total oracle calls despite using more gates per iteration.
  • The framework supplies a template for translating other manifold optimization algorithms into quantum circuits, suggesting a systematic route from optimization theory to quantum circuit design.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • An unstated consequence is that the 2D classical simulation implies this particular circuit family is classically simulable and can reach at most the Grover speedup; the interesting open question is which tasks survive if the Grover-compatible retraction is generalized beyond two-level subspace dynamics.
  • The paper does not address the case where the number of marked items M is unknown; a practical version would need to estimate M or use adaptive step sizes, and a proof of the speedup under uncertainty would be a natural extension.
  • The same PL-inequality analysis might explain the convergence of imaginary-time evolution and double-bracket quantum algorithms, as those methods are known to follow Riemannian gradient flows; if so, the optimization viewpoint could unify a broader class of quantum algorithms.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. The paper recasts Grover's unstructured search as the maximization problem max_{U∈U(N)} f(U)=Tr(HUψ0U†), where H is the marked-subspace projector and ψ0 is the uniform initial state. The authors introduce 'Grover-compatible retractions' whose updates are finite products of the oracle and diffusion gates e^{iθH} and e^{iθψ0}, prove that the Riemannian gradient remains in a fixed two-dimensional subspace (Theorem 3.3), and show that the full iteration is classically simulable via a 2×2 dynamical system (Theorem 3.6). On the analysis side, they prove a pullback descent inequality with Riemannian Lipschitz constant L_Rie=O(√(N/M)) (Prop. 4.3), a local PL inequality with μ=1/2 (Prop. 4.5), and derive T=O(L_Rie log(1/ε)) iterations, i.e. O(√(N/M) log(1/ε)) oracle calls (Theorem 4.6). Numerical experiments with fixed step size and exact line search are reported for the 5-, 6-, and 8-factor retractions.

Significance. If the claims hold, the paper provides a genuinely new conceptual bridge: Grover's quadratic speedup is recovered through standard Riemannian optimization tools, with explicit retractions that are both physically implementable and classically precomputable. The self-contained proof of the invariant 2D gradient subspace, the tight first/second-order retraction bounds, and the PL-based linear convergence analysis are valuable contributions. The paper also ships reproducible NumPy simulations and is transparent about the use of LLM assistance. The main caveat is that the central complexity claim is conditional on knowing the number of marked items M, and one of the theorem's hypotheses (ε≤q0) excludes many of the reported experiments; these are fixable but should be stated precisely.

major comments (3)
  1. [Algorithm 1 / Eq. (30) / Theorem 3.6] The algorithm and complexity guarantee presuppose knowledge of M. Algorithm 1 line 1 sets q0:=Tr(Hψ0); Eq. (30) defines L_Rie through c0=√(2M(N−M))/N; Theorem 3.6 initializes the 2D classical simulation with Ψ0 built from q0=M/N. In the black-box Grover setting, the oracle g does not reveal M, so the fixed step size t_k=1/L_Rie cannot be formed and the classical precomputation of angles cannot be initialized. The paper never flags this as an assumption. The authors should either explicitly state that M is known (as in variants of Grover with a known number of marked items) or add a quantum-counting preprocessing step and include its query cost in the total O(√(N/M) log(1/ε)) bound. As written, the abstract's claim overreaches the usual unstructured-search problem.
  2. [Theorem 4.6 and §5] Theorem 4.6 is proved only under 0<ε≤q0, but the numerical section uses ε=10^{-4} with M=1 and n up to 25; for n≥14, q0=2^{-n}<ε, so the theorem does not cover those runs. Moreover, for fixed ε and growing N, q0 eventually falls below ε, so the claimed O(√(N/M) log(1/ε)) bound is not established in the regime most relevant to large N. The proof actually yields the more general sufficient condition T≥4L_Rie log(1/(2q0))+1+2L_Rie log(1/ε), i.e. O(√(N/M) log(N/ε)) when M=1. The paper should either state this general bound as the main complexity result, or justify restricting to ε≤q0, and should align the experiments with the proven range.
  3. [Theorem 4.4 / Definition 3.4] Theorem 4.4 is stated as a direct application of [13, Corollary 2.9], but Definition 3.4 defines the retraction only on the two-dimensional subbundle WU, not on the full tangent bundle TU. The hypotheses of [13] are formulated for retractions on the whole manifold, and the paper does not prove that a subbundle retraction satisfies those hypotheses or that the proof carries over unchanged. Since Theorem 4.6 is self-contained, this gap does not invalidate the main O(L_Rie log(1/ε)) claim, but Theorem 4.4 as stated needs a proof or a caveat.
minor comments (3)
  1. [Eq. (27) / Lemma 3.2] Eq. (27) attributes the equality ∥[X0,ψ0]∥_F=c0 to Lemma 3.2, but Lemma 3.2 only proves the norm identities for X0=[H,ψ0] and Y0=i[H,X0], not for [X0,ψ0]. A short proof of the latter identity should be added, since it is used in Lemma 4.2.
  2. [Section 5] The text says the total number of H-exp calls equals the iteration count T multiplied by 2, 3, or 4 for the 5-, 6-, and 8-factor retractions. However, the 5-factor retraction in Eq. (20) contains three e^{iθH} factors, the 6-factor retraction also contains three, and the 8-factor retraction contains four. The multipliers should be corrected, or the counting convention clarified, so that the numerical verification is consistent with the query model.
  3. [Algorithm 1] Line 4 of Algorithm 1 contains the garbled symbol 'ggrad f(Uk)'; this should be the skew-Hermitian part of grad f(Uk), e.g. \widetilde{grad} f(Uk), consistent with the notation used in Section 3.2.

Circularity Check

0 steps flagged

No significant circularity: the convergence and complexity claims are derived in-paper from the projector structure; the only caveat is an unstated precondition that M (hence q0 and LRie) is known, which is a modeling gap rather than a circular reduction.

full rationale

The central derivation is self-contained. The invariant 2D subspace (Theorem 3.3) is proved from H^2 = H and the explicit forms of the gates, not imported from Grover or from the authors' prior work. The 2D classical simulation (Theorem 3.6) follows by direct matrix representation. The Lipschitz constant LRie = 2 + N/sqrt(2M(N-M)) in Proposition 4.3 is obtained from the tight retraction bounds in Lemma 4.2, which are proved for the 5-factor retraction, together with the standard Euclidean descent lemma; no fitted parameter is renamed as a prediction. The PL inequality in Proposition 4.5 is an exact identity from Lemma 3.2 (||grad f(U)||^2 = 2q(1-q)), and Theorem 4.6 is a standard PL-plus-descent argument. The only self-citation in the load-bearing mathematical sense is absent: [35] is cited solely for the human-AI proof methodology and is not used to justify any theorem. The genuine caveat is that Algorithm 1 sets q0 := Tr(H psi0) and the fixed step size uses LRie = 2 + N/sqrt(2M(N-M)), both requiring the marked-set size M; in the black-box unstructured-search setting the oracle does not reveal M, and the paper does not flag this assumption. That is a precondition/assumption gap, not an equivalence of inputs and outputs, so it does not constitute circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 0 invented entities

No new physical entities are postulated. The Grover-compatible retractions are explicit mathematical constructions (products of given gates) whose properties are proved, not brute-force postulated entities. The free-parameter list is empty because the step size and angles are computed from the problem data (N, M, and the state), not fitted to simulation results.

axioms (5)
  • domain assumption H is an idempotent Hermitian projector (H† = H = H²), and M = rank(H) < N.
    Problem 1 and all later structure (eigenvalues {0,1}, ad³_H = ad_H, Lemma 3.2, Theorem 3.3) require H to be a projector. This is the quantum-search setup (marked-subspace projector).
  • domain assumption Gates e^{iθH} and e^{iθψ0} are available as physical primitives for arbitrary real θ.
    Grover-compatible retractions (Definition 3.4, Proposition 3.5) and the circuit implementations (Figs. 1-2) consist of such exponentials; the query complexity counts calls to e^{iθH}.
  • domain assumption The initial state |ψ0⟩ is known, fixed, and satisfies q0 = ⟨ψ0|H|ψ0⟩ ∈ (0,1).
    Problem 1, Theorem 3.6 ('initialization triplet (x0,y0,q0) := (1,0,q0)'), and the phase-I analysis in Theorem 4.6.
  • standard math Euclidean descent lemma and generic Riemannian nonconvex machinery.
    Used in Proposition 4.3 (descent lemma [42, Lemma 1.2.3]) and Theorem 4.4 ([13, Corollary 2.9]).
  • domain assumption The number of marked items M (hence q0 = M/N) is known to the algorithm (used to choose t_k = 1/L_Rie and to run the 2D simulation).
    Theorems 4.4 and 4.6 set t_k = 1/L_Rie with L_Rie = O(√(N/M)); Theorem 3.6 needs q0. This is not flagged as a limitation in the paper.

pith-pipeline@v1.3.0-alltime-deepseek · 24046 in / 41536 out tokens · 345936 ms · 2026-08-03T17:43:38.052742+00:00 · methodology

0 comments
read the original abstract

Grover's algorithm is a fundamental quantum algorithm that offers a quadratic speedup for the unstructured search problem by alternately applying physically implementable oracle and diffusion operators. In this paper, we reformulate the unstructured search as a maximization problem on the unitary manifold and solve it via the Riemannian gradient ascent (RGA) method. To overcome the difficulty that generic RGA updates do not, in general, correspond to physically implementable quantum operators, we introduce Grover-compatible retractions to restrict RGA updates to valid oracle and diffusion operators. Theoretically, we establish a local Riemannian $\mu$-Polyak-{\L}ojasiewicz (PL) inequality with $\mu = \tfrac{1}{2}$, which yields a linear convergence rate of $1 - \kappa^{-1}$ toward the global solution. Here, the condition number $\kappa = L_{\mathrm{Rie}} / \mu$, where $L_{\mathrm{Rie}}$ denotes the Riemannian Lipschitz constant of the gradient. Taking into account both the geometry of the unitary manifold and the special structure of the cost function, we show that $L_{\mathrm{Rie}} = O(\sqrt{N})$ for problem size $N = 2^n$. Consequently, the resulting iteration complexity is $O(\sqrt{N} \log(1/\varepsilon))$ for attaining an $\varepsilon$-accurate solution, which matches the quadratic speedup of $O(\sqrt{N})$ achieved by Grover's algorithm. These results demonstrate that an optimization-based viewpoint can offer fresh conceptual insights and lead to new advances in the design of quantum algorithms.

Figures

Figures reproduced from arXiv: 2512.08432 by Dong An, Jiang Hu, Zaiwen Wen, Zhijian Lai.

Figure 1
Figure 1. Figure 1: Quantum circuit generated by Riemannian gradient ascent method with the exponen [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Quantum circuit generated by Riemannian gradient ascent method with the Grover [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Representative optimization trajectory for [PITH_FULL_IMAGE:figures/full_fig_p021_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Scaling of the total number of H-exp calls with problem size √ N for different Grover-compatible retractions and step size strategies. Left: total H-exp calls for the 5-, 6-, and 8-factor retractions under fixed step size tk = 1/LRie and exact line search (els). The fitted slopes s ≈ const, shown in the legend, confirm the linear dependence T = O( √ N). Right: zoom on the exact line search, highlighting th… view at source ↗
Figure 5
Figure 5. Figure 5: Scaling of the total number of H-exp calls with the target accuracy ε for different Grover-compatible retractions and step size strategies at fixed problem size N = 215. Left: total H-exp calls for the 5-factor retraction with fixed step size tk = 1/LRie and for the 5-, 6-, and 8-factor retractions under exact line search (els), plotted against log10(1/ε). The fitted slopes s ≈ const, shown in the legend, … view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Achieving double-logarithmic precision dependence in optimization-based quantum unstructured search

    quant-ph 2026-03 unverdicted novelty 7.0

    Riemannian modified Newton optimization on quantum search achieves quadratic convergence and O(√(N/M) log log(1/ε)) complexity when M/N is known.

Reference graph

Works this paper leans on

59 extracted references · 3 linked inside Pith · cited by 1 Pith paper

  1. [1]

    Efficient Riemannian algorithms for optimization under unitary matrix constraint

    Traian Abrudan, Jan Eriksson, and Visa Koivunen. Efficient Riemannian algorithms for optimization under unitary matrix constraint. In2008 IEEE International Conference on Acoustics, Speech and Signal Processing, pages 2353–2356. IEEE, 2008

    2008

  2. [2]

    Princeton University Press, 2008

    P-A Absil, Robert Mahony, and Rodolphe Sepulchre.Optimization Algorithms on Matrix Manifolds. Princeton University Press, 2008. 23

    2008

  3. [3]

    Projection-like retractions on matrix manifolds

    P-A Absil and Jérôme Malick. Projection-like retractions on matrix manifolds. SIAM Journal on Optimization, 22(1):135–158, 2012

    2012

  4. [4]

    Using quantum amplitude amplification in genetic algorithms.Expert Systems with Applications, 209:118203, 2022

    Giovanni Acampora, Roberto Schiattarella, and Autilia Vitiello. Using quantum amplitude amplification in genetic algorithms.Expert Systems with Applications, 209:118203, 2022

    2022

  5. [5]

    Newton’s method on Riemannian manifolds and a geometric model for the human spine

    Roy L Adler, Jean-Pierre Dedieu, Joseph Y Margulies, Marco Martens, and Mike Shub. Newton’s method on Riemannian manifolds and a geometric model for the human spine. IMA Journal of Numerical Analysis, 22(3):359–390, 2002

    2002

  6. [6]

    Variable time amplitude amplification and quantum algorithms for linear algebra problems

    Andris Ambainis. Variable time amplitude amplification and quantum algorithms for linear algebra problems. In STACS’12 (29th Symposium on Theoretical Aspects of Computer Science), volume 14, pages 636–647. LIPIcs, 2012

    2012

  7. [7]

    Grover’s quantum algorithm applied to global optimization.SIAM Journal on Optimization, 15(4):1170–1184, 2005

    William P Baritompa, David W Bulger, and Graham R Wood. Grover’s quantum algorithm applied to global optimization.SIAM Journal on Optimization, 15(4):1170–1184, 2005

    2005

  8. [8]

    Quan- tum lower bounds by polynomials.Journal of the ACM (JACM), 48(4):778–797, 2001

    Robert Beals, Harry Buhrman, Richard Cleve, Michele Mosca, and Ronald De Wolf. Quan- tum lower bounds by polynomials.Journal of the ACM (JACM), 48(4):778–797, 2001

    2001

  9. [9]

    Strengths and weaknesses of quantum computing.SIAM Journal on Computing, 26(5):1510–1523, 1997

    Charles H Bennett, Ethan Bernstein, Gilles Brassard, and Umesh Vazirani. Strengths and weaknesses of quantum computing.SIAM Journal on Computing, 26(5):1510–1523, 1997

    1997

  10. [10]

    Post-quantum cryptography

    Daniel J Bernstein. Post-quantum cryptography. InEncyclopedia of Cryptography, Security and Privacy, pages 1846–1847. Springer, 2025

    2025

  11. [11]

    Quantum machine learning.Nature, 549(7671):195–202, 2017

    Jacob Biamonte, Peter Wittek, Nicola Pancotti, Patrick Rebentrost, Nathan Wiebe, and Seth Lloyd. Quantum machine learning.Nature, 549(7671):195–202, 2017

    2017

  12. [12]

    An Introduction to Optimization on Smooth Manifolds

    Nicolas Boumal. An Introduction to Optimization on Smooth Manifolds. Cambridge Uni- versity Press, 2023

    2023

  13. [13]

    Global rates of convergence for nonconvex optimization on manifolds

    Nicolas Boumal, Pierre-Antoine Absil, and Coralia Cartis. Global rates of convergence for nonconvex optimization on manifolds. IMA Journal of Numerical Analysis, 39(1):1–33, 2019

    2019

  14. [14]

    Searching a quantum phone book.Science, 275(5300):627–628, 1997

    Gilles Brassard. Searching a quantum phone book.Science, 275(5300):627–628, 1997

    1997

  15. [15]

    Quantum amplitude ampli- fication and estimation.arXiv preprint quant-ph/0005055, 2000

    Gilles Brassard, Peter Hoyer, Michele Mosca, and Alain Tapp. Quantum amplitude ampli- fication and estimation.arXiv preprint quant-ph/0005055, 2000

    Pith/arXiv arXiv 2000

  16. [16]

    On Grover’s search algorithm from a quantum in- formation geometry viewpoint

    Carlo Cafaro and Stefano Mancini. On Grover’s search algorithm from a quantum in- formation geometry viewpoint. Physica A: Statistical Mechanics and its Applications, 391(4):1610–1625, 2012

    2012

  17. [17]

    Quantum reinforcement learning

    Daoyi Dong, Chunlin Chen, Hanxiong Li, and Tzyh-Jong Tarn. Quantum reinforcement learning. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 38(5):1207–1220, 2008

    2008

  18. [18]

    A Grover-search based quantum learning scheme for classification.New Journal of Physics, 23(2):023020, 2021

    Yuxuan Du, Min-Hsiu Hsieh, Tongliang Liu, and Dacheng Tao. A Grover-search based quantum learning scheme for classification.New Journal of Physics, 23(2):023020, 2021

    2021

  19. [19]

    A quantum algorithm for finding the minimum.arXiv preprint quant-ph/9607014, 1996

    Christoph Durr and Peter Hoyer. A quantum algorithm for finding the minimum.arXiv preprint quant-ph/9607014, 1996

    Pith/arXiv arXiv 1996

  20. [20]

    On the Barzilai-Borwein method

    Roger Fletcher. On the Barzilai-Borwein method. InOptimization and control with appli- cations, pages 235–256. Springer, 2005. 24

    2005

  21. [21]

    Grover adaptive search for constrained polynomial binary optimization.Quantum, 5:428, 2021

    Austin Gilliam, Stefan Woerner, and Constantin Gonciulea. Grover adaptive search for constrained polynomial binary optimization.Quantum, 5:428, 2021

    2021

  22. [22]

    Quantum singular value transformation and beyond: exponential improvements for quantum matrix arithmetics

    András Gilyén, Yuan Su, Guang Hao Low, and Nathan Wiebe. Quantum singular value transformation and beyond: exponential improvements for quantum matrix arithmetics. In Proceedings of the 51st Annual ACM SIGACT Symposium on Theory of Computing, pages 193–204, 2019

    2019

  23. [23]

    Double-bracket quantum algorithms for diagonalization.Quantum, 8:1316, 2024

    Marek Gluza. Double-bracket quantum algorithms for diagonalization.Quantum, 8:1316, 2024

    2024

  24. [24]

    Double-bracket quantum algorithmsfor quantum imaginary- time evolution.arXiv preprint arXiv:2412.04554, 2024

    Marek Gluza, Jeongrak Son, Bi Hong Tiang, René Zander, Raphael Seidel, Yudai Suzuki, Zoë Holmes, and NellyHY Ng. Double-bracket quantum algorithmsfor quantum imaginary- time evolution.arXiv preprint arXiv:2412.04554, 2024

    arXiv 2024

  25. [25]

    Applying Grover’s algorithm to AES: quantum resource estimates

    Markus Grassl, Brandon Langenberg, Martin Roetteler, and Rainer Steinwandt. Applying Grover’s algorithm to AES: quantum resource estimates. In International Workshop on Post-Quantum Cryptography, pages 29–43. Springer, 2016

    2016

  26. [26]

    A fast quantum mechanical algorithm for database search

    Lov K Grover. A fast quantum mechanical algorithm for database search. InProceedings of the twenty-eighth annual ACM symposium on Theory of computing, pages 212–219, 1996

    1996

  27. [27]

    Fixed-point quantum search.Physical Review Letters, 95(15):150501, 2005

    Lov K Grover. Fixed-point quantum search.Physical Review Letters, 95(15):150501, 2005

    2005

  28. [28]

    Brian C. Hall. Lie Groups, Lie Algebras, and Representations: An Elementary Introduc- tion, volume 222 of Graduate Texts in Mathematics. Springer International Publishing Switzerland, 2 edition, 2015

    2015

  29. [29]

    Quantum state tomography via nonconvex Riemannian gradient descent.Physical Review Letters, 132(24):240804, 2024

    Ming-ChienHsu, En-JuiKuo, Wei-HsuanYu, Jian-FengCai, andMin-HsiuHsieh. Quantum state tomography via nonconvex Riemannian gradient descent.Physical Review Letters, 132(24):240804, 2024

    2024

  30. [30]

    A brief introduction to manifold optimization

    Jiang Hu, Xin Liu, Zai-Wen Wen, and Ya-Xiang Yuan. A brief introduction to manifold optimization. Journal of the Operations Research Society of China, 8(2):199–248, 2020

    2020

  31. [31]

    The Riemannian Barzilai-Borwein method with nonmonotone line search and the matrix geometric mean computation.IMA Journal of Numerical Analysis, 38(1):495–517, 2018

    Bruno Iannazzo and Margherita Porcelli. The Riemannian Barzilai-Borwein method with nonmonotone line search and the matrix geometric mean computation.IMA Journal of Numerical Analysis, 38(1):495–517, 2018

    2018

  32. [32]

    Linear convergence of gradient and proximal-gradient methods under the Polyak–Łojasiewicz condition

    Hamed Karimi, Julie Nutini, and Mark Schmidt. Linear convergence of gradient and proximal-gradient methods under the Polyak–Łojasiewicz condition. InJoint European con- ference on machine learning and knowledge discovery in databases, pages 795–811. Springer, 2016

    2016

  33. [33]

    Oxford University Press., 2006

    Phillip Kaye, Raymond Laflamme, and Michele Mosca.An Introduction to Quantum Com- puting. Oxford University Press., 2006

    2006

  34. [34]

    The Art of Computer Programming, volume 3

    Donald Ervin Knuth. The Art of Computer Programming, volume 3. Pearson Education, 1997

    1997

  35. [35]

    Advancing mathematical research via human–AI interactive theorem proving.arXiv preprint arXiv:2512.09443, 2025

    Chenyi Li, Zhijian Lai, Dong An, Jiang Hu, and Zaiwen Wen. Advancing mathematical research via human–AI interactive theorem proving.arXiv preprint arXiv:2512.09443, 2025

    arXiv 2025

  36. [36]

    Quantum comb tomography via learning isometries on Stiefel manifold

    Ze-Tong Li, Xin-Lin He, Cong-Cong Zheng, Yu-Qian Dong, Tian Luan, Xu-Tao Yu, and Zai-Chen Zhang. Quantum comb tomography via learning isometries on Stiefel manifold. Physical Review Letters, 134(1):010803, 2025. 25

    2025

  37. [37]

    A topological property of real analytic subsets.Coll

    Stanislaw Lojasiewicz. A topological property of real analytic subsets.Coll. du CNRS, Les équations aux dérivées partielles, 117(87-89):2, 1963

    1963

  38. [38]

    QGOpt: Rie- mannian optimization for quantum technologies.Scipost physics, 10(3):079, 2021

    Ilia Luchnikov, Alexander Ryzhov, Sergey Filippov, and Henni Ouerdane. QGOpt: Rie- mannian optimization for quantum technologies.Scipost physics, 10(3):079, 2021

    2021

  39. [39]

    Introduction to Information Retrieval

    Christopher D Manning. Introduction to Information Retrieval. Cambridge University Press, 2008

    2008

  40. [40]

    Grand unification of quantum algorithms

    John M Martyn, Zane M Rossi, Andrew K Tan, and Isaac L Chuang. Grand unification of quantum algorithms. PRX Quantum, 2(4):040203, 2021

    2021

  41. [41]

    Geometric strategy for the optimal quantum search

    Akimasa Miyake and Miki Wadati. Geometric strategy for the optimal quantum search. Physical Review A, 64(4):042317, 2001

    2001

  42. [42]

    Springer US, Boston, MA, 2004

    Yurii Nesterov.Introductory Lectures on Convex Optimization: A Basic Course, volume 87 of Applied Optimization. Springer US, Boston, MA, 2004

    2004

  43. [43]

    Cambridge University Press, 2010

    Michael A Nielsen and Isaac L Chuang.Quantum Computation and Quantum Information. Cambridge University Press, 2010

    2010

  44. [44]

    Springer, 2006

    Jorge Nocedal and Stephen J Wright.Numerical Optimization. Springer, 2006

    2006

  45. [45]

    Gradient methods for minimizing functionals.Zhurnal Vychis- litel’noi Matematiki i Matematicheskoi Fiziki, 3(4):643–653, 1963

    Boris Teodorovich Polyak. Gradient methods for minimizing functionals.Zhurnal Vychis- litel’noi Matematiki i Matematicheskoi Fiziki, 3(4):643–653, 1963

    1963

  46. [46]

    On the Barzilai and Borwein choice of steplength for the gradient method

    Marcos Raydan. On the Barzilai and Borwein choice of steplength for the gradient method. IMA Journal of Numerical Analysis, 13(3):321–326, 1993

    1993

  47. [47]

    Fast convergence to non-isolated minima: four equivalent conditions forC2 functions

    Quentin Rebjock and Nicolas Boumal. Fast convergence to non-isolated minima: four equivalent conditions forC2 functions. Mathematical Programming, pages 1–49, 2024

    2024

  48. [48]

    Double- bracket quantum algorithms for high-fidelity ground state preparation

    Matteo Robbiati, Edoardo Pedicillo, Andrea Pasquale, Xiaoyue Li, Oriel Kiss, Andrew Wright, Renato Farias, Khanh Uyen Giang, Jeongrak Son, Johannes Knörzer, et al. Double- bracket quantum algorithms for high-fidelity ground state preparation. arXiv preprint arXiv:2408.03987, 2024

    arXiv 2024

  49. [49]

    Cholesky QR-based retraction on the generalized Stiefel manifold

    Hiroyuki Sato and Kensuke Aihara. Cholesky QR-based retraction on the generalized Stiefel manifold. Computational Optimization and Applications, 72(2):293–308, 2019

    2019

  50. [50]

    Amplitude estimation without phase estimation.Quantum Information Processing, 19(2):75, 2020

    Yohichi Suzuki, Shumpei Uno, Rudy Raymond, Tomoki Tanaka, Tamiya Onodera, and Naoki Yamamoto. Amplitude estimation without phase estimation.Quantum Information Processing, 19(2):75, 2020

    2020

  51. [51]

    Grover’s algorithm is an approximation of imaginary-time evolution.arXiv preprint, 2025

    Yudai Suzuki, Marek Gluza, Jeongrak Son, Bi Hong Tiang, Nelly HY Ng, and Zoë Holmes. Grover’s algorithm is an approximation of imaginary-time evolution.arXiv preprint, 2025. arXiv:2507.15065

    arXiv 2025

  52. [52]

    Double-bracket algorithm for quantum signal processing without post-selection

    Yudai Suzuki, Bi Hong Tiang, Jeongrak Son, Nelly HY Ng, Zoë Holmes, and Marek Gluza. Double-bracket algorithm for quantum signal processing without post-selection. arXiv preprint arXiv:2504.01077, 2025

    arXiv 2025

  53. [53]

    Quantum speed-up of Markov chain based algorithms

    Mario Szegedy. Quantum speed-up of Markov chain based algorithms. In 45th Annual IEEE Symposium on Foundations of Computer Science, pages 32–41. IEEE, 2004

    2004

  54. [54]

    A feasible method for optimization with orthogonality con- straints

    Zaiwen Wen and Wotao Yin. A feasible method for optimization with orthogonality con- straints. Mathematical Programming, 142(1):397–434, 2013. 26

    2013

  55. [55]

    Optimizing quantum circuits with Riemannian gradient flow.Physical Review A, 107(6):062421, 2023

    Roeland Wiersema and Nathan Killoran. Optimizing quantum circuits with Riemannian gradient flow.Physical Review A, 107(6):062421, 2023

    2023

  56. [56]

    Strategies for optimizing double-bracket quantum algorithms

    Li Xiaoyue, Matteo Robbiati, Andrea Pasquale, Edoardo Pedicillo, Andrew Wright, Stefano Carrazza, and Marek Gluza. Strategies for optimizing double-bracket quantum algorithms. arXiv preprint arXiv:2408.07431, 2024

    Pith/arXiv arXiv 2024

  57. [57]

    Fixed-point quantum search with an optimal number of queries.Physical Review Letters, 113(21):210501, 2014

    Theodore J Yoder, Guang Hao Low, and Isaac L Chuang. Fixed-point quantum search with an optimal number of queries.Physical Review Letters, 113(21):210501, 2014

    2014

  58. [58]

    Grover’s quantum searching algorithm is optimal

    Christof Zalka. Grover’s quantum searching algorithm is optimal. Physical Review A, 60(4):2746, 1999

    1999

  59. [59]

    Role of Riemannian geome- try in double-bracket quantum imaginary-time evolution

    René Zander, Raphael Seidel, Li Xiaoyue, and Marek Gluza. Role of Riemannian geome- try in double-bracket quantum imaginary-time evolution. InInternational Conference on Geometric Science of Information, pages 105–114. Springer, 2025. 27

    2025