Pith. sign in

REVIEW 2 major objections 5 minor 33 references

For large MaxCut graphs of fixed local structure, QAOA needs fewer measurement shots to hit the same relative accuracy as the graph grows.

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 · grok-4.5

2026-07-12 03:10 UTC pith:SS6VWHLM

load-bearing objection Clean, usable inverse-shot scaling for relative-error QAOA MaxCut under extensivity; the math holds and the result is new relative to the concentration literature. the 2 major comments →

arxiv 2607.03340 v1 pith:SS6VWHLM submitted 2026-07-03 quant-ph

Measurements Number Scaling in the Quantum Approximate Optimization Algorithm for MaxCut: A Statistical Analysis

Inbar Chefer , Uri Shaham , Adi Makmal This is my paper
classification quant-ph
keywords QAOAMaxCutshot complexityrelative concentrationstochastic gradient descentBenjamini-Schramm convergencefinite-difference gradientsvariational quantum algorithms
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.

This paper asks how many circuit measurements (shots) QAOA needs for the MaxCut problem, and how that number scales with graph size. Using concentration inequalities and optimization theory, the authors prove that when the expected cost grows linearly with the number of edges and the landscape smoothness constants scale the same way, the shot count per cost evaluation that is enough to keep a fixed relative error actually falls as one over the number of edges. The number of stochastic-gradient iterations needed stays independent of size. The same local-structure conditions that earlier work used to transfer angles from small graphs to large ones therefore also make the sampling budget shrink rather than grow. The result is backed by explicit sufficient bounds, a practitioner calibration recipe, and numerical checks on regular, Erdős–Rényi and random connected graphs. If the picture holds, large MaxCut instances become cheaper per relative-performance target, not more expensive.

Core claim

Under an extensive-cost lower bound and linear scaling of the Polyak-Łojasiewicz and Lipschitz constants, a sufficient shot budget per QAOA cost evaluation that guarantees fixed relative estimation error (or fixed relative SGD suboptimality) decreases as 1/m with the number of edges m, while the iteration count remains Θ(1).

What carries the argument

Relative concentration of the MaxCut cost: both expectation and variance scale linearly with m, so the relative standard deviation shrinks as 1/√m; Janson’s inequality then yields an inverse shot bound, and the same scaling plus PL/Lipschitz assumptions carries the argument through SGD.

Load-bearing premise

The expected QAOA cost must stay at least a fixed positive fraction of the number of edges throughout the optimization; if the cost stops being extensive, the inverse-shot claim fails.

What would settle it

On a sequence of bounded-degree graphs that keep the same local structure, measure the minimal shots needed to keep relative cost error below a fixed δ with fixed confidence: if that shot count does not fall roughly as 1/m, the central scaling claim is false.

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

If this is right

  • For Benjamini–Schramm convergent families (regular, sparse Erdős–Rényi, cycles) the total shot budget for fixed relative performance improves with size.
  • Angle-transfer methods that already avoid outer-loop optimization can also run with a shrinking measurement budget.
  • Finite-difference gradient estimators inherit the 1/m shot scaling; gate-wise parameter-shift estimators keep a size-independent shot count.
  • Practitioners can calibrate shots on small graphs and extrapolate by the inverse-size rule for larger instances of the same family.

Where Pith is reading between the lines

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

  • The same relative-concentration argument may apply to other local-cost combinatorial problems (e.g., Max-k-SAT or Ising models with bounded degree) once an extensive lower bound is verified.
  • If absolute rather than relative accuracy is required, as in many VQE energy targets, the 1/m advantage disappears.
  • Hardware-aware implementations that already use shallow fixed-angle QAOA could further cut classical post-processing by deliberately lowering the shot schedule with instance size.

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

2 major / 5 minor

Summary. The manuscript analyzes the measurement (shot) complexity of QAOA for MaxCut. Using Janson’s inequality on dependent edge terms and standard L-smooth + Polyak–Łojasiewicz SGD theory, it derives sufficient shot budgets to (a) estimate the cost to relative error δ with confidence 1−ε and (b) keep the SGD relative suboptimality below a target ξ*. Under an extensive-cost lower bound |⟨Cp⟩|≥κm and linear scaling of the PL and Lipschitz constants (Assumptions 1–3), the sufficient shots per cost evaluation scale as 1/m for finite-difference gradients (and as Θ(1) for parameter-shift), while the number of SGD iterations is Θ(1). The authors characterize the relevant graph classes via Benjamini–Schramm local limits, give practitioner calibration rules, and support the scalings with noiseless finite-shot simulations on regular, sparse Erdős–Rényi, and random connected graphs.

Significance. If the extensivity and linear PL/Lipschitz hypotheses hold for the intended families, the result is a genuine and useful contribution: it shows that a fixed relative performance target can become cheaper in shots as MaxCut instances grow, complementing known parameter-transfer / tree-QAOA concentration results. The finite-difference versus parameter-shift variance prefactors are treated carefully, the concentration argument via Janson is correctly applied to QAOA causal cones, and the paper supplies explicit sufficient conditions, practitioner heuristics, and public code. The claims are properly conditional rather than universal, which is appropriate. Strengths include transparent derivations, local-positivity lemmas supporting extensivity, and numerical checks that track the predicted 1/m trend on small but structured instances.

major comments (2)
  1. Assumption 1 (and Appendix A, Reasoning I) is load-bearing for the 1/m claim. Reasoning I lower-bounds |⟨Cp⟩| by m/2 using the (0,0) initialization together with monotonicity of the expected cost along the SGD trajectory. Noisy SGD need not be monotone pathwise. Either restrict the trajectory claim to expected progress / noise-free descent, or replace it by a weaker, rigorously controlled condition (e.g., that the optimizer remains in a region where |F|≥κm). Reasoning II (ensemble average under Benjamini–Schramm) and the local-positivity lemmas already give a cleaner route; elevating that route would remove the gap.
  2. Assumptions 2–3 and Appendix C.2: the scalings µ=Θ(m) and L=Θ(m) rest on strong convexity of the limiting per-edge objective on a local region U. That landscape hypothesis is not checked numerically (e.g., Hessian spectra or empirical PL constants on the calibration graphs). Because Results B–C and the claim T=Θ(1) depend on µ/L=Θ(1), a short numerical check or a more prominent caveat that the iteration/shot conclusions are conditional on this local strong-convexity property would make the load-bearing hypothesis transparent.
minor comments (5)
  1. Notation for the sample mean of the cost is inconsistent in places (ˆCp vs ⟨ˆCp⟩, e.g. Lemma 3 / Appendix B). Pick one convention and use it throughout.
  2. Fig. 4 reports minimal shot counts from a single graph instance per size; adding error bars or medians over the 10 instances already used elsewhere would better match the multi-instance protocol of Figs. 3 and 6–8.
  3. In Sec. 4.3 the relative gap dt is evaluated from exact state-vector costs while only the gradient is shot-noisy. This is fine for testing the theory, but the practitioner heuristics (Sec. 3.5) should note that full shot noise on both cost and gradient may require a modestly larger budget.
  4. Typographical: “Polyak- Lojasiewicz” should be “Polyak–Łojasiewicz” (or “Polyak-Lojasiewicz”) consistently; a few missing spaces after commas appear in the abstract and Sec. 1.
  5. The open Conjecture 10 is interesting but not needed for the main theorems; a one-sentence pointer that the proved lemmas already cover the tree-like and path-repeatable cases used in the scaling claims would help readers.

Circularity Check

0 steps flagged

No significant circularity: shot-scaling bounds follow from external concentration inequalities plus scoped extensivity/PL assumptions justified by independent local-limit arguments; only minor non-load-bearing self-reference to prior tree-QAOA work.

full rationale

The central claims (Result A: np = Ω(1/m) for fixed relative error δ under |Fp| ≥ κ m; Result B: same for finite-difference SGD under μ = Θ(m); Result C: T = Θ(1) under L = Θ(m)) are derived from Janson’s inequality (Eq. 2 / Lemma 3), the Farhi et al. variance bound Var(C) ≤ 2 cvp m (Eq. 34), and standard L-smooth + PL SGD analysis (Lemma 5, Thm. 4/6). Assumptions 1–3 are explicitly stated and supported by external Benjamini–Schramm convergence plus original local-positivity lemmas (App. A Lemmas 7–9) and Hessian-convergence arguments (App. C.2); none of these reduce the target scaling to a fitted constant or to a definition. Numerical experiments independently check the predicted scalings rather than fitting and re-predicting. The only self-reference is a non-load-bearing complement to prior tree-QAOA parameter-transfer results (including one co-authored paper), which is not used to force the shot bounds. The derivation chain is therefore self-contained against its stated external inputs and assumptions.

Axiom & Free-Parameter Ledger

0 free parameters · 6 axioms · 0 invented entities

The inverse-shot claim rests on three extensivity assumptions (cost, PL constant, Lipschitz constant) plus standard concentration and optimization tools. No free parameters are fitted; the constants κ, μ, L are existence statements. No new physical entities are introduced.

axioms (6)
  • domain assumption Assumption 1: |Fp(θ;G)| ≥ κ m for some κ > 0 independent of m (extensive cost).
    Invoked after Thm. 2 and Thm. 4 to convert absolute variance bounds into 1/m shot scalings; justified in Appendix A via initialization and Benjamini-Schramm arguments but not proved for every optimization trajectory.
  • domain assumption Assumption 2: the PL constant satisfies μ = Θ(m).
    Used in Thm. 4 and Cor. 6.1; argued via local strong convexity of the per-edge cost under Benjamini-Schramm convergence (Appendix C.2).
  • domain assumption Assumption 3: the gradient Lipschitz constant satisfies L = Θ(m).
    Used to obtain size-independent iteration count; same local-limit reasoning as Assumption 2.
  • standard math Janson’s inequality for dependent summands with maximum dependency degree Λ ≤ 2 c_vp.
    Applied in Lemma 3 / Thm. 2 to bound the relative deviation of the sample-mean cost.
  • domain assumption Local invexity / PL inequality holds in the region reached by the SGD trajectory.
    Required for the relative-gap recursion of Lemma 5; the paper notes the landscape is neither globally convex nor invex.
  • domain assumption Variance bound Var(C) ≤ 2 c_vp m of Farhi et al. (2014).
    Used throughout to upper-bound gradient estimator variance; the constant c_vp depends only on max degree and depth.

pith-pipeline@v1.1.0-grok45 · 40293 in / 2771 out tokens · 24929 ms · 2026-07-12T03:10:00.050543+00:00 · methodology

0 comments
read the original abstract

We provide a statistical analysis of the measurement (shot) requirements of the quantum approximate optimization algorithm (QAOA) for the MaxCut problem. We derive sufficient conditions on the number of shots per cost operator evaluation to: (a) estimate the expected cost to within a relative error $\delta$ and a confidence $1-\epsilon$, and (b) ensure SGD-based parameter optimization converges to a target relative suboptimality level with high probability. In addition, we provide an explicit bound on the number of SGD iterations required to reach the target accuracy. Our analysis reveals an unexpected scaling phenomenon: for specific graph classes, which we formally characterize, the total shot budget needed to achieve a fixed relative-performance metric decreases as the instance size grows. This result complements earlier cost function concentration arguments regarding parameter optimization redundancy, thereby highlighting the potential for high-performance, low-overhead QAOA implementations for large-scale MaxCut instances. To assist practitioners, we translate our analytical findings into practical rules of thumb for shot-budget allocation and validate these results with numerical simulations, offering new insights into the interplay between graph size, structural complexity, and resource requirements in QAOA.

Figures

Figures reproduced from arXiv: 2607.03340 by Adi Makmal, Inbar Chefer, Uri Shaham.

Figure 1
Figure 1. Figure 1: Schematic illustration of the main prediction developed in this paper. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Schematic illustration of the concentration mechanism underlying the shot scaling results. At fixed depth p and under our extensivity assumptions, the expectation ⟨Cp⟩ scales linearly with the number of edges m, while the standard deviation σ(Cp) scales only as √m. Consequently, the relative cost Cp/⟨Cp⟩ becomes more concentrated as m increases, with σ(Cp/⟨Cp⟩) ∝ 1/ √m, so fewer shots suffice to achieve th… view at source ↗
Figure 3
Figure 3. Figure 3: Scaling of the QAOA cost variance with graph size. [PITH_FULL_IMAGE:figures/full_fig_p028_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Required shots for relative-error cost estimation versus graph size. ˆ [PITH_FULL_IMAGE:figures/full_fig_p029_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Numerical illustration of inverse shot scaling on random 3-regular graphs. [PITH_FULL_IMAGE:figures/full_fig_p030_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Sample variance of depth-1 gradient estimators on random 3-regular graphs, comparing (a) the finite-difference approximation (left) and (b) the parameter-shift esti￾mator (right). Solid blue and red lines denote the estimators for ∂Cp/∂γ1 and ∂Cp/∂β1, respectively. A fixed number of shots np per cost evaluation is used in each experiment: np = 1000 in the finite-difference experiment, and np = 10 in the pa… view at source ↗
Figure 7
Figure 7. Figure 7: Numerical performance of QAOA-SGD with shots scaled inversely with graph [PITH_FULL_IMAGE:figures/full_fig_p032_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Relative cost gap (dt) trajectories on random 3-regular graphs under inverse shot scaling using the finite-difference approximation. Graphs with N ∈ {10, 14, 18, 22} nodes, corresponding to m ∈ {15, 21, 27, 33} edges, were considered, with 10 graph instances per size and 5 independent finite-shot SGD trajectories per instance, at circuit depth p = 1. All trajectories were initialized from the same fixed pa… view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

33 extracted references · 3 canonical work pages · 1 internal anchor

  1. [1]

    Farhi, E., Goldstone, J., Gutmann, S.: A Quantum Approximate Opti- mization Algorithm. arXiv. https://doi.org/10.48550/arXiv.1411.4028. http://arxiv.org/abs/1411.4028

    Pith/arXiv arXiv doi:10.48550/arxiv.1411.4028

  2. [2]

    Kashapogu, R., Hasib, S., Rasool, A.: Exploring the Versatility of QAOA: A Comprehensive Review. (2024). https://doi.org/10.1109/ ICCCNT61001.2024.10725610

    Pith/arXiv arXiv 2024

  3. [3]

    Nature Reviews Physics3(9), 625–644 (2021)

    Cerezo, M., Arrasmith, A., Babbush, R., Benjamin, S.C., Endo, S., Fujii, K., McClean, J.R., Mitarai, K., Yuan, X., Cincio, L.,et al.: Variational quantum algorithms. Nature Reviews Physics3(9), 625–644 (2021). https: //doi.org/10.1038/s42254-021-00348-9

    doi:10.1038/s42254-021-00348-9 2021

  4. [4]

    https://arxiv.org/abs/1812.04170

    Brandao, F.G.S.L., Broughton, M., Farhi, E., Gutmann, S., Neven, H.: For Fixed Control Parameters the Quantum Approximate Optimization Algorithm’s Objective Function Value Concentrates for Typical Instances (2018). https://arxiv.org/abs/1812.04170

    Pith/arXiv arXiv 2018

  5. [5]

    Physical Review X10(2), 021067 (2020)

    Zhou, L., Wang, S.-T., Choi, S., Pichler, H., Lukin, M.D.: Quan- tum approximate optimization algorithm: Performance, mechanism, and implementation on near-term devices. Physical Review X10(2), 021067 (2020). https://doi.org/10.1103/PhysRevX.10.021067

    doi:10.1103/physrevx.10.021067 2020

  6. [6]

    Wauters, M.M., Mbeng, G.B., Santoro, G.E.: Polynomial scaling of the quantum approximate optimization algorithm for ground-state prepa- ration of the fully connectedp-spin ferromagnet in a transverse field. Phys. Rev. A102, 062404 (2020). https://doi.org/10.1103/PhysRevA. 102.062404

    doi:10.1103/physreva 2020

  7. [7]

    Measurements Number Scaling in QAOA: A Statistical Analysis51 https://doi.org/10.1103/PhysRevA.97.022304

    Wang, Z., Hadfield, S., Jiang, Z., Rieffel, E.G.: Quantum approximate optimization algorithm for MaxCut: A fermionic view97(2), 022304. Measurements Number Scaling in QAOA: A Statistical Analysis51 https://doi.org/10.1103/PhysRevA.97.022304. Accessed 2024-11-16

    doi:10.1103/physreva.97.022304 2024

  8. [8]

    In: Proceedings of the 21st ACM International Conference on Computing Frontiers, pp

    Rajakumar, J., Golden, J., B¨ artschi, A., Eidenbenz, S.: Trainability bar- riers in low-depth qaoa landscapes. In: Proceedings of the 21st ACM International Conference on Computing Frontiers, pp. 199–206 (2024). https://doi.org/10.1145/3649153.3649204

    doi:10.1145/3649153.3649204 2024

  9. [9]

    Nature Reviews Physics, 1–16 (2025)

    Larocca, M., Thanasilp, S., Wang, S., Sharma, K., Biamonte, J., Coles, P.J., Cincio, L., McClean, J.R., Holmes, Z., Cerezo, M.: Barren plateaus in variational quantum computing. Nature Reviews Physics, 1–16 (2025). https://doi.org/10.1038/s42254-025-00813-9

    doi:10.1038/s42254-025-00813-9 2025

  10. [10]

    Quantum5, 491 (2021)

    Sack, S.H., Serbyn, M.: Quantum annealing initialization of the quantum approximate optimization algorithm. Quantum5, 491 (2021). https:// doi.org/10.22331/q-2021-07-01-491

    doi:10.22331/q-2021-07-01-491 2021

  11. [11]

    Quantum Machine Intelligence6(2), 38 (2024)

    Amosy, O., Danzig, T., Lev, O., Porat, E., Chechik, G., Makmal, A.: Iteration-free quantum approximate optimization algorithm using neural networks. Quantum Machine Intelligence6(2), 38 (2024). https://doi.org/ 10.1007/s42484-024-00159-y

    work page doi:10.1007/s42484-024-00159-y 2024

  12. [12]

    https: //arxiv.org/abs/1908.08862

    Streif, M., Leib, M.: Training the Quantum Approximate Optimization Algorithm without access to a Quantum Processing Unit (2019). https: //arxiv.org/abs/1908.08862

    Pith/arXiv arXiv 2019

  13. [13]

    Wurtz, J., Lykov, D.: Fixed-angle conjectures for the quantum approx- imate optimization algorithm on regular maxcut graphs. Phys. Rev. A 104, 052419 (2021). https://doi.org/10.1103/PhysRevA.104.052419

    doi:10.1103/physreva.104.052419 2021

  14. [14]

    Quantum9, 1892 (2025)

    Wybo, E., Leib, M.: Missing puzzle pieces in the performance landscape of the quantum approximate optimization algorithm. Quantum9, 1892 (2025). https://doi.org/10.22331/q-2025-10-22-1892

    doi:10.22331/q-2025-10-22-1892 2025

  15. [15]

    https: //arxiv.org/abs/2503.12789

    Farhi, E., Gutmann, S., Ranard, D., Villalonga, B.: Lower bounding the MaxCut of high girth 3-regular graphs using the QAOA (2025). https: //arxiv.org/abs/2503.12789

    arXiv 2025

  16. [16]

    Random Structures and Algorithms24(2004)

    Janson, S.: Large deviations for sums of partly dependent random vari- ables. Random Structures and Algorithms24(2004). https://doi.org/10. 1002/rsa.20008

    2004

  17. [17]

    Physical Review A99(3), 032331 (2019)

    Schuld, M., Bergholm, V., Gogolin, C., Izaac, J., Killoran, N.: Evaluat- ing analytic gradients on quantum hardware. Physical Review A99(3), 032331 (2019). https://doi.org/10.1103/PhysRevA.99.032331

    doi:10.1103/physreva.99.032331 2019

  18. [18]

    https://arxiv.org/abs/1811.04968

    Bergholm, V., Izaac, J., Schuld, M., Gogolin, C., Ahmed, S., Ajith, V., Alam, M.S., Alonso-Linaje, G., AkashNarayanan, B., Asadi, A., Arrazola, 52Measurements Number Scaling in QAOA: A Statistical Analysis J.M., Azad, U., Banning, S., Blank, C., Bromley, T.R., Cordier, B.A., Ceroni, J., Delgado, A., Matteo, O.D., Dusko, A., Garg, T., Guala, D., Hayes, A.,...

    Pith/arXiv arXiv 2022

  19. [19]

    https://github.com/PennyLaneAI/pennylane

    Accessed: 2023-03-23. https://github.com/PennyLaneAI/pennylane

    2023

  20. [20]

    Quantum6, 677 (2022)

    Wierichs, D., Izaac, J., Wang, C., Lin, C.Y.-Y.: General parameter-shift rules for quantum gradients. Quantum6, 677 (2022). https://doi.org/10. 22331/q-2022-03-30-677

    2022

  21. [21]

    https://www-m5.ma.tum.de/foswiki/pub/M5/Allgemeines/ MA4801 2018S/ML notes main.pdf

    Wolf, M.M.: Mathematical Foundations of Supervised Learning (2018). https://www-m5.ma.tum.de/foswiki/pub/M5/Allgemeines/ MA4801 2018S/ML notes main.pdf

    2018

  22. [22]

    Electronic Journal of Probability6(none), 1–13 (2001)

    Benjamini, I., Schramm, O.: Recurrence of Distributional Limits of Finite Planar Graphs. Electronic Journal of Probability6(none), 1–13 (2001). https://doi.org/10.1214/EJP.v6-96

    doi:10.1214/ejp.v6-96 2001

  23. [23]

    MIT OpenCourseWare, 18.S096: Topics in Mathematics of Data Science (Fall 2015), Lec- ture Notes

    Rahman, M.: Session 17: Local Convergence of Graphs and Enu- meration of Spanning Trees. MIT OpenCourseWare, 18.S096: Topics in Mathematics of Data Science (Fall 2015), Lec- ture Notes. Accessed 2026-02-15 (2015). https://ocw.mit.edu/ courses/18-s096-topics-in-mathematics-of-data-science-fall-2015/ 55ff9f23be313f3beefe692dda95aff9 MIT18 S096F15 Ses17.pdf

    2015

  24. [24]

    Lecture notes / book draft

    van der Hofstad, R.: Random Graphs and Complex Networks, Volume I. Lecture notes / book draft. Accessed 2026-02-18 (2017). https://rhofstad. win.tue.nl/NotesRGCN.pdf

    2026

  25. [25]

    https: //doi.org/10.22331/q-2024-01-18-1231

    Sureshbabu, S.H., Herman, D., Shaydulin, R., Basso, J., Chakrabarti, S., Sun, Y., Pistoia, M.: Parameter setting in quantum approximate opti- mization of weighted problems8, 1231 2305.15201 [quant-ph]. https: //doi.org/10.22331/q-2024-01-18-1231. Accessed 2024-11-16

    Pith/arXiv arXiv doi:10.22331/q-2024-01-18-1231 2024

  26. [26]

    https://arxiv.org/abs/2005.08747 Measurements Number Scaling in QAOA: A Statistical Analysis53

    Farhi, E., Gamarnik, D., Gutmann, S.: The Quantum Approximate Opti- mization Algorithm Needs to See the Whole Graph: Worst Case Examples (2020). https://arxiv.org/abs/2005.08747 Measurements Number Scaling in QAOA: A Statistical Analysis53

    Pith/arXiv arXiv 2020

  27. [27]

    Physics Reports986, 1–128 (2022)

    Tilly, J., Chen, H., Cao, S., Picozzi, D., Setia, K., Li, Y., Grant, E., Woss- nig, L., Rungger, I., Booth, G.H., Tennyson, J.: The variational quantum eigensolver: A review of methods and best practices. Physics Reports986, 1–128 (2022). https://doi.org/10.1016/j.physrep.2022.08.003

    doi:10.1016/j.physrep.2022.08.003 2022

  28. [28]

    Journal of Chemical Theory and Computation20(6), 2390–2403 (2024)

    Zhu, L., Liang, S., Yang, C., Li, X.: Optimizing shot assignment in vari- ational quantum eigensolver measurement. Journal of Chemical Theory and Computation20(6), 2390–2403 (2024). https://doi.org/10.1021/acs. jctc.3c01113

    doi:10.1021/acs 2024

  29. [29]

    PRX Quantum1, 020312 (2020)

    Sanders, Y.R., Berry, D.W., Costa, P.C.S., Tessler, L.W., Wiebe, N., Gidney, C., Neven, H., Babbush, R.: Compilation of fault-tolerant quan- tum heuristics for combinatorial optimization. PRX Quantum1, 020312 (2020). https://doi.org/10.1103/PRXQuantum.1.020312

    work page doi:10.1103/prxquantum.1.020312 2020

  30. [30]

    A novel framework for Shot number minimization in Quantum Variational Algorithms

    Kahani, S.S., Nobakhti, A.: A novel framework for shot number minimiza- tion in quantum variational algorithms. arXiv preprint arXiv:2307.04035 (2023). https://doi.org/10.48550/arXiv.2307.04035

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2307.04035 2023

  31. [31]

    https://arxiv.org/abs/1401

    Lyons, R.: Factors of IID on Trees (2016). https://arxiv.org/abs/1401. 4197

    2016

  32. [32]

    Accessed: 2023-03-23 (2023)

    PennyLane: PennyLane: A Library for Quantum Machine Learning, Quantum Chemistry, and Quantum Optimization. Accessed: 2023-03-23 (2023). https://pennylane.ai/

    2023

  33. [33]

    Applied Optimization, vol

    Nesterov, Y.: Introductory Lectures on Convex Optimization: A Basic Course. Applied Optimization, vol. 87. Kluwer Academic Publishers, Boston, MA (2004)

    2004