Pith. sign in

REVIEW 2 major objections 4 minor 130 references

Bitflip gauges that align best solutions with the device ground state turn amplitude-damping noise into an asset for 100-qubit QAOA-style optimization.

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

2026-07-13 03:32 UTC pith:RQOQWN6M

load-bearing objection Clean 100-qubit head-to-head showing that bitflip gauge + iterative warm-start improves AR at zero circuit cost; the comparison is controlled enough to be useful even if classical post-processing and offline angles share some of the credit. the 2 major comments →

arxiv 2607.09368 v1 pith:RQOQWN6M submitted 2026-07-10 quant-ph

Quantum Approximate Optimization via Noise-Directed Adaptive Warm-Starting

Filip B. Maciejewski , Stuart Hadfield , Oscar Wallis , George Pennington , Sebastian Brandhofer , Stefan Woerner , Daniel J. Egger , Davide Venturelli This is my paper
classification quant-ph
keywords QAOAwarm-startnoise-directed remappingbitflip gaugecombinatorial optimizationIsing HamiltoniansNISQamplitude damping
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.

Hardware noise and shallow circuits still limit quantum approximate optimization of combinatorial problems. This paper introduces Noise-Directed Adaptive Warm-Starting (ND-AWS): an iterative loop that repeatedly samples a warm-started, time-block QAOA ansatz, then applies a bitflip gauge so that the best-found bitstring becomes logically identical to the device’s physical all-zero ground state. Because amplitude-damping noise also drives the device toward that state, the algorithm and the noise pull in the same direction. On 100-qubit Erdős–Rényi and 3-regular Ising instances run on a superconducting processor, the gauge-transformed loop produces higher and smoother approximation ratios than the identical iterative warm-start procedure without gauges, at exactly the same circuit depth and shot budget. The authors therefore argue that simple, noise-aware re-encoding can extract better solutions from present-day hardware without extra gates.

Core claim

On 100-qubit random Ising Hamiltonians, the ND-AWS loop (p=1 time-block warm-start QAOA plus local Hamming-distance post-processing) yields approximation ratios 0.974–1.0 (ER-10), 0.969–1.0 (ER-20) and 0.989–1.0 (RG-3). These figures are generally higher, less rugged, and obtained with comparable or fewer iterations than the same iterative warm-start algorithm run without the bitflip gauge, while using identical circuit resources.

What carries the argument

Noise-Directed Adaptive Warm-Starting: at each iteration the cost and phase-separator Hamiltonians are conjugated by a bitflip operator P_y built from the previous best bitstring y; this maps y onto the device’s |0…0⟩ state while the warm-start bias parameter c keeps the mixer and initial state fixed toward that physical ground state.

Load-bearing premise

That variational angles optimized offline on a noiseless simulator transfer well enough to the real device that the observed performance gap can be attributed mainly to the gauge transform rather than to angle mismatch or device drift.

What would settle it

Rerun the identical 100-qubit instances with on-device re-optimization of the angles (or with a noise model that includes the actual angle-transfer error) and check whether the approximation-ratio advantage of the gauge-transformed loop over the non-gauge loop disappears.

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

If this is right

  • Shallow QAOA-style circuits can be made more noise-tolerant simply by re-encoding the best sample into the device ground state at every iteration, without increasing gate count.
  • The same bitflip-gauge idea can be combined with adaptive bias schedules, multi-solution tree search, or classical local solvers inside the feedback loop.
  • Once deeper circuits become reliable, the same noise-directed warm-start loop is expected to improve further, as indicated by the paper’s MPS simulations.
  • The framework supplies a practical template for testing whether other hardware platforms (ions, atoms, spins) also benefit from aligning algorithmic bias with their dominant relaxation channels.

Where Pith is reading between the lines

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

  • If amplitude damping is the dominant error on a given platform, ND-AWS may convert that platform’s noise into a free local-search bias, reducing the classical post-processing needed to reach high approximation ratios.
  • The method’s simplicity suggests it could be inserted as a drop-in outer loop around any warm-start or reverse-annealing ansatz that already admits a bitflip gauge.
  • Because the gauge lives entirely in classical post-processing of the Hamiltonian coefficients, the technique is immediately portable to any gate-model or analog device that can implement signed ZZ interactions.

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 / 4 minor

Summary. The paper introduces Noise-Directed Adaptive Warm-Starting (ND-AWS), an iterative warm-start QAOA variant that applies bitflip gauge transformations so the best sample from the previous iteration is logically mapped to the device ground state |0…0⟩, thereby aligning the ansatz bias with amplitude-damping-like noise. Using a Time-Block p=1 phase separator (fraction of largest-magnitude edges), offline noiseless angle optimization, and optional Hamming-distance quadratic search (HDQS), the authors run the method on 100 qubits of ibm_boston for 30 random Ising instances (ER-10, ER-20, RG-3). They report approximation ratios in the 0.97–1.0 range (best of three runs) and show that the gauge-transformed loop generally yields higher or smoother ARs than an otherwise identical non-gauge iterative warm-start baseline at identical circuit depth and shot budget, with supporting amplitude-damping simulations and MPS baselines.

Significance. If the reported AR gaps and noise-adaptivity hold, ND-AWS supplies a simple, zero-extra-depth practical heuristic that improves the reliability of shallow warm-start QAOA on present-day superconducting hardware. The controlled head-to-head (identical shots, post-processing, and termination), the small-scale AD-noise study (Fig. 2), the MPS scaling checks, and the planned open-source release constitute concrete, reproducible contributions that advance the NDAR/warm-start literature and give the community a usable framework for adaptive bias schedules and hybrid classical–quantum loops. Even without a quantum-advantage claim, high-quality 100-qubit Ising results of this type remain scarce and therefore useful for benchmarking.

major comments (2)
  1. [Section III A, Appendix B 2] Section III A and Appendix B 2: Variational angles are obtained from a noiseless p=1 expectation-value simulator (COBYQA + basin-hopping) and transferred to hardware for both ND-AWS and Standard IWS. Because the physical circuits differ (uniform bias + gauge-transformed RZZ signs versus qubit-dependent RY/mixer angles with fixed signs), any systematic mismatch between the noiseless landscape and the real noisy landscape can interact differently with the two encodings. The relative AR gap is therefore not cleanly isolated to the gauge. The manuscript should quantify the transfer quality (e.g., noiseless versus hardware energy for the same angles) or at least discuss this confound explicitly; the small-scale AD simulations of Fig. 2 use an analogous transfer and still favor ND, which should be highlighted as supporting evidence.
  2. [Appendix D, Tables II–III] Appendix D and Tables II–III: Primary claims and the “highest-quality” framing rest on results that include HDQS classical local search after every QPU sample. Without HDQS the absolute ARs drop substantially and the ND–Standard gap becomes smaller and less consistent (especially ER-20, where mean ARs are nearly identical). Because the abstract and introduction attribute the improvement primarily to the bitflip gauge, the no-HDQS data should be presented more prominently as the purer quantum comparison, and the contribution of classical post-processing to the reported quality should be stated more carefully.
minor comments (4)
  1. [Table I, Section III A] Table I and Section III A: The phase-separator fractions (25 % / 10 % / 80 %) and the piecewise bias schedule c are presented as fixed heuristics. A short justification or sensitivity check (beyond the 500-qubit mean-AR scan of App. B 2) would help readers assess robustness.
  2. [Figure 3] Figure 3 caption and main text: Optimal solutions are clipped to 10^{-3} on the inverted log scale; this is fine for visualization but should be stated once in the caption so that AR = 1.0 is not misread as a numerical floor.
  3. [Appendix A 4 b] Appendix A 4 b: The noiseless equivalence proof is clear, yet the experimental section never explicitly reminds the reader that any observed gap must therefore arise from noise (or from differential angle transfer). A one-sentence cross-reference would tighten the narrative.
  4. Minor typographical issues: “anastz” (p. 7), inconsistent hyphenation of “Warm-Start”/“warm-start”, and a few missing spaces around citations. These are easily cleaned.

Circularity Check

0 steps flagged

No significant circularity: empirical AR claims rest on direct QPU sampling plus classical energy evaluation against independent Emin/Emax; self-citations supply only background technique.

full rationale

The paper’s load-bearing claims are experimental approximation ratios obtained by sampling bit-strings from ibm_boston, evaluating the classical Ising energy of those bit-strings, and comparing against Emin/Emax found by independent classical solvers (Burer-Monteiro / TABU). The ND-AWS algorithm itself is a transparent composition of previously published Warm-Start QAOA, Time-Block QAOA and NDAR gauge transforms; the only new content is the iterative combination and the 100-qubit hardware comparison. Appendix A proves noiseless equivalence of the gauge-transformed and standard warm-start circuits by direct algebraic manipulation (Eqs. A19–A20), which is a self-contained identity, not a circular derivation. Offline COBYQA angles are transferred from a noiseless p=1 expectation-value simulator, but the paper never re-labels those fitted angles as a “prediction” of the hardware AR; the AR values reported in Table II and Fig. 3 are measured, not fitted. Self-citations to the authors’ earlier NDAR papers supply the gauge idea and are not used to justify uniqueness or to forbid alternatives. Consequently the derivation chain does not reduce any claimed result to its own inputs by construction. A residual score of 1 acknowledges the minor, non-load-bearing self-citation background without elevating it to circularity.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

The central experimental claim rests on standard quantum-circuit and Ising-model assumptions plus a handful of hand-chosen algorithmic hyperparameters (bias schedule, interaction fraction, termination rule). No new physical entities are postulated; the free parameters are purely algorithmic knobs whose values are stated explicitly.

free parameters (3)
  • bias schedule c
    Hand-chosen piecewise schedule (c=0.5 then {0.1,0.05} then {0.05,0.025}); two values tried per iteration and the better kept. Directly controls the strength of the warm-start bias that drives the reported AR curves.
  • phase-separator interaction fraction
    25 % (ER-10), 10 % (ER-20), 80 % (RG-3) of largest-magnitude edges; chosen to keep circuit depth feasible. Changes the expressivity/noise trade-off of every ansatz.
  • termination patience
    Stop after 3 consecutive non-improving iterations (or when known optimum is hit). Affects total shot budget and final reported AR.
axioms (3)
  • domain assumption Amplitude-damping-like noise is the dominant noise component that the gauge transform can usefully exploit on the target superconducting device.
    Invoked throughout Sec. II and II E; supported by prior NDAR literature and by the controlled simulations of Fig. 2, but not independently re-measured on ibm_boston for these circuits.
  • ad hoc to paper Offline noiseless p=1 angle optimization transfers to the real device sufficiently well that observed AR differences can be attributed to the gauge rather than to angle mismatch.
    Stated in Sec. III A; no on-device re-optimization is performed.
  • domain assumption Classical solvers (Burer-Monteiro / TABU) correctly identify Emin and Emax for the 100-qubit instances, so reported approximation ratios are accurate.
    Used for every AR number in Tables II–III and Figs. 3,6.

pith-pipeline@v1.1.0-grok45 · 31111 in / 2618 out tokens · 36195 ms · 2026-07-13T03:32:00.234998+00:00 · methodology

0 comments
read the original abstract

Progress towards a quantum advantage using known heuristic methods for combinatorial optimization is impeded by hardware noise and limited qubit count. Here, we propose a noise-aware adaptive approach to quantum approximate optimization, Noise-Directed Adaptive Warm-Starting (ND-AWS), that builds on recent concepts such as Warm-Start QAOA and Noise-Directed Adaptive Remapping. By leveraging bitflip gauge transformations, our algorithm exploits amplitude-damping-like noise components. We experimentally implement high-performance quantum optimization ans\"atze on 100-qubit Ising Hamiltonians, showing that ND-AWS generally improves the performance over a non-gauge-transformed iterative Warm-Starting variant, at no additional circuit cost. This places our results among the highest-quality demonstrations of quantum optimization with similar ans\"atze at this scale. Crucially, the simplicity of the framework opens the door for future enhancements such as adaptive bias schedules, and integration with classical solvers.

Figures

Figures reproduced from arXiv: 2607.09368 by Daniel J. Egger, Davide Venturelli, Filip B. Maciejewski, George Pennington, Oscar Wallis, Sebastian Brandhofer, Stefan Woerner, Stuart Hadfield.

Figure 1
Figure 1. Figure 1: FIG. 1. Illustration of the Noise-Directed Adaptive Warm-Starting algorithm implementation introduced in this paper. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Distance to the optimal solution, 1 [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Distance to the optimal solution, 1 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Distance to the optimal solution, 1 [PITH_FULL_IMAGE:figures/full_fig_p018_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Optimized mean Approximation Ratio of [PITH_FULL_IMAGE:figures/full_fig_p019_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Distance to the optimal solution, 1 [PITH_FULL_IMAGE:figures/full_fig_p021_6.png] 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

130 extracted references · 1 canonical work pages

  1. [1]

    optimize variational parameters of the ansatz (e.g., offline in simulations),

  2. [2]

    sample on quantum hardware from the ansatz,

  3. [3]

    gauge-transformH C so that the (ideal) ground state of the QPU is logically equivalent to the best- found sample (see Eq. (4)),

  4. [4]

    (6) and (7)),

    optionally, update the hyperparameters of the method, such as the WS bias valuec(see Eqs. (6) and (7)),

  5. [5]

    The procedure stops when it meets a termination crite- rion, for example, a shot budget is reached or the best so- lution does not improve over the past few iterations [46]

    move to the next iteration. The procedure stops when it meets a termination crite- rion, for example, a shot budget is reached or the best so- lution does not improve over the past few iterations [46]. Fig. 1 illustrates our algorithm. As the utilized quantum resources increase, we expect our approach to explore in- creasingly non-local neighborhoods of t...

  6. [6]

    5 Noise-Directed (solid) Noiseless Standard (dashed) 0.01 Noise strength q 0.02 0.03 0.04 0.05 0.10 FIG

    gives a further related approach, where sampled bit- strings are iteratively used to set spin-dependent phases for the oscillating drive terms therein. 5 Noise-Directed (solid) Noiseless Standard (dashed) 0.01 Noise strength q 0.02 0.03 0.04 0.05 0.10 FIG. 2. Distance to the optimal solution, 1−AR, of the best-found sample obtained in simulated runs ofn= ...

    1917

  7. [7]

    Abbas, A

    A. Abbas, A. Ambainis, B. Augustino, A. B¨ artschi, H. Buhrman, C. Coffrin, G. Cortiana, V. Dunjko, D. J. Egger, B. G. Elmegreen,et al., Challenges and opportu- nities in quantum optimization, Nature Reviews Physics , 718–735 (2024)

    2024

  8. [8]

    Farhi, J

    E. Farhi, J. Goldstone, and S. Gutmann, A Quantum Approximate Optimization Algorithm, arXiv preprint arXiv:1411.4028 (2014)

    Pith/arXiv arXiv 2014

  9. [9]

    Hadfield, Z

    S. Hadfield, Z. Wang, B. O’gorman, E. G. Rieffel, D. Venturelli, and R. Biswas, From the quantum ap- proximate optimization algorithm to a quantum alter- nating operator ansatz, Algorithms12, 34 (2019)

    2019

  10. [10]

    T. Koch, D. E. B. Neira, Y. Chen, G. Cortiana, D. J. Eg- ger, R. Heese, N. N. Hegade, A. G. Cadavid, R. Huang, T. Itoko, T. Kleinert, P. M. Xavier, N. Mohseni, J. A. Montanez-Barrera, K. Nakano, G. Nannicini, C. O’Meara, J. Pauckert, M. Proissl, A. Ramesh, M. Schicker, N. Shimada, M. Takeori, V. Valls, D. V. Bulck, S. Woerner, and C. Zoufal, Quantum optim...

    Pith/arXiv arXiv 2025

  11. [11]

    Bravyi, A

    S. Bravyi, A. Kliesch, R. Koenig, and E. Tang, Obstacles to Variational Quantum Optimization from Symmetry Protection, Physical Review Letters125, 260505 (2020)

    2020

  12. [12]

    Dupont, B

    M. Dupont, B. Evert, M. J. Hodson, B. Sundar, S. Jef- frey, Y. Yamaguchi, D. Feng, F. B. Maciejewski, S. Had- field, M. S. Alam,et al., Quantum-enhanced greedy combinatorial optimization solver, Science Advances9, eadi0487 (2023)

    2023

  13. [13]

    L. T. Brady and S. Hadfield, Iterative quantum algo- rithms for maximum independent set, Physical Review A110, 052435 (2024)

    2024

  14. [14]

    J. R. Finˇ zgar, A. Kerschbaumer, M. J. Schuetz, C. B. Mendl, and H. G. Katzgraber, Quantum-informed re- cursive optimization algorithms, PRX Quantum5, 020327 (2024)

    2024

  15. [15]

    L. T. Brady and S. Hadfield, Quantum DPLL and gen- eralized constraints in iterative quantum algorithms, arXiv preprint arXiv:2509.02689 (2025)

    Pith/arXiv arXiv 2025

  16. [16]

    Dupont and B

    M. Dupont and B. Sundar, Extending relax-and- round combinatorial optimization solvers with quantum correlations, Physical Review A109, 10.1103/phys- reva.109.012429 (2024)

    doi:10.1103/phys- 2024

  17. [17]

    Dupont, T

    M. Dupont, T. Oberoi, and B. Sundar, Optimization via Quantum Preconditioning, Physical Review Applied24, 10.1103/9prw-684p (2025)

    work page doi:10.1103/9prw-684p 2025

  18. [18]

    ˇCepait˙ e, N

    I. ˇCepait˙ e, N. Vaishnav, L. Zhou, and A. Montanaro, Quantum-enhanced optimization by warm starts, arXiv preprint arXiv:2508.16309 (2025)

    Pith/arXiv arXiv 2025

  19. [19]

    D. J. Egger, J. Mareˇ cek, and S. Woerner, Warm-starting quantum optimization, Quantum5, 479 (2021)

    2021

  20. [20]

    R. Tate, M. Farhadi, C. Herold, G. Mohler, and S. Gupta, Bridging classical and quantum with Sdp ini- tialized warm-starts for QAOA, ACM Transactions on Quantum Computing4, 1 (2023)

    2023

  21. [21]

    R. Tate, J. Moondra, B. Gard, G. Mohler, and S. Gupta, Warm-Started QAOA with Custom Mixers Prov- ably Converges and Computationally Beats Goemans- Williamson's Max-Cut at Low Circuit Depths, Quan- tum7, 1121 (2023)

    2023

  22. [22]

    M. Cain, E. Farhi, S. Gutmann, D. Ranard, and E. Tang, The qaoa gets stuck starting from a good clas- sical string, arXiv preprint arXiv:2207.05089 (2022)

    Pith/arXiv arXiv 2022

  23. [23]

    Augustino, M

    B. Augustino, M. Cain, E. Farhi, S. Gupta, S. Gutmann, D. Ranard, E. Tang, and K. Van Kirk, Strategies for running the qaoa at hundreds of qubits, arXiv preprint arXiv:2410.03015 (2024)

    Pith/arXiv arXiv 2024

  24. [24]

    K. N. Okada, H. Nishi, T. Kosugi, and Y.-i. Matsushita, Systematic study on the dependence of the warm-start quantum approximate optimization algorithm on ap- proximate solutions, Scientific Reports14, 1167 (2024)

    2024

  25. [25]

    Y. Chai, K. Jansen, S. K¨ uhn, T. Schw¨ agerl, and T. Stol- lenwerk, Warm start of variational quantum algorithms for quadratic unconstrained binary optimization prob- lems, EPJ Quantum Technology13, 9 (2025)

    2025

  26. [26]

    H. Yuan, C. K. Long, H. V. Lepage, and C. H. W. Barnes, Quantifying the advantages of applying quan- tum approximate algorithms to portfolio optimisation, Quantum Science and Technology11, 025034 (2026)

    2026

  27. [27]

    Dehn and T

    V. Dehn and T. Wellens, A hybrid quantum-classical approach to warm-starting optimization, inQuantum Computing, Communication, and Simulation IV, Vol. 12911 (SPIE, 2024) pp. 216–227

    2024

  28. [28]

    Graß, Quantum annealing with longitudinal bias fields, Phys

    T. Graß, Quantum annealing with longitudinal bias fields, Phys. Rev. Lett.123, 120501 (2019)

    2019

  29. [29]

    Y. Yu, C. Cao, C. Dewey, X.-B. Wang, N. Shannon, and R. Joynt, Quantum approximate optimization al- gorithm with adaptive bias fields, Physical Review Re- search4, 023249 (2022)

    2022

  30. [30]

    Y. Yu, C. Cao, X.-B. Wang, N. Shannon, and R. Joynt, Solution of sat problems with the adaptive-bias quan- tum approximate optimization algorithm, Physical Re- view Research5, 023147 (2023)

    2023

  31. [31]

    Yu, X.-B

    Y. Yu, X.-B. Wang, N. Shannon, and R. Joynt, Warm- start adaptive-bias quantum approximate optimization algorithm, Physical Review A112, 012422 (2025)

    2025

  32. [32]

    Feeney, R

    S. Feeney, R. Tate, and S. Eidenbenz, The better solu- tion probability metric: Optimizing qaoa to outperform its warm-start solution, in2025 International Confer- ence on Quantum Communications, Networking, and Computing (QCNC)(2025) pp. 450–458

    2025

  33. [33]

    Bhattacharyya, M

    B. Bhattacharyya, M. Capriotti, and R. Tate, Solving general qubos with warm-start qaoa via a reduction to max-cut, arXiv preprint arXiv:2504.06253 (2025)

    arXiv 2025

  34. [34]

    P. N. Ha Huy, V. H. Nguyen, and A. S. Ta, Difference of convex algorithm for warm-start quantum approximate optimization algorithm, Advanced Quantum Technolo- gies , 2400253 (2025)

    2025

  35. [35]

    H. Yuan, S. Yang, and C. H. Barnes, Iterative quantum optimisation with a warm-started quantum state, arXiv preprint arXiv:2502.09704 (2025)

    Pith/arXiv arXiv 2025

  36. [36]

    M. A. Lopez-Ruiz, E. L. Tucker, E. M. Arnold, E. Epifanovsky, A. Kaushik, and M. Roetteler, A non- variational quantum approach to the job shop schedul- ing problem, arXiv preprint arXiv:2510.26859 (2025)

    arXiv 2025

  37. [37]

    K. V. Marshall, D. J. Egger, M. Garn, F. Schiavello, S. Brandhofer, C. Zoufal, and S. Woerner, Quantum- enhanced markov chain monte carlo for combinatorial optimization, arXiv preprint arXiv:2602.06171 (2026)

    arXiv 2026

  38. [38]

    Bucher, M

    D. Bucher, M. Janetschek, M. Poppel, J. Stein, C. Linnhoff-Popien, and S. Feld, Constrained quantum 10 optimization via iterative warm-start xy-mixers, arXiv preprint arXiv:2604.02083 (2026)

    Pith/arXiv arXiv 2026

  39. [39]

    P. C. Lotshaw, T. Morris, S. Hadfield, and R. Bennink, Iterative warm-start optimization with quantum imag- inary time evolution, arXiv preprint arXiv:2604.26047 (2026)

    Pith/arXiv arXiv 2026

  40. [40]

    F. B. Maciejewski, J. Biamonte, S. Hadfield, and D. Venturelli, Improving quantum approximate opti- mization by noise-directed adaptive remapping, Quan- tum9, 1906 (2025)

    1906

  41. [41]

    F. B. Maciejewski, B. G. Bach, M. Dupont, P. A. Lott, B. Sundar, D. E. B. Neira, I. Safro, and D. Venturelli, A multilevel approach for solving large-scale qubo prob- lems with noisy hybrid quantum approximate optimiza- tion, in2024 IEEE High Performance Extreme Com- puting Conference (HPEC)(IEEE, 2024) pp. 1–10

    2024

  42. [42]

    W.-H. Tam, H. Matsuyama, R. Sakai, and Y. Ya- mashiro, Enhancing ndar with delay-gate-induced am- plitude damping, arXiv preprint arXiv:2504.12628 (2025)

    Pith/arXiv arXiv 2025

  43. [43]

    F. B. Maciejewski, S. Hadfield, B. Hall, M. Hodson, M. Dupont, B. Evert, J. Sud, M. S. Alam, Z. Wang, S. Jeffrey,et al., Design and execution of quantum cir- cuits using tens of superconducting qubits and thou- sands of gates for dense Ising optimization problems, Physical Review Applied22, 044074 (2024)

    2024

  44. [44]

    G. Li, Y. Ding, and Y. Xie, Tackling the qubit mapping problem for nisq-era quantum devices, inProceedings of the Twenty-Fourth International Conference on Archi- tectural Support for Programming Languages and Oper- ating Systems, ASPLOS ’19 (Association for Computing Machinery, New York, NY, USA, 2019) p. 1001–1014

    2019

  45. [45]

    H. Zou, M. Treinish, K. Hartman, A. Ivrii, and J. Lish- man, Lightsabre: A lightweight and enhanced sabre al- gorithm, arXiv preprint arXiv:2409.08368 (2024)

    Pith/arXiv arXiv 2024

  46. [46]

    Pelofske, A

    E. Pelofske, A. B¨ artschi, L. Cincio, J. Golden, and S. Ei- denbenz, Scaling whole-chip qaoa for higher-order ising spin glass models on heavy-hex graphs, npj Quantum Information10(2024)

    2024

  47. [47]

    Z. Wang, J. Mandell, Y. Xu, and J. Shi, A depth- independent linear chain ansatz for large-scale quantum approximate optimization (2025), arXiv:2509.17296 [quant-ph]

    arXiv 2025

  48. [48]

    Mohseni, J.-P

    N. Mohseni, J.-P. Houle, I. Shehzad, G. Cortiana, C. O’Meara, and A. B. Watts, Constrained quantum optimization at utility scale: Application to the knap- sack problem, arXiv preprint arXiv:2603.00260 (2026)

    arXiv 2026

  49. [49]

    J. A. Montanez-Barrera, K. Michielsen, and D. E. B. Neira, Evaluating the performance of quantum pro- cessing units at large width and depth, arXiv preprint arXiv:2502.06471 (2025)

    Pith/arXiv arXiv 2025

  50. [50]

    J. A. Monta˜ nez-Barrera and K. Michielsen, Toward a linear-ramp qaoa protocol: evidence of a scaling advan- tage in solving some combinatorial optimization prob- lems, npj Quantum Information11, 10.1038/s41534- 025-01082-1 (2025)

    doi:10.1038/s41534- 2025

  51. [51]

    A. B. Rava, K. Michielsen, and J. A. Montanez-Barrera, Benchmarking neutral atom-based quantum processors at scale, arXiv preprint arXiv:2511.22967 (2025)

    arXiv 2025

  52. [52]

    D. E. Bernal Neira, R. Brown, P. Sathe, F. Wudarski, M. Pavone, E. Rieffel, and D. Venturelli, Benchmarking the operation of quantum heuristics and ising machines: scoring parameter setting strategies on optimization ap- plications, Quantum Machine Intelligence7, 86 (2025)

    2025

  53. [53]

    Sherrington and S

    D. Sherrington and S. Kirkpatrick, Solvable model of a spin-glass, Physical review letters35, 1792 (1975)

    1975

  54. [54]

    Lucas, Ising formulations of many NP problems, Frontiers in physics2, 74887 (2014)

    A. Lucas, Ising formulations of many NP problems, Frontiers in physics2, 74887 (2014)

    2014

  55. [55]

    Hadfield, On the representation of Boolean and real functions as Hamiltonians for quantum computing, ACM Transactions on Quantum Computing2, 1 (2021)

    S. Hadfield, On the representation of Boolean and real functions as Hamiltonians for quantum computing, ACM Transactions on Quantum Computing2, 1 (2021)

    2021

  56. [56]

    Z. He, R. Shaydulin, S. Chakrabarti, D. Herman, C. Li, Y. Sun, and M. Pistoia, Alignment between initial state and mixer improves qaoa performance for constrained optimization, npj Quantum Information9, 121 (2023)

    2023

  57. [57]

    Burer, R

    S. Burer, R. D. Monteiro, and Y. Zhang, Rank-two relaxation heuristics for max-cut and other binary quadratic programs, SIAM Journal on Optimization12, 503 (2002)

    2002

  58. [58]

    Dunning, S

    I. Dunning, S. Gupta, and J. Silberholz, What works best when? a systematic evaluation of heuristics for max-cut and qubo, INFORMS Journal on Computing 30, 608 (2018)

    2018

  59. [59]

    F. B. Maciejewski, B. G. Bach, J. Biamonte, S. Hadfield, and D. Venturelli, quapopt – open source GitHub repository for quantum approxi- mate optimization,https://github.com/usra-riacs/ quantum-approximate-optimization(2025)

    2025

  60. [60]

    Koch, Private communication (2026)

    D. Koch, Private communication (2026)

    2026

  61. [61]

    M. X. Goemans and D. P. Williamson, Improved ap- proximation algorithms for maximum cut and satisfia- bility problems using semidefinite programming, Jour- nal of the ACM (JACM)42, 1115 (1995)

    1995

  62. [62]

    R. S. d. Carmo, M. Santana, F. F. Fanchini, V. H. C. de Albuquerque, and J. P. Papa, Warm-starting qaoa with xy mixers: A novel approach for quantum- enhanced vehicle routing optimization, arXiv preprint arXiv:2504.19934 (2025)

    Pith/arXiv arXiv 2025

  63. [63]

    R. S. d. Carmo, R. G. d. Reis, S. F. F. Silva, L. G. E. Ar- ruda, and F. F. Fanchini, Warm-starting PCE for travel- ing salesman problem, arXiv preprint arXiv:2509.14414 (2025)

    arXiv 2025

  64. [64]

    Truger, J

    F. Truger, J. Barzen, M. Bechtold, M. Beisel, F. Ley- mann, A. Mandl, and V. Yussupov, Warm-starting and quantum computing: A systematic mapping study, ACM Computing Surveys56, 1 (2024)

    2024

  65. [65]

    Truger, J

    F. Truger, J. Barzen, M. Beisel, F. Leymann, and V. Yussupov, Warm-Starting Patterns for Quantum Al- gorithms, inProceedings of the 16 th International Con- ference on Pervasive Patterns and Applications (PAT- TERNS 2024)(Xpert Publishing Services (XPS), 2024) pp. 25–31

    2024

  66. [66]

    Developers, Reverse quantum annealing for local refinement of solutions, D-Wave Systems Inc., Tech

    D.-W. Developers, Reverse quantum annealing for local refinement of solutions, D-Wave Systems Inc., Tech. Rep (2021)

    2021

  67. [67]

    Ohkuwa, H

    M. Ohkuwa, H. Nishimori, and D. A. Lidar, Reverse annealing for the fully connected p-spin model, Physical Review A98, 022314 (2018)

    2018

  68. [68]

    Kechedzhi, V

    K. Kechedzhi, V. Smelyanskiy, J. R. McClean, V. S. Denchev, M. Mohseni, S. Isakov, S. Boixo, B. Altshuler, and H. Neven, Efficient Population Transfer via Non- Ergodic Extended States in Quantum Spin Glass, Leib- niz Int. Proc. Inf.111, 9:1 (2018)

    2018

  69. [69]

    Marshall, D

    J. Marshall, D. Venturelli, I. Hen, and E. G. Rieffel, Power of pausing: Advancing understanding of ther- malization in experimental quantum annealers, Physical Review Applied11, 044083 (2019). 11

    2019

  70. [70]

    Venturelli and A

    D. Venturelli and A. Kondratyev, Reverse quantum an- nealing approach to portfolio optimization problems, Quantum Machine Intelligence1, 17 (2019)

    2019

  71. [71]

    Mehta, H

    V. Mehta, H. De Raedt, K. Michielsen, and F. Jin, Un- raveling reverse annealing: A study of d-wave quantum annealers, Physical Review A112, 012414 (2025)

    2025

  72. [72]

    Barton, J

    B. Barton, J. Sagal, S. Feeney, G. Grattan, P. Patnaik, V. Oganesyan, L. D. Carr, and E. Kapit, Iterative quan- tum optimization of spin glass problems with rapidly oscillating transverse fields, Quantum Science and Tech- nology10, 045063 (2025)

    2025

  73. [73]

    T. M. Ragonneau, Model-based derivative-free op- timization methods and software, arXiv preprint arXiv:2210.12018 (2023)

    Pith/arXiv arXiv 2023

  74. [74]

    D. J. Wales and J. P. Doye, Global optimization by basin-hopping and the lowest energy structures of lennard-jones clusters containing up to 110 atoms, The Journal of Physical Chemistry A101, 5111 (1997)

    1997

  75. [75]

    Javadi-Abhari, M

    A. Javadi-Abhari, M. Treinish, K. Krsulich, C. J. Wood, J. Lishman, J. Gacon, S. Martiel, P. D. Nation, L. S. Bishop, A. W. Cross, B. R. Johnson, and J. M. Gam- betta, Quantum computing with Qiskit, arXiv preprint arXiv:2405.08810 (2024)

    Pith/arXiv arXiv 2024

  76. [76]

    As explained in the text, 0th iteration corresponds to stan- dard (Time-Block) QAOA withc= 0.5

    Please note that we label iterations starting from 0. As explained in the text, 0th iteration corresponds to stan- dard (Time-Block) QAOA withc= 0.5. Each consec- utive iteration is implemented with 2 distinct values of c, hence the total number of samples generated until iterationris 10 4 (2r+ 1)

  77. [77]

    Matsuo, S

    A. Matsuo, S. Yamashita, and D. J. Egger, A SAT approach to the initial mapping problem in SWAP gate insertion for commuting gates, IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences , 2022EAP1159 (2023)

    2023

  78. [78]

    Kotil, F

    A. Kotil, F. Simkovic, and M. Leib, Improved qubit routing for qaoa circuits, arXiv preprint arXiv:2312.15982 (2023)

    Pith/arXiv arXiv 2023

  79. [79]

    Y. Quek, D. Stilck Fran¸ ca, S. Khatri, J. J. Meyer, and J. Eisert, Exponentially tighter bounds on limita- tions of quantum error mitigation, Nature Physics20, 1648–1658 (2024)

    2024

  80. [80]

    Viola and S

    L. Viola and S. Lloyd, Dynamical suppression of deco- herence in two-state quantum systems, Physical Review A58, 2733 (1998)

    1998

Showing first 80 references.