Pith. sign in

REVIEW 2 major objections 7 minor 2 cited by

Evaluating the Performance of Direct Higher-Order Formulations in Combinatorial Optimization Problems

T0 review · 2 major / 7 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read Directly solving higher-order optimization problems in PUBO form outperforms reducing them to QUBO on the same annealing hardware.

desk verdict Useful benchmark evidence for PUBO vs QUBO on a commercial SA engine, but the fixed p=1 penalty undercuts the claim of an intrinsic advantage. read the letter →

arxiv 2510.24237 v1 pith:BEPO7BX3 submitted 2025-10-28 cond-mat.stat-mech

Kazuki Ikeuchi , Yoshiki Matsuda , Shu Tanaka This is my paper
classification cond-mat.stat-mech
keywords PUBOQUBOorderreductionIsingmachinessimulatedannealinglowautocorrelationbinarysequencevehicleroutingproblemmulti-objectiveoptimization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tests whether the standard step of converting higher-order optimization problems into quadratic QUBO form—required by most Ising-type hardware—actually degrades solution quality. Using a simulated-annealing engine that can natively handle up to fourth-order terms, the authors compare a PUBO solver against a QUBO solver that first applies the substitution method for order reduction. On the low autocorrelation binary sequence (LABS) problem for lengths N=5 to 99, the PUBO solver consistently reaches lower normalized energies with smaller run-to-run variance, and the gap widens as N grows. On a vehicle routing problem with distance balancing, where the variance objective is fourth-order, PUBO achieves larger hypervolume on all 30 random instances, meaning its Pareto fronts are broader and better. The paper attributes this to the auxiliary variables and penalty terms that order reduction introduces, which inflate the search space and depend on a penalty coefficient with no single optimal value across problem sizes.

What carries the argument

The substitution method for order reduction, illustrated by the replacement x1x2x3 = y12x3 + p[x1x2 - 2y12(x1+x2) + 3y12], introduces an auxiliary binary variable y12 and a penalty whose strength is set by p. The paper contrasts this with a PUBO solver that evaluates up to fourth-order terms directly. The mechanism driving the results is the combinatorial growth of auxiliary variables and penalty sensitivity: as problem size increases, the QUBO search space expands far beyond PUBO's, and no fixed p works well across all N, as shown in the penalty-sweep heatmap.

What would settle it

Re-run the LABS comparison for representative lengths (e.g., N=41, 67, 97) with a QUBO solver whose penalty p is tuned per N by grid search over 1–10, and show that PUBO no longer dominates in average normalized energy; or, for the VRP, include all infeasible solutions in the hypervolume computation and show the PUBO advantage disappears.

Watch

Extended reading notes

Core claim

On a GPU-parallelized simulated annealing platform that can evaluate polynomial objective functions up to fourth order, preserving higher-order terms yields consistently better solutions than first reducing the problem to QUBO. For LABS (N=5–99), the average normalized energy E/E*_N is lower and more stable for PUBO, with the performance gap increasing with problem size. For the distance-balancing VRP (P=11, V=3), all 30 generated instances give larger hypervolume for PUBO, indicating more diverse and faithful Pareto fronts between total travel distance and distance variance. The QUBO route requires far more variables (up to tens of times more for larger LABS instances) and its solution qual

Load-bearing premise

The central claim rests on treating the fixed default penalty p=1 as a fair representative of QUBO order reduction; since the paper's own sweep shows the optimal p varies with problem size, a tuned QUBO baseline could close much of the observed gap.

Editorial extensions

If this is right

  • Practitioners targeting annealing hardware can skip order-reduction compilation; a direct PUBO formulation is both simpler and yields better solutions in the tested regimes.
  • Hardware and solver developers should consider native support for higher-order terms, since the quality loss from order reduction grows with problem size rather than being a constant offset.
  • For multi-objective problems with variance-type objectives, QUBO may systematically fail to reach the true Pareto front, which matters for fairness-related objectives in logistics.
  • The penalty coefficient of order reduction should be treated as a per-instance hyperparameter; benchmarks with fixed defaults can misrepresent the QUBO baseline.
  • The PUBO advantage is achieved without sacrificing computation time, since both solvers ran under identical 60-second limits.

Reading between the lines

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

  • The claimed 'intrinsic' advantage of PUBO may shrink against a QUBO baseline with carefully tuned penalty coefficients; the paper's own sweep shows optimal p varies with N, so a tuned QUBO is a fairer and possibly stronger opponent.
  • The VRP comparison omits infeasible solutions before computing hypervolume; if QUBO produces many infeasible solutions, discarding them is lenient toward QUBO, and reporting feasibility rates could change the interpretation.
  • The variable-explosion mechanism suggests a testable prediction: problems with dense higher-order structure (many interacting triples and quartets) should show a larger PUBO advantage than problems with sparse higher-order terms.
  • Although the experiments use simulated annealing, the structural argument—penalty sensitivity plus auxiliary-variable growth—likely carries over to any solver that requires QUBO input, including gate-based algorithms that would otherwise need extra qubits.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 7 minor

Summary. This paper reports an empirical comparison between directly solving higher-order polynomial unconstrained binary optimization (PUBO) formulations and first reducing them to QUBO via the substitution method. The experiments use Fixstars Amplify AE on the same platform for two benchmarks: the low autocorrelation binary sequence (LABS) problem for N=5..99, and a distance-balancing vehicle routing problem (VRP) with P=11 customers and V=3 vehicles. The paper reports that the PUBO solver gives lower normalized LABS energies with smaller variance across trials, larger hypervolume on all 30 VRP instances, fewer variables, and no need for order-reduction compilation. The QUBO baseline is built with the substitution penalty fixed to the default value p=1, although Section V-B later shows that the optimal p varies with N and requires problem-specific tuning.

Significance. The problem choice is apt: LABS and distance-balancing VRP both have natural higher-order interactions, and comparing formulations on one hardware/software platform is a reasonable way to isolate the effect of order reduction. If the main claim survived a fairer baseline, the paper would provide useful evidence for the practical development of native PUBO solvers. The concrete documentation of variable-count growth in Fig. 7 is also useful. However, the present evidence is not yet sufficient to support the abstract's 'consistently superior' and 'intrinsic advantage' claims: the QUBO side is a single reduction method with an untuned penalty parameter, and the VRP comparison omits feasibility-rate information. In its current form, the paper is best read as a default-parameter comparison on one commercial solver, not as a general statement about PUBO versus QUBO.

major comments (2)
  1. [§IV-A, §V-B, Fig. 8] The main comparison fixes the order-reduction penalty to p=1 for every problem size (Section IV-A), but Section V-B/Fig. 8 shows that for LABS the best p varies strongly with N and no single p is optimal across N. This is load-bearing: the reduced QUBO objective equals the original objective only when the substitution constraints y=x_i x_j are satisfied, and the manuscript itself notes in Section II-C that solution quality is sensitive to p. With p=1 the QUBO baseline may be mistuned, particularly for large N where Fig. 7 shows a very large number of auxiliary variables. The same fixed p=1 is used for the VRP experiments, whose higher-order structure is different. To support the headline claim, please either sweep p (and report substitution-constraint satisfaction rates), or restrict the abstract and conclusions to a default-parameter comparison. As written, the claim of an 'intrinsic ad
  2. [§IV-C1, Fig. 5, Fig. 6] The VRP analysis plots and compares only feasible solutions (Fig. 5 caption), but the paper never reports how many solutions are infeasible for each formulation. If PUBO and QUBO return different feasibility rates, filtering changes the comparison and can hide important differences in constraint-handling quality. The penalty coefficient mu is also fixed by Eq. (24) without a sensitivity study. Please report feasibility counts/rates for both formulations, and ideally present the hypervolume comparison both with and without the feasibility filter, plus at least a small sweep over mu.
minor comments (7)
  1. [Eq. (17)] The index set in (17) uses '0≤i≤N', but the VRP section defines P customer locations; N is not defined in that section. This is likely a typo for P.
  2. [Eq. (23)] The formula for S is typeset ambiguously. Please write S = floor(P/(V-1)) explicitly.
  3. [Abstract, §IV-A] Since the runtime is fixed at 60 seconds for both solvers, 'comparable computational time' is true by construction. Describe this as a fixed-budget comparison rather than a finding.
  4. [§IV-B, Eq. (22)] The source of E*_N, the best-known LABS energies, is not given. For larger N these values should be cited or explicitly described as best-known rather than proven optimum.
  5. [§IV-C2] The hypervolume reference point is described as 'maximum observed value of each objective function plus a margin of 0.1'. Please clarify whether the maximum is taken over the union of both solvers' solution sets or separately for each solver; the current wording is ambiguous.
  6. [§IV-B, §IV-C2] No statistical tests are reported. The error bars in Fig. 3 and the 30/30 points below y=x in Fig. 6 are descriptive; a paired nonparametric test (e.g., Wilcoxon signed-rank for the hypervolume comparison) would be a simple and appropriate addition.
  7. [Data/code availability] Code and data are not provided. For an empirical comparison of this type, making the instances and solver settings available would substantially improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the PUBO-vs-QUBO comparison is an empirical benchmark with externally defined problems; the p=1 baseline caveat is a validity concern, not a circular derivation.

full rationale

The paper contains no derivation chain in which an output is defined in terms of an input or a fitted parameter is renamed as a prediction. LABS and VRP are defined independently from their problem statements (Eqs. 11–21), and the normalized LABS metric (Eq. 14) uses external best-known energies E*_N. The QUBO baseline is generated by the standard substitution formula Eq. (9), and the paper does not claim Eq. (9) is an identity for all p; it explicitly states that solution accuracy is sensitive to p. The main comparison is a direct empirical measurement on the same solver platform, not a mathematical consequence of the formulations. The principal caveat—Sec. IV-A fixes p=1 for all main runs, while Sec. V-B/Fig. 8 shows no single p is best—could weaken the claim of an 'intrinsic advantage' if a tuned QUBO baseline performs better, but this is a baseline-tuning/confound issue, not circularity: the paper never fits a parameter to the target result and then reports the fit as a prediction. There are self-citations in the reference list, but none is load-bearing for the central empirical claim, and the problems are benchmarked against external known optima. The Fixstars affiliation of the second author is an independence/COI consideration, not a form of derivational circularity under the requested taxonomy. Accordingly, no circular steps are identified; score 0 is the honest finding.

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

The central claim rests on an empirical comparison rather than a derivation; the main free parameters are penalty coefficients and hyperparameters. No new physical entities are introduced. The load-bearing assumptions concern the fairness of the QUBO baseline and the claim that both solvers share an algorithmic core.

free parameters (5)
  • penalty coefficient p for order reduction = 1 (default; fixed in main experiments; swept 1-10 in Sec V-B)
    Substitution-method penalty in Eq. (9); main comparison fixes p=1 although Fig. 8 shows the optimal p varies with N, so the QUBO baseline is not tuned.
  • VRP constraint penalty coefficient mu = max d_ij + (max d_ij / V)^2 (Eq. 24)
    Chosen by a heuristic formula, not tuned; controls feasibility of solutions and affects all VRP results.
  • hyperparameter alpha (variance weight) = swept 0 to 1 in 0.01 increments
    Weights f1 and f2 in Eq. (21); swept rather than fitted, but choice shapes the Pareto front and hypervolume comparison.
  • hypervolume reference point margin = 0.1
    Reference point = max observed objective + 0.1; chosen by hand, same for both solvers, but affects hypervolume magnitudes.
  • number of travel steps S = ceil(P/(V-1)) (Eq. 23)
    A modeling choice that fixes the encoding size of the VRP and affects the search space.
assumptions (5)
  • standard math The substitution penalty AND(x1,x2,y) in Eq. (10) enforces y=x1*x2 for sufficiently large p in Eq. (9).
    Standard order-reduction identity, verified by Table 1; central to the QUBO construction.
  • domain assumption Amplify AE's PUBO and QUBO solvers share the same simulated-annealing core and hardware, so observed differences are due to formulation/order reduction.
    Stated in Secs. II-B and IV-A; if the two solver implementations differ internally, the attribution to order reduction fails.
  • domain assumption Setting p=1 in the main comparison is a representative QUBO configuration.
    Adopted as the Amplify default; Sec. V-B shows this is not optimal across N, so the assumption is questionable.
  • domain assumption Best-known LABS energies E*_N used in Eq. (14) are correct for N=5..99.
    Normalized energy depends on E*_N; the paper does not list or cite per-N values, so this input is unverified.
  • domain assumption Filtering VRP results to feasible solutions does not bias the comparison.
    Infeasible QUBO solutions are omitted without reporting rates (Sec. IV-C1); if rates differ, hypervolume comparison is confounded with constraint violation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Evaluating the Performance of Direct Higher-Order Formulations in Combinatorial Optimization Problems." pith.science (2026). https://pith.science/paper/BEPO7BX3

@misc{pith2026251024237,
  author       = {Pith},
  title        = {Pith review of: Evaluating the Performance of Direct Higher-Order Formulations in Combinatorial Optimization Problems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BEPO7BX3}},
  note         = {Machine review of arXiv:2510.24237}
}
read the original abstract

Ising machines, including quantum annealing machines, are promising next-generation computers for combinatorial optimization problems. However, due to hardware limitations, most Ising-type hardware can only solve objective functions expressed in linear or quadratic terms of binary variables. Therefore, problems with higher-order terms require an order-reduction process, which increases the number of variables and constraints and may degrade solution quality. In this study, we evaluate the effectiveness of directly solving such problems without order reduction by using a high-performance simulated annealing-based optimization solver capable of handling polynomial unconstrained binary optimization (PUBO) formulations. We compare its performance against a conventional quadratic unconstrained binary optimization (QUBO) solver on the same hardware platform. As benchmarks, we use the low autocorrelation binary sequence (LABS) problem and the vehicle routing problem with distance balancing, both of which naturally include higher-order interactions. Results show that the PUBO solver consistently achieves superior solution quality and stability compared to its QUBO counterpart, while maintaining comparable computational time and requiring no order-reduction compilation indicating potential advantages of directly handling higher-order terms in practical optimization problems.

Figures

Figures reproduced from arXiv: 2510.24237 by the authors.

Figure 1
Figure 1. FIGURE 1 [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. FIGURE 2 [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. shows the average normalized energy E˜ obtained from ten independent trials for each sequence length N. The shaded area represents the standard deviation over the trials. The results show that the QUBO solver exhibited degraded solution accuracy even for relatively small N, accompanied by large variations in the obtained results. In contrast, the PUBO solver consistently achieved lower average E˜ values with smaller… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: shows the relationships among α, the total travel distance, the variance of travel distances, and the objective function values. The solid lines represent the averages of ten independent trials, and the surrounding shaded regions indi￾cate the standard deviations. As s…
Figure 5
Figure 5. Figure 5: shows a scatter plot of all feasible solutions obtained over all values of α and all trials, where the horizontal axis represents the total travel distance and the vertical axis represents the variance of travel distances. In the scatter plot, the solutions obtained by…
Figure 7
Figure 7. Figure 7: illustrates the increase in the total number of vari￾ables as a function of problem size [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: shows a heatmap where the horizontal axis rep￾resents the penalty coefficient used for order reduction, the vertical axis represents the sequence length N, and the color indicates the ranking of solutions obtained under each con￾dition. Brighter colors correspond to sm…

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Structural Comparison of Error Mitigation Methods for Ising Machines: Penalty-Spin Model versus Stacked Model

    cond-mat.stat-mech 2026-01 conditional novelty 6.0 of 10

    On sparse assignment problems, ferromagnetically coupled stacked replicas of an Ising model outperform the centralized penalty-spin design, which loses solution structure when many replicas are averaged in its auxilia...

  2. Simulated Annealing for Quadratic and Higher-Order Unconstrained Integer Optimization

    cond-mat.stat-mech 2025-11 conditional novelty 6.0 of 10

    Direct simulated annealing on QUIO and HUIO problems achieves higher efficiency and better solution quality than QUBO conversions.

Reference graph

Works this paper leans on

65 extracted references · 5 linked inside Pith · cited by 2 Pith papers

  1. [1]

    Traffic flow optimization using a quantum annealer,

    F. Neukart, G. Compostella, C. Seidel, D. von Dollen, S. Yarkoni, and B. Parney, “Traffic flow optimization using a quantum annealer,” Front. ICT, vol. 4, p. 29, 2017

    2017

  2. [2]

    An approach to the vehicle routing problem with balanced pick-up using Ising machines,

    S. Bao, M. Tawada, S. Tanaka, and N. Togawa, “An approach to the vehicle routing problem with balanced pick-up using Ising machines,” in 2021 International Symposium on VLSI Design, Automation and Test (VLSI- DAT), 2021, pp. 1–4

    2021

  3. [3]

    Annealing-assisted column generation for inequality-constrained combinatorial optimization problems,

    H. Kanai, M. Yamashita, K. Tanahashi, and S. Tanaka, “Annealing-assisted column generation for inequality-constrained combinatorial optimization problems,” IEEE Access, vol. 12, pp. 157 669–157 685, 2024

    2024

  4. [4]

    Designing metamaterials with quantum annealing and factorization ma- chines,

    K. Kitai, J. Guo, S. Ju, S. Tanaka, K. Tsuda, J. Shiomi, and R. Tamura, “Designing metamaterials with quantum annealing and factorization ma- chines,” Phys. Rev. Res., vol. 2, no. 1, p. 013319, 2020

    2020

  5. [5]

    Towards optimization of photonic-crystal surface-emitting lasers via quantum annealing,

    T. Inoue, Y . Seki, S. Tanaka, N. Togawa, K. Ishizaki, and S. Noda, “Towards optimization of photonic-crystal surface-emitting lasers via quantum annealing,” Opt. Express, vol. 30, no. 24, pp. 43 503–43 512, 2022

    2022

  6. [6]

    Hashiguchi, A

    K. Hashiguchi, A. Maruo, S. Iwane, H. Jippo, and Y . Suga, “Material discovery using quantum-inspired technologies: validating a band gap 10 VOLUME 4, 2016 Kazuki Ikeuchiet al.: Evaluating the Performance of Direct Higher-Order Formulations in Combinatorial Optimization Problems engineering approach for inorganic materials,” Bull. Chem. Soc. Jpn., vol. 98...

    2016

  7. [7]

    Solving the optimal trading trajectory problem using a quantum annealer,

    G. Rosenberg, P. Haghnegahdar, P. Goddard, P. Carr, K. Wu, and M. L. de Prado, “Solving the optimal trading trajectory problem using a quantum annealer,” IEEE J. Sel. Top. Signal Processing, vol. 10, no. 6, pp. 1053– 1060, 2016

    2016

  8. [8]

    Application of Ising machines and a software development for Ising machines,

    K. Tanahashi, S. Takayanagi, T. Motohashi, and S. Tanaka, “Application of Ising machines and a software development for Ising machines,” J. Phys. Soc. Jpn., vol. 88, no. 6, p. 061010, 2019

    2019

Show all 65 references
  1. [9]

    Correlation- diversified portfolio construction by finding maximum independent set in large-scale market graph,

    R. Hidaka, Y . Hamakawa, J. Nakayama, and K. Tatsumura, “Correlation- diversified portfolio construction by finding maximum independent set in large-scale market graph,” IEEE Access, vol. 11, pp. 142 979–142 991, 2023

    2023

  2. [10]

    Real-time trading system based on selections of potentially profitable, uncorrelated, and balanced stocks by np-hard combinatorial optimization,

    K. Tatsumura, R. Hidaka, J. Nakayama, T. Kashimata, and M. Yamasaki, “Real-time trading system based on selections of potentially profitable, uncorrelated, and balanced stocks by np-hard combinatorial optimization,” IEEE Access, vol. 11, pp. 120 023–120 033, 2023

    2023

  3. [11]

    Quantum annealing in the transverse Ising model,

    T. Kadowaki and H. Nishimori, “Quantum annealing in the transverse Ising model,” Phys. Rev. E, vol. 58, no. 5, p. 5355, 1998

    1998

  4. [12]

    Tanaka, R

    S. Tanaka, R. Tamura, and B. K. Chakrabarti, Quantum spin glasses, annealing and computation. Cambridge University Press, 2017

    2017

  5. [13]

    Quantum annealing and computation: challenges and perspectives,

    B. K. Chakrabarti, H. Leschke, P. Ray, T. Shirai, and S. Tanaka, “Quantum annealing and computation: challenges and perspectives,” Phil. Trans. R. Soc. A, vol. 381, no. 2241, p. 20210419, 2023

    2023

  6. [14]

    How to reduce the bit- width of an Ising model by adding auxiliary spins,

    D. Oku, M. Tawada, S. Tanaka, and N. Togawa, “How to reduce the bit- width of an Ising model by adding auxiliary spins,” IEEE Trans. Comput., vol. 71, no. 1, pp. 223–234, 2020

    2020

  7. [15]

    Ising machines as hardware solvers of combinatorial optimization problems,

    N. Mohseni, P. L. McMahon, and T. Byrnes, “Ising machines as hardware solvers of combinatorial optimization problems,” Nat. Rev. Phys., vol. 4, no. 6, pp. 363–379, 2022

    2022

  8. [16]

    Effectiveness of hybrid op- timization method for quantum annealing machines,

    S. Kikuchi, N. Togawa, and S. Tanaka, “Effectiveness of hybrid op- timization method for quantum annealing machines,” arXiv preprint arXiv:2507.15544, 2025

    2025 arXiv

  9. [17]

    A hybrid solution method for the capacitated vehicle routing problem using a quantum annealer,

    S. Feld, C. Roch, T. Gabor, C. Seidel, F. Neukart, I. Galter, W. Mauerer, and C. Linnhoff-Popien, “A hybrid solution method for the capacitated vehicle routing problem using a quantum annealer,” Front. ICT, vol. 6, p. 13, 2019

    2019

  10. [18]

    Implementation of a hybrid classical-quantum annealing algorithm for logistic network design,

    Y . Ding, X. Chen, L. Lamata, E. Solano, and M. Sanz, “Implementation of a hybrid classical-quantum annealing algorithm for logistic network design,” SN Comput. Sci., vol. 2, no. 2, p. 68, 2021

    2021

  11. [19]

    Hybrid optimization method using simulated-annealing-based ising machine and quantum annealer,

    S. Kikuchi, N. Togawa, and S. Tanaka, “Hybrid optimization method using simulated-annealing-based ising machine and quantum annealer,” J. Phys. Soc. Jpn., vol. 92, no. 12, p. 124002, 2023

    2023

  12. [20]

    Black-box optimization using factorization and ising machines,

    R. Tamura, Y . Seki, Y . Minamoto, K. Kitai, Y . Matsuda, S. Tanaka, and K. Tsuda, “Black-box optimization using factorization and ising machines,” arXiv preprint arXiv:2507.18003, 2025

    2025 arXiv

  13. [21]

    Hybrid quantum annealing via molecular dynamics,

    H. Irie, H. Liang, T. Doi, S. Gongyo, and T. Hatsuda, “Hybrid quantum annealing via molecular dynamics,” Sci. Rep., vol. 11, no. 1, p. 8426, 2021

    2021

  14. [22]

    Spin-variable reduction method for handling linear equality constraints in ising machines,

    T. Shirai and N. Togawa, “Spin-variable reduction method for handling linear equality constraints in ising machines,” IEEE Trans. Comput., vol. 72, no. 8, pp. 2151–2164, 2023

    2023

  15. [23]

    Extending sample persistence variable reduction for constrained combinatorial optimization problems,

    S. Ide, S. Kikuchi, and S. Tanaka, “Extending sample persistence variable reduction for constrained combinatorial optimization problems,” arXiv preprint arXiv:2509.19280, 2025

    2025

  16. [24]

    Performance comparison of typical binary-integer encodings in an ising machine,

    K. Tamura, T. Shirai, H. Katsura, S. Tanaka, and N. Togawa, “Performance comparison of typical binary-integer encodings in an ising machine,” IEEE Access, vol. 9, pp. 81 032–81 039, 2021

    2021

  17. [25]

    Guiding principle for minor- embedding in simulated-annealing-based ising machines,

    T. Shirai, S. Tanaka, and N. Togawa, “Guiding principle for minor- embedding in simulated-annealing-based ising machines,” IEEE Access, vol. 8, pp. 210 490–210 502, 2020

    2020

  18. [26]

    Multi- objective qubo solver: Bi-objective quadratic assignment problem,

    M. Ayodele, R. Allmendinger, M. López-Ibáñez, and M. Parizy, “Multi- objective qubo solver: Bi-objective quadratic assignment problem,” in Proceedings of the Genetic and Evolutionary Computation Conference, 2022, pp. 467–475

    2022

  19. [27]

    Dynamical process of a bit-width reduced ising model with simulated annealing,

    S. Kikuchi, N. Togawa, and S. Tanaka, “Dynamical process of a bit-width reduced ising model with simulated annealing,” IEEE Access, vol. 11, pp. 95 493–95 506, 2023

    2023

  20. [28]

    Impact of fixing spins in a quantum annealer with energy rescaling,

    T. Hattori, H. Irie, T. Kadowaki, and S. Tanaka, “Impact of fixing spins in a quantum annealer with energy rescaling,” J. Phys. Soc. Jpn., vol. 94, no. 7, p. 074001, 2025

    2025

  21. [29]

    Quantum annealing for industry applications: Introduction and review,

    S. Yarkoni, E. Raponi, T. Bäck, and S. Schmitt, “Quantum annealing for industry applications: Introduction and review,” Rep. Prog. Phys., vol. 85, no. 10, p. 104001, 2022

    2022

  22. [30]

    Implementation of quantum annealing: A systematic review,

    L. P. Yulianti and K. Surendro, “Implementation of quantum annealing: A systematic review,” IEEE Access, vol. 10, pp. 73 156–73 177, 2022

    2022

  23. [31]

    Perspectives of quantum annealing: methods and implementations,

    P. Hauke, H. G. Katzgraber, W. Lechner, H. Nishimori, and W. D. Oliver, “Perspectives of quantum annealing: methods and implementations,” Rep. Prog. Phys., vol. 83, no. 5, p. 054401, 2020

    2020

  24. [32]

    Polynomial reduction methods and their impact on qaoa circuits,

    L. Schmidbauer, K. Wintersperger, E. Lobe, and W. Mauerer, “Polynomial reduction methods and their impact on qaoa circuits,” arXiv preprint arXiv:2406.08889, 2024

    2024 arXiv

  25. [33]

    Nonperturbative k-body to two-body commuting con- version hamiltonians and embedding problem instances into ising spins,

    J. D. Biamonte, “Nonperturbative k-body to two-body commuting con- version hamiltonians and embedding problem instances into ising spins,” Phys. Rev. A, vol. 77, p. 069901, 2008

    2008

  26. [34]

    Resource efficient gadgets for compiling adiabatic quantum optimization problems,

    R. Babbush, B. O. Gorman, and A. Aspuru-Guzik, “Resource efficient gadgets for compiling adiabatic quantum optimization problems,” Ann. Phys., vol. 525, no. 10-11, pp. 877–888, 2013

    2013

  27. [35]

    Polymer physics by quantum computing,

    C. Micheletti, P. Hauke, and P. Faccioli, “Polymer physics by quantum computing,” Phys. Rev. Lett., vol. 127, p. 080501, 2021

    2021

  28. [36]

    Quantum-inspired encoding enhances stochastic sampling of soft matter systems,

    F. Slongo, P. Hauke, P. Faccioli, and C. Micheletti, “Quantum-inspired encoding enhances stochastic sampling of soft matter systems,” Sci. Adv., vol. 9, no. 43, 2023

    2023

  29. [37]

    Investigating the potential for a limited quantum speedup on protein lattice problems,

    C. Outeiral, G. M. Morris, J. Shi, M. Strahm, S. C. Benjamin, and C. M. Deane, “Investigating the potential for a limited quantum speedup on protein lattice problems,” New J. Phys., vol. 23, p. 103030, 2021

    2021

  30. [38]

    Poly- nomial unconstrained binary optimisation inspired by optical simulation,

    D. A. Chermoshentsev, A. O. Malyshev, M. Esencan, E. S. Tiunov, D. Mendoza, A. Aspuru-Guzik, A. K. Fedorov, and A. I. Lvovsky, “Poly- nomial unconstrained binary optimisation inspired by optical simulation,” arXiv preprint arXiv:2106.13167, 2022

    2022 arXiv

  31. [39]

    Optimizing adsorption config- urations on alloy surfaces using tensor train optimizer,

    T. M. Do, T. Shiota, and W. Mizukami, “Optimizing adsorption config- urations on alloy surfaces using tensor train optimizer,” arXiv preprint arXiv:2507.20827, 2025

    2025 arXiv

  32. [40]

    Fixstars Amplify,

    Fixstars Corporation, “Fixstars Amplify,” https://amplify.fixstars.com/en/

  33. [41]

    Boosting quantum annealing performance through direct polynomial unconstrained binary optimization,

    S. Nagies, K. T. Geier, J. Akram, D. Bantounas, M. Johanning, and P. Hauke, “Boosting quantum annealing performance through direct polynomial unconstrained binary optimization,” Quantum Sci. Technol., vol. 10, no. 3, p. 035008, 2025

    2025

  34. [42]

    Evidence that pubo outperforms qubo when solving con- tinuous optimization problems with the qaoa,

    J. Stein, F. Chamanian, M. Zorn, J. Nüßlein, S. Zielinski, M. Kölle, and C. L. Popien, “Evidence that pubo outperforms qubo when solving con- tinuous optimization problems with the qaoa,” GECCO ’23 Companion: Proceedings of the Companion Conference on Genetic and Evolutionary ...

    2023

  35. [43]

    Simulated bifurcation for higher-order cost functions,

    T. Kanao and H. Goto, “Simulated bifurcation for higher-order cost functions,” Appl. Phys. Express, vol. 16, p. 014501, 2023

    2023

  36. [44]

    Optimization by simulated annealing,

    S. Kirkpatrick, C. D. Gelatt Jr, and M. P. Vecchi, “Optimization by simulated annealing,” Science, vol. 220, no. 4598, pp. 671–680, 1983

    1983

  37. [45]

    Transformation of general binary mrf minimization to the first-order case,

    H. Ishikawa, “Transformation of general binary mrf minimization to the first-order case,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 33, no. 6, pp. 1234–1249, 2010

    2010

  38. [46]

    What energy functions can be minimized via graph cuts?

    V . Kolmogorov and R. Zabih, “What energy functions can be minimized via graph cuts?” IEEE Trans. Pattern Anal. Mach. Intell., vol. 26, no. 2, pp. 147–159, 2004

    2004

  39. [47]

    Fast approximate energy minimization with label costs,

    A. Delong, A. Osokin, H. N. Isack, and Y . Boykov, “Fast approximate energy minimization with label costs,” Int. J. Comput. Vision, vol. 96, pp. 1–27, 2012

    2012

  40. [48]

    Energy minimization via graph cuts: Settling what is possible,

    D. Freedman and P. Drineas, “Energy minimization via graph cuts: Settling what is possible,” in 2005 IEEE Computer Society Conference on Com- puter Vision and Pattern Recognition (CVPR’05), vol. 2. IEEE, 2005, pp. 939–946

    2005

  41. [49]

    Sieves for low autocorrelation binary sequences,

    M. J. E. Golay, “Sieves for low autocorrelation binary sequences,” IEEE Trans. Inf. Theory, vol. 23, pp. 43–51, 1977

    1977

  42. [50]

    Synthesis of low-peak-factor signals and binary sequences with low autocorrelation,

    M. R. Schroeder, “Synthesis of low-peak-factor signals and binary sequences with low autocorrelation,” IEEE Trans. Inf. Theory, vol. 16, no. 1, pp. 85–89, 1970. [Online]. Available: https://ieeexplore.ieee.org/ document/1054411

    1970

  43. [51]

    Determination of the merit factor of legendre sequences,

    T. Hoholdt and H. E. Jensen, “Determination of the merit factor of legendre sequences,” IEEE Trans. Inf. Theory, vol. 34, pp. 161–164, 2002

    2002

  44. [52]

    Low autocorrelation binary sequences: Statistical mechanics and configuration space analysis,

    J. Bernasconi, “Low autocorrelation binary sequences: Statistical mechanics and configuration space analysis,” J. Phys., vol. 48, no. 4, pp. 559–567, 1987. [Online]. Available: https://www.semanticscholar.org/ paper/298a052b623e329bf57dc4819699983c61371691

    1987

  45. [53]

    Binary sequences with a maximally flat amplitude spectrum,

    G. F. M. Beenker, T. A. C. M. Claasen, and P. W. C. Hermens, “Binary sequences with a maximally flat amplitude spectrum,” Philips J. Res., vol. 40, pp. 289–304, 1985

    1985

  46. [54]

    Binary sequences with optimal autocorrelation,

    Y . Cai and C. Ding, “Binary sequences with optimal autocorrelation,” Theor. Comput. Sci., vol. 410, pp. 2316–2322, 2009. VOLUME 4, 2016 11 Kazuki Ikeuchiet al.: Evaluating the Performance of Direct Higher-Order Formulations in Combinatorial Optimization Problems

    2009

  47. [55]

    Analysis of pseudo- random sequence correlation identification parameters and anti-noise per- formance,

    X. Song, X. Wang, Z. Dong, X. Zhao, and X. Feng, “Analysis of pseudo- random sequence correlation identification parameters and anti-noise per- formance,” Energies, vol. 11, no. 10, p. 2586, 2018

    2018

  48. [56]

    Compilation of fault-tolerant quantum heuristics for combinatorial optimization,

    Y . R. Sanders, D. W. Berry, P. C. Costa, L. W. Tessler, N. Wiebe, C. Gid- ney, H. Neven, and R. Babbush, “Compilation of fault-tolerant quantum heuristics for combinatorial optimization,” PRX quantum, vol. 1, no. 2, p. 020312, 2020

    2020

  49. [57]

    Combinatorial optimization with quantum imaginary time evolution,

    N. M. Bauer, R. Alam, G. Siopsis, and J. Ostrowski, “Combinatorial optimization with quantum imaginary time evolution,” Phys. Rev. A, vol. 109, no. 5, p. 052430, 2024

    2024

  50. [58]

    A competitive NISQ and qubit-efficient solver for the LABS problem,

    M. Sciorilli, G. Camilo, T. O. Maciel, A. Canabarro, L. Borges, and L. Aolita, “A competitive NISQ and qubit-efficient solver for the LABS problem,” arXiv preprint arXiv:2506.17391, 2025

    2025

  51. [59]

    Evidence of scaling advantage for the quantum approximate optimization algorithm on a classically intractable problem,

    R. Shaydulin, C. Li, S. Chakrabarti, M. DeCross, D. Herman, N. Kumar, J. Larson, D. Lykov, P. Minssen, Y . Sun, Y . Alexeev, J. M. Dreiling, J. P. Gaebler, T. M. Gatterman, J. A. Gerber, K. Gilmore, D. Gresh, N. Hewitt, C. V . Horst, S. Hu, J. Johansen, M. Matheny, T. Mengle, ...

    2024

  52. [60]

    Optimization perfor- mance of factorization machine with annealing under limited training data,

    M. Nakano, Y . Seki, S. Kikuchi, and S. Tanaka, “Optimization perfor- mance of factorization machine with annealing under limited training data,” arXiv preprint arXiv:2507.21024, 2025

    2025

  53. [61]

    The merit factor of long low autocorrelation skew- symmetric binary sequences,

    M. J. E. Golay, “The merit factor of long low autocorrelation skew- symmetric binary sequences,” IEEE Trans. Inf. Theory, vol. 28, pp. 543– 549, 1982

    1982

  54. [62]

    A new search for skewsym- metric binary sequences with optimal merit factors,

    M. J. E. Golay and D. B. Harris, “A new search for skewsym- metric binary sequences with optimal merit factors,” IEEE Trans. Inf. Theory, vol. 36, pp. 1163–1166, 1990

    1990

  55. [63]

    The truck dispatching problem,

    G. B. Dantzig and J. H. Ramser, “The truck dispatching problem,” Manage. Sci., vol. 6, no. 1, pp. 80–91, 1959

    1959

  56. [64]

    Multiobjective evolutionary algorithms: a com- parative case study and the strength pareto approach,

    E. Zitzler and L. Thiele, “Multiobjective evolutionary algorithms: a com- parative case study and the strength pareto approach,” IEEE Trans. Evol. Comput., vol. 3, no. 4, pp. 257–271, 1999

    1999

  57. [65]

    The hypervolume indicator: Computational problems and algorithms,

    A. P. Guerreiro, C. M. Fonseca, and L. Paquete, “The hypervolume indicator: Computational problems and algorithms,” ACM Comput. Surv., vol. 54, no. 6, 2021. KAZUKI IKEUCHIreceived the B. Eng. degree in applied physics and physico-informatics from Keio University, Kanagawa, Jap...

    2021

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.