Pith. sign in

REVIEW 2 major objections 5 minor 37 references

COMET: Combinatorial Optimization for Multiplex Editing Targets Via Constraint-Preserving QAOA

T0 review · 2 major / 5 minor · reviewed 2026-07-12 · grok-4.5

Pith's one-line read On a multiplex CRISPR guide-RNA selection task, XY-mixer QAOA finds the optimum far more reliably than penalty-based QAOA and stays closer to simulation on real hardware.

desk verdict Clean empirical methods paper: XY-mixer vs penalty QAOA on a synthetic 12-qubit CRISPR QUBO, with honest Heron hardware and no overclaim. read the letter →

arxiv 2607.02622 v1 pith:Z5VYLMNW submitted 2026-07-02 quant-ph cs.AI

classification quant-phcs.AI
keywords QAOAXY-mixerQUBOconstraintenforcementmultiplexCRISPRguideRNAselectionquantumoptimizationerrormitigation
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

Multiplex CRISPR editing needs one guide RNA per gene while accounting for cross-gene interactions, which is a constrained combinatorial problem. The usual quantum approach folds the one-guide-per-gene rule into the cost function with penalty terms, but the penalty size is heuristic and the extra terms make hardware noise worse. This paper shows that the same rule can be built into the mixer instead: the XY-mixer keeps every state feasible by construction, so the cost Hamiltonian only encodes biology. On a three-gene, twelve-qubit instance for the immune-checkpoint genes PDCD1, LAG3 and HAVCR2, the XY-mixer reaches more than 95 percent probability of sampling an optimal selection by depth three in simulation, while three penalty variants spanning an order of magnitude in penalty strength stay below 6 percent at every depth. On IBM Heron hardware the XY-mixer’s energy stays within 0.8 of the noiseless value, whereas the worst-tuned penalty drifts by more than 50. The structural feasibility guarantee erodes under gate noise, yet the energy and sampling advantages remain at practical depths. The instance is classically trivial; the contribution is a clean methodological comparison of how constraints should be enforced when quantum optimizers are applied to biologically motivated selection problems.

What carries the argument

The XY-mixer: a mixer Hamiltonian that swaps states only inside each gene’s candidate group, thereby preserving the one-hot subspace by construction and removing the need for penalty terms in the cost Hamiltonian.

What would settle it

Re-run the identical QAOA comparison on a larger multiplex panel whose linear costs and cross-gene couplings are taken from measured on-target and off-target scores rather than synthetic values; if the XY-mixer advantage in optimum-sampling probability and energy gap disappears, the central claim fails for realistic biology.

Watch

Extended reading notes

Core claim

For one-hot multiplex guide-RNA selection formulated as QAOA, structural enforcement with the XY-mixer produces dramatically higher probability of the combinatorial optimum in simulation and far smaller simulator–hardware energy gaps on a Heron processor than any of the three penalty-coefficient variants tested, while the idealized 100 percent feasibility guarantee only partially survives gate-level noise.

Load-bearing premise

That the hand-crafted synthetic scores and pairwise interaction pattern used for the three genes are representative enough of real CRISPR scoring data for the ranking of the two constraint strategies to carry over.

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 / 5 minor

Summary. The manuscript formulates multiplex CRISPR gRNA selection (one guide per gene with cross-gene interactions) as a one-hot constrained QUBO and compares two QAOA constraint-enforcement strategies on a three-gene, twelve-qubit synthetic instance (PDCD1, LAG3, HAVCR2; four candidates each). Penalty-based QAOA (transverse-field mixer plus quadratic penalties with P in {2.0, 3.0, 5.0, ~13.65}) is contrasted with penalty-free XY-mixer QAOA that preserves the feasible subspace by construction. In noiseless simulation the XY-mixer reaches >95% probability of sampling a cost-optimal feasible solution by depth p=3 while all penalty variants stay below 6%; on ibm_kingston (Heron r2) the XY-mixer sim–hardware energy gap remains within |0.8| across p=1–5 while the worst penalty gap reaches +53.9. The authors report feasibility leakage under gate noise, ZNE over-correction at larger depth, classical triviality of the instance, and synthetic scoring data, framing the work as a methodological comparison rather than a claim of quantum advantage or biological transfer.

Significance. If the reported ranking holds, the paper supplies a concrete, hardware-validated demonstration that structural constraint enforcement via the XY-mixer can outperform penalty methods both in sampling concentration and in noise resilience for one-hot combinatorial problems of this type. The contribution is methodological rather than algorithmic novelty: the XY-mixer construction is known, but the systematic penalty sweep, honest disclosure of where the structural guarantee fails under gate noise, and execution on a Heron r2 device in a biologically motivated domain are useful for the near-term QAOA literature. Strengths include consistent metric definitions (P(opt), feasibility, energy gap), classical baselines that recover the known optimum, multi-restart COBYLA optimization, and explicit reporting of ZNE artifacts and feasibility decay. The 12-qubit scale and synthetic L_i/Q_ij limit claims of biological or quantum-advantage impact, which the authors already acknowledge.

major comments (2)
  1. §V.C and §VI.B–C: The decisive simulation ranking (XY-mixer P(opt)>95% vs penalty <6%) rests on COBYLA with 15 random restarts and max 500 iterations. The non-monotonic energy-vs-depth behavior for P=5.0 and P=13.65 is correctly interpreted as landscape pathology, yet the manuscript does not quantify whether a stronger classical optimizer (more restarts, warm-starts, or parameter concentration) would close the gap. Because the central claim is an empirical ranking of constraint-enforcement strategies under a fixed optimizer, a short sensitivity check or explicit statement that the ranking is conditional on this optimizer budget is needed to keep the claim load-bearing.
  2. §IV.B and §VIII.D: The synthetic linear costs L_i and pairwise Q_ij are constructed to produce a “greedy trap” (individually best candidates share the strongest positive couplings). While the authors correctly disclaim biological transfer, the abstract and introduction still present the instance as “targeting” named immune-checkpoint genes. A clearer separation—e.g., stating that the numerical values are synthetic and chosen to illustrate combinatorial structure—would prevent readers from over-interpreting the biological framing as part of the validated result.
minor comments (5)
  1. Fig. 4: Transpiled depth and two-qubit gate counts for the XY-mixer are higher than for the penalty variants at equal p; a brief remark in §VII on how this extra gate cost interacts with the observed energy-gap advantage would help readers weigh circuit overhead against noise resilience.
  2. §V.A: The automatic penalty heuristic returns P≈13.65; stating the exact formula or Qiskit version used would improve reproducibility.
  3. Fig. 5c caption and §VI.E: The distribution is labeled “p=5” while the text refers to “best-performing depth”; aligning the caption with the depth that actually achieves 95.2% would avoid minor confusion.
  4. References [19] and [20] appear with 2026 arXiv/journal dates; confirm that these are the intended citations and that they are publicly available at submission time.
  5. Notation: both S_k and G_k are used for gene groups (Eq. 4 vs Fig. 1); standardizing on one symbol would improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: empirical ranking of known constraint-enforcement strategies on a fixed synthetic instance

full rationale

The paper’s load-bearing claims are measured quantities (P(opt) vs depth on a noiseless simulator; sim–HW energy gaps and feasibility rates on ibm_kingston) obtained by running four explicitly constructed QAOA circuits (three penalty coefficients plus the complete-graph XY-mixer) under a fixed classical optimizer (COBYLA + 15 restarts). The XY-mixer’s ideal-circuit feasibility is a standard algebraic property of the mixer (commutes with Hamming-weight operators; cited to Hadfield/Wang/Fuchs), not redefined or fitted here. Synthetic L_i and Q_ij are stated as hand-chosen to produce a greedy trap; they are inputs, not parameters reverse-engineered from the reported ranking. Penalty coefficients are swept over an order of magnitude rather than tuned to force the claimed superiority. No uniqueness theorem, self-citation chain, or fitted constant is re-labeled as a prediction. Classical triviality and synthetic data are disclosed as limitations. The derivation chain therefore reduces only to ordinary numerical experiment, not to a circular identity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The load-bearing content is an empirical methods comparison on a hand-constructed 12-variable QUBO. Free parameters are the synthetic biology scores, the penalty sweep, and classical optimizer settings. Background axioms are standard QAOA/XY theory and the one-hot CRISPR encoding; no new physical entities are postulated.

free parameters (4)
  • synthetic linear costs L_i and pairwise Q_ij
    Hand-chosen on/off-target and cross-gene interaction numbers that create the greedy trap and the reported optimum; not measured from experiment.
  • penalty coefficients P ∈ {2.0, 3.0, 5.0, ~13.65}
    Manually and auto-selected values that define the three penalty variants; the ranking of methods depends on this sweep.
  • w_on = w_off = 1
    Equal weighting of on-target efficacy and off-target risk in L_i; changes the numerical landscape.
  • COBYLA settings (15 random restarts, max 500 iterations)
    Classical outer-loop hyperparameters that affect whether penalty landscapes are fairly optimized relative to the XY landscape.
assumptions (4)
  • standard math The complete-graph XY mixer within each gene group commutes with the Hamming-weight operator and therefore preserves one-hot feasibility when started from a feasible state (ideal unitary level).
    Invoked throughout §§III.E, V.B; taken from Wang et al. / Hadfield et al. Quantum Alternating Operator Ansatz literature.
  • domain assumption Multiplex gRNA selection with cross-gene interactions is faithfully represented by a QUBO with linear L_i and pairwise Q_ij plus one-hot constraints per gene.
    Problem formulation §IV; structural form is biologically motivated but numerical values are synthetic.
  • domain assumption QAOA performance is non-decreasing in depth p for ideal circuits, and COBYLA with random restarts is an adequate classical optimizer for comparing the four variants.
    Used to interpret non-monotonic penalty energies as landscape pathology rather than optimizer failure alone (§VI.C).
  • domain assumption ibm_kingston Heron-r2 noise and the chosen error-mitigation stack (measurement twirling, XY4 DD, digital ZNE) are representative enough for the reported sim–HW gaps to be informative.
    Hardware protocol §V.D–E; device-specific and calibration-dependent.

how reviews work

0 comments
Cite this review

Pith. "Pith review of COMET: Combinatorial Optimization for Multiplex Editing Targets Via Constraint-Preserving QAOA." pith.science (2026). https://pith.science/paper/Z5VYLMNW

@misc{pith2026260702622,
  author       = {Pith},
  title        = {Pith review of: COMET: Combinatorial Optimization for Multiplex Editing Targets Via Constraint-Preserving QAOA},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z5VYLMNW}},
  note         = {Machine review of arXiv:2607.02622}
}
read the original abstract

Multiplex CRISPR-Cas9 gene editing requires selecting one guide RNA per target gene subject to cross-gene interactions: a constrained combinatorial problem that can be formulated as a Quadratic Unconstrained Binary Optimization (QUBO) and solved via the Quantum Approximate Optimization Algorithm (QAOA). The one-hot per-gene constraint is conventionally enforced by adding quadratic penalty terms to the cost Hamiltonian, but penalty coefficient selection is heuristic and penalties amplify hardware noise. An alternative is to enforce the constraint structurally via the XY-mixer, which preserves feasibility by construction. We present COMET, a systematic comparison of penalty-based and XY-mixer QAOA on a three-gene, twelve-qubit multiplex editing instance targeting the immune-checkpoint genes PDCD1, LAG3, and HAVCR2. In simulation, the XY-mixer exceeds 95% probability of the optimum by QAOA depth p=3, while three penalty variants spanning an order of magnitude in penalty coefficient remain below 6% at every depth. On IBM's ibm_kingston (Heron r2) processor, the XY-mixer's simulator-hardware energy gap stays within |0.8| across all depths, while the worst-tuned penalty variant's gap reaches +53.9. We provide an honest account of where the structural guarantee partially breaks under gate-level noise. The twelve-qubit instance is classically trivial; our contribution is a methodological comparison of constraint-enforcement strategies in a biologically motivated domain, with real-hardware validation.

Figures

Figures reproduced from arXiv: 2607.02622 by the authors.

Figure 1
Figure 1. COMET problem structure and constraint-handling strategies. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The penalty approach searches all 2 12 = 4096 bitstrings (98.4% infeasible); the XY-mixer confines evolution to the 64 feasible states. A. QUBO Construction and Penalty Variants We construct the constrained problem of Eq. 5–4 as a QuadraticProgram and convert it to QUBO form via Qiskit’s QuadraticProgramToQubo, mapping to an Ising Hamiltonian HC through xi = (I − Zi)/2 .We evaluate three manual penalty coefficients … view at source ↗
Figure 3
Figure 3. All 64 feasible solutions ranked by cost. The com [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Transpiled circuit depth and two-qubit gate count for each QAOA variant as a function of depth [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Simulation results for the four QAOA variants across depths [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Hardware energy and feasibility measurements on [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: XY-mixer output distributions at best hardware depth: simulator (left), raw hardware (center), mitigated hardware (right). [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

37 extracted references · 4 linked inside Pith

  1. [1]

    The new frontier of genome engi- neering with crispr-cas9,

    J. A. Doudna and E. Charpentier, “The new frontier of genome engi- neering with crispr-cas9,”Science, vol. 346, no. 6213, p. 1258096, 2014

  2. [2]

    Multiplexed crispr technologies for gene editing and transcriptional regulation,

    N. S. McCartyet al., “Multiplexed crispr technologies for gene editing and transcriptional regulation,”Nature communications, vol. 11, no. 1, p. 1281, 2020

  3. [3]

    Crispr-cas9-mediated multiplex gene editing in car-t cells,

    X. Liuet al., “Crispr-cas9-mediated multiplex gene editing in car-t cells,”Cell research, vol. 27, no. 1, pp. 154–157, 2017

  4. [4]

    Multiplex genome engineering using crispr/cas systems,

    L. Cong, F. A. Ranet al., “Multiplex genome engineering using crispr/cas systems,”Science, vol. 339, no. 6121, pp. 819–823, 2013

  5. [5]

    Rna-guided human genome engineering via cas9,

    P. Maliet al., “Rna-guided human genome engineering via cas9,” Science, vol. 339, no. 6121, pp. 823–826, 2013

  6. [6]

    Simultaneous genetic ablation of pd-1, lag-3, and tim- 3 in cd8 t cells delays tumor growth and improves survival outcome,

    E. Ciraoloet al., “Simultaneous genetic ablation of pd-1, lag-3, and tim- 3 in cd8 t cells delays tumor growth and improves survival outcome,” International Journal of Molecular Sciences, vol. 23, no. 6, p. 3207, 2022

  7. [7]

    Optimized sgrna design to maximize activity and minimize off-target effects of crispr-cas9,

    J. G. Doenchet al., “Optimized sgrna design to maximize activity and minimize off-target effects of crispr-cas9,”Nature biotechnology, vol. 34, no. 2, pp. 184–191, 2016

  8. [8]

    Induced expression of pd-1, a novel member of the immunoglobulin gene superfamily, upon programmed cell death

    Y . Ishida, Y . Agata, K. Shibahara, and T. Honjo, “Induced expression of pd-1, a novel member of the immunoglobulin gene superfamily, upon programmed cell death.”The EMBO journal, vol. 11, no. 11, pp. 3887– 3895, 1992

Show all 37 references
  1. [9]

    Understanding lag-3 signaling,

    L. Chocarroet al., “Understanding lag-3 signaling,”International jour- nal of molecular sciences, vol. 22, no. 10, p. 5282, 2021

  2. [10]

    Th1-specific cell surface protein tim-3 regulates macrophage activation and severity of an autoimmune disease,

    L. Monneyet al., “Th1-specific cell surface protein tim-3 regulates macrophage activation and severity of an autoimmune disease,”Nature, vol. 415, no. 6871, pp. 536–541, 2002

  3. [11]

    A quantum approximate optimization algorithm,

    E. Farhi, J. Goldstone, and S. Gutmann, “A quantum approximate optimization algorithm,”arXiv preprint arXiv:1411.4028, 2014

  4. [12]

    The effect of penalty factors of constrained hamiltonians on the eigenspectrum in quantum annealing,

    C. Rochet al., “The effect of penalty factors of constrained hamiltonians on the eigenspectrum in quantum annealing,”ACM Transactions on Quantum Computing, vol. 4, no. 2, pp. 1–18, 2023

  5. [13]

    From the quantum approximate optimization algo- rithm to a quantum alternating operator ansatz,

    S. Hadfieldet al., “From the quantum approximate optimization algo- rithm to a quantum alternating operator ansatz,”Algorithms, vol. 12, no. 2, p. 34, 2019

  6. [14]

    Xy-mixers: analytical and numerical results for qaoa,

    E. Rieffel, J. M. Dominy, N. Rubin, and Z. Wang, “Xy-mixers: analytical and numerical results for qaoa,”Phys. Rev. A, vol. 101, p. 012320, 2020

  7. [15]

    A review on quantum approximate optimization algorithm and its variants,

    K. Blekoset al., “A review on quantum approximate optimization algorithm and its variants,”Physics Reports, vol. 1068, pp. 1–66, 2024

  8. [16]

    Xy-mixers: ana- lytical and numerical results for qaoa,

    Z. Wang, N. C. Rubin, J. M. Dominy, and E. G. Rieffel, “Xy-mixers: ana- lytical and numerical results for qaoa,”arXiv preprint arXiv:1904.09314, 2019

  9. [17]

    Constraint preserving mixers for the quantum approximate optimization algorithm,

    F. G. Fuchs, K. O. Lye, H. Møll Nilsen, A. J. Stasik, and G. Sartor, “Constraint preserving mixers for the quantum approximate optimization algorithm,”Algorithms, vol. 15, no. 6, p. 202, 2022

  10. [18]

    Analytical framework for quan- tum alternating operator ans ¨atze,

    S. Hadfield, T. Hogg, and E. G. Rieffel, “Analytical framework for quan- tum alternating operator ans ¨atze,”Quantum Science and Technology, vol. 8, no. 1, p. 015017, 2023

  11. [19]

    Constrained quantum optimization via iterative warm- start xy-mixers,

    D. Bucheret al., “Constrained quantum optimization via iterative warm- start xy-mixers,”arXiv preprint arXiv:2604.02083, 2026

  12. [20]

    Constraint-aware quantum optimization via hamming weight operators,

    Y . Hao, Q. Ding, X. Yuan, and X. Wang, “Constraint-aware quantum optimization via hamming weight operators,”Science China Physics, Mechanics & Astronomy, vol. 69, no. 5, p. 250314, 2026

  13. [21]

    Ising formulations of many np problems,

    A. Lucas, “Ising formulations of many np problems,”Frontiers in physics, vol. 2, p. 74887, 2014

  14. [22]

    A tutorial on formulating and using qubo models,

    F. Glover, G. Kochenberger, and Y . Du, “A tutorial on formulating and using qubo models,”arXiv preprint arXiv:1811.11538, 2018

  15. [23]

    Molecular docking via quantum approximate optimization algorithm,

    Q.-M. Ding, Y .-M. Huang, and X. Yuan, “Molecular docking via quantum approximate optimization algorithm,”Physical Review Applied, vol. 21, no. 3, p. 034036, 2024

  16. [24]

    Genome assembly using quantum and quantum-inspired annealing,

    A. Boev, A. Rakitko, S. Usmanov, A. Kobzeva, I. Popov, V . Ilinsky, E. Kiktenko, and A. Fedorov, “Genome assembly using quantum and quantum-inspired annealing,”Scientific Reports, vol. 11, no. 1, p. 13183, 2021

  17. [25]

    Quantum computing for genomics: conceptual challenges and practical perspectives,

    A. Maurizio and G. Mazzola, “Quantum computing for genomics: conceptual challenges and practical perspectives,”PRX Life, vol. 3, no. 4, p. 047001, 2025

  18. [26]

    Quantum biological insights into crispr-cas9 sgrna efficiency from explainable-ai driven feature engineering,

    J. M. Noshay, T. Walker, W. G. Alexander, D. M. Klingeman, J. Romero, A. M. Walker, E. Prates, C. Eckert, S. Irle, D. Kaineret al., “Quantum biological insights into crispr-cas9 sgrna efficiency from explainable-ai driven feature engineering,”Nucleic acids research, vol. 51, n...

  19. [27]

    Error mitigation for short- depth quantum circuits,

    K. Temme, S. Bravyi, and J. M. Gambetta, “Error mitigation for short- depth quantum circuits,”Physical review letters, vol. 119, no. 18, p. 180509, 2017

  20. [28]

    Efficient variational quantum simulator incorporating active error minimization,

    Y . Li and S. C. Benjamin, “Efficient variational quantum simulator incorporating active error minimization,”Physical Review X, vol. 7, no. 2, p. 021050, 2017

  21. [29]

    Digital zero noise extrapolation for quantum error mitigation,

    T. Giurgica-Tiron, Y . Hindy, R. LaRose, A. Mari, and W. J. Zeng, “Digital zero noise extrapolation for quantum error mitigation,” in2020 IEEE international conference on quantum computing and engineering (QCE). IEEE, 2020, pp. 306–316

  22. [30]

    Best practices for quantum error mitigation with digital zero-noise extrapola- tion,

    R. Majumdar, P. Rivero, F. Metz, A. Hasan, and D. S. Wang, “Best practices for quantum error mitigation with digital zero-noise extrapola- tion,” in2023 IEEE International Conference on Quantum Computing and Engineering (QCE), vol. 1. IEEE, 2023, pp. 881–887

  23. [31]

    Quantum computing with qiskit,

    A. Javadi-Abhari, M. Treinish, K. Krsulich, C. J. Wood, J. Lishman, J. Gacon, S. Martiel, P. D. Nation, L. S. Bishop, A. W. Crosset al., “Quantum computing with qiskit,”arXiv preprint arXiv:2405.08810, 2024

  24. [32]

    Dynamical decoupling of open quantum systems,

    L. Viola, E. Knill, and S. Lloyd, “Dynamical decoupling of open quantum systems,”Physical Review Letters, vol. 82, no. 12, p. 2417, 1999

  25. [33]

    Dy- namical decoupling for superconducting qubits: A performance survey,

    N. Ezzell, B. Pokharel, L. Tewala, G. Quiroz, and D. A. Lidar, “Dy- namical decoupling for superconducting qubits: A performance survey,” Physical Review Applied, vol. 20, no. 6, p. 064027, 2023

  26. [34]

    Synergistic dynamical decoupling and circuit design for enhanced algorithm performance on near-term quantum devices,

    Y . Ji and I. Polian, “Synergistic dynamical decoupling and circuit design for enhanced algorithm performance on near-term quantum devices,” Entropy, vol. 26, no. 7, p. 586, 2024

  27. [35]

    Evidence for the utility of quantum computing before fault tolerance,

    Y . Kim, A. Eddins, S. Anand, K. X. Wei, E. Van Den Berg, S. Rosenblatt, H. Nayfehet al., “Evidence for the utility of quantum computing before fault tolerance,”Nature, vol. 618, no. 7965, pp. 500–505, 2023

  28. [36]

    Empirical evaluation of QAOA with zero noise extrapo- lation on NISQ hardware for carbon credit portfolio optimization in the Brazilian Cerrado,

    H. J. Ribeiro, “Empirical evaluation of QAOA with zero noise extrapo- lation on NISQ hardware for carbon credit portfolio optimization in the Brazilian Cerrado,”arXiv preprint arXiv:2602.09047, 2026

  29. [37]

    Direct analysis of zero-noise extrapo- lation: Polynomial methods, error bounds, and simultaneous physical- algorithmic error mitigation,

    P. Mohammadipour and X. Li, “Direct analysis of zero-noise extrapo- lation: Polynomial methods, error bounds, and simultaneous physical- algorithmic error mitigation,”Quantum, vol. 9, p. 1909, 2025

Pith tools

Reviewed July 12, 2026 · model on record in the stance chip above.