Pith. sign in

REVIEW 3 major objections 5 minor 49 references

A classical simulator can draw exact samples from quantum circuits by letting a bitstring evolve stochastically alongside the wavefunction, reaching 476-qubit QAOA circuits that full-state methods cannot handle.

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

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

T0 review · deepseek-v4-flash

2026-08-04 07:48 UTC pith:WMBE3GHH

load-bearing objection The noiseless sampler is solid and worth knowing; the noisy extension has a state-dependent sampling bug that breaks the exactness claim for amplitude-damping-like noise. the 3 major comments →

arxiv 2510.24218 v2 pith:WMBE3GHH submitted 2025-10-28 quant-ph

Pilot-Wave Simulator: Exact Classical Sampling from Ideal and Noisy Quantum Circuits up to Hundreds of Qubits

Gleb Kalachev , Pavel Mosharev , Zuoheng Zou , Pavel Panteleev , Man-Hong Yung This is my paper
classification quant-ph MSC 81P6868Q12
keywords quantum circuit simulationexact samplingtensor networkspilot-wave methodQAOAnoisy quantum circuitsclassical simulationMarkov chain sampling
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.

Exact classical sampling from a quantum circuit — producing bitstrings with the true measurement probabilities — is normally out of reach for large circuits, because the full state vector is exponentially large. This paper claims that the task can be reduced to a sequence of small stochastic updates: keep a classical bitstring that changes after each gate, and compute the few transition amplitudes needed at each step by tensor-network contraction of the prefix circuit. The main bound makes the sampling cost polynomial in the number of non-monomial gates, the gate width, and the cost of a single amplitude, so the problem becomes no harder than single-amplitude simulation. The authors demonstrate the approach on QAOA (a quantum optimization ansatz) circuits for spin-glass problems, producing exact samples from ideal and noisy circuits with up to 476 qubits at depth 1 and 49 qubits at depth 3, and use those samples to study QAOA's thermal-like output distributions and its competition with a simple classical local-update rule. If the central claim holds, exact sampling becomes practical for shallow structured circuits that full-state simulation cannot touch.

Core claim

At its core, the paper establishes a polynomial reduction from exact sampling to single-amplitude simulation. For any circuit C, generating one sample costs O(qsize(C)·2^ℓ(C)·Ĉ_amp(C)), where qsize(C) is the number of gates whose matrices are not monomial (a product of a permutation and a diagonal), ℓ(C) is the maximum number of target qubits per gate, and Ĉ_amp(C) is the largest cost of computing one amplitude of any prefix subcircuit. The algorithm maintains a classical bitstring s_t together with the quantum wavefunction ψ_t, using transition probabilities chosen so that Pr(s_t=x)=|⟨x|ψ_t⟩|² at every step. Each non-monomial gate is split into blocks according to the connected components o

What carries the argument

The central mechanism is a pilot-wave Markov chain driven by a tensor-network amplitude oracle. A classical n-bit string s_t evolves in parallel with the quantum state so that at every time the string's distribution equals the circuit's measurement distribution. The transition rule uses the block structure of each gate matrix: from the current string, the possible next strings are the connected component of the nonzero-support graph of the gate matrix restricted to the target qubits, and the transition probabilities are the normalized measurement probabilities of the amplitudes in that block. The amplitudes are computed on demand by contracting the tensor network of the prefix subcircuit; be

Load-bearing premise

The load-bearing premise is that every amplitude needed during the walk can be evaluated exactly and cheaply by tensor-network contraction of the prefix subcircuit; for circuits that are not shallow and sparse this per-amplitude cost is typically exponential, so the polynomial sampling bound does not by itself guarantee scalability to hundreds of qubits, and the noisy version additionally assumes all noise operators are local-submonomial so they do not change the tensor-netwo

What would settle it

Take a 100-qubit random circuit with two-qubit gates arranged in a brickwork pattern for depth 50 and measure the time per sample as depth grows. If the runtime grows exponentially in depth because the tensor-network contraction width of the prefix subcircuits blows up, then the 'up to hundreds of qubits' demonstration does not generalize beyond shallow circuits; the paper's own bound would still hold but its practical reach would collapse. A sharper probe: insert a single non-monomial gate acting on 50 target qubits (ℓ=50) into an otherwise easy 100-qubit circuit; the algorithm would need to

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

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Exact samples were produced from ideal and noisy QAOA circuits with 476 qubits at depth 1 and 49 qubits at depth 3, using only CPU servers, far beyond what full-state simulators can sample exactly.
  • For any circuit whose amplitude oracle is efficient, the sampling problem is only a factor O(qsize·2^ℓ) harder than computing one amplitude, so exact sampling and single-amplitude simulation rise and fall together.
  • The sampled ground-state probability of fixed-depth QAOA decreases exponentially with system size, supporting the conclusion that shallow QAOA is unlikely to be an effective large-scale optimizer.
  • The sampled output distributions show a pseudo-Boltzmann profile whose effective temperature drops with circuit depth; noise raises the effective temperature and suppresses the low-energy tail in a predictable way.
  • On the tested 100- and 256-qubit spin-glass instances, a simple classical local-update rule matches the performance of QAOA at the same depth, and beats it when realistic depolarizing and amplitude-damping noise is added.

Where Pith is reading between the lines

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

  • The 'hundreds of qubits' headline is a property of the demonstrated circuit class — shallow, sparse, structured QAOA circuits — not of arbitrary quantum circuits; the sampling bound inherits the per-amplitude cost, which can be exponential for dense deep circuits. The paper itself demonstrates rather than proves scalability for generic circuits.
  • Any independent improvement in exact single-amplitude tensor-network contraction would automatically translate into faster exact sampling, making the algorithm a beneficiary of all progress in tensor-network simulation.
  • The pilot-wave updating scheme is agnostic to the amplitude backend; replacing the tensor-network oracle with another exact or approximate amplitude method would yield a family of samplers with different speed-fidelity trade-offs, though the paper does not explore this.
  • The observed parity between a classical local-update rule and QAOA at equal depth, if it extends to denser graphs and higher depth, would strengthen the case against a shallow-depth quantum advantage for spin-glass optimization; testing at higher depth and with structured problem classes would settle that.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes a classical sampling algorithm for quantum circuits, building on the pilot-wave/sequential-sampling ideas of [13,14] and using a tensor-network amplitude oracle [15]. The method maintains a classical bitstring s whose marginal distribution matches the Born distribution of the current circuit state; for each gate, it decomposes the gate matrix into blocks and updates s according to the block probabilities, computing only a few amplitudes per gate. Proposition 1 bounds the sampling cost as O(qsize(C) 2^{ℓ(C)} \hat C_amp(C)), reducing sampling to single-amplitude simulation. The algorithm is extended to noisy circuits under local-submonomial (LSM) Kraus operators, using X-gate propagation so that the tensor-network shape is preserved. Demonstrations include QAOA circuits with up to 476 qubits at depth p=1 and 49 qubits at p=3, under depolarizing and amplitude-damping noise, together with pseudo-Boltzmann energy distributions and a comparison with Hastings' classical local-update algorithm.

Significance. If correct, the algorithm provides an exact and practical sampling method for shallow structured circuits, extending prior sequential-sampling frameworks with modern tensor-network contraction and giving access to scales beyond full-state simulation. The central noiseless derivation is clean and non-circular: condition (1) is verified from unitarity and the Born rule, and Proposition 1 is a genuine reduction of sampling to amplitude evaluation. The numerical experiments, if reproducible, would be a useful benchmark for QAOA and for classical-versus-quantum optimization comparisons. Main weaknesses are the absence of shipped code, lack of statistical error bars on several scaling claims, and, most importantly, a correctness gap in the noisy-sampling algorithm described below.

major comments (3)
  1. [§3.1, Eq. (4) and Appendix A, Algorithm 2] The stochastic choice of Kraus operator in the noisy extension appears to use the wrong classical state. Eq. (4) defines Pr(i|s) = ⟨s[q]|E_i†E_i|s[q]⟩ for the classical state s. In Algorithm 2, however, s is the state in the Pauli frame, while the physical state is s⊕x (the algorithm returns (s+x) mod 2). For state-dependent noise such as amplitude damping, E_i†E_i is not proportional to identity, so the sampling probability must be evaluated on s⊕x, not on s. Concretely, for a single qubit prepared in |+⟩ and subjected to amplitude damping Λ_γ twice, the exact output distribution is P(0)=(1+2γ−γ²)/2, P(1)=(1−γ)²/2. If Eq. (4) is evaluated on the frame state s, the simulated distribution is P(0)=1/2+γ−γ², P(1)=1/2−γ+γ², an O(γ²) error for γ>0. The X-toggle in lines 13–15 handles deterministic propagation only; it does not compensate for using the wrong state when drawing the Kraus index.
  2. [§2, Proposition 1; §4.3, Fig. 4] The sampling-complexity bound is honest, but the title/abstract claim of sampling 'up to hundreds of qubits' is broader than what Proposition 1 guarantees. The cost bound contains \hat C_amp(C), the cost of the tensor-network oracle for prefix subcircuits, which can be exponential for general circuits. The method is efficient only when these amplitudes can be contracted cheaply, as for the shallow sparse QAOA circuits used here. The experimental support in Fig. 4 is limited to p=1, shows no error bars or confidence intervals, and reports average wall-clock time without accounting for variance across instances. The 'hundreds of qubits' claim should therefore be explicitly qualified as demonstrated for low-depth circuits with cheaply contractible tensor networks, not as a generic property of arbitrary circuits.
  3. [§3.1, Proposition 2 and Appendix A] The paper nowhere proves that Algorithm 2 produces samples from the distribution of the noisy circuit. Proposition 2 is a complexity statement; it does not establish correctness. A formal invariant is needed: after each gate, the returned classical configuration (including the Pauli-frame offset x) should be distributed according to the Born probabilities of the actual noisy state. Without such an invariant, the noisy-sampling claim is unsupported even if the Kraus-selection rule is corrected as suggested above.
minor comments (5)
  1. [Appendix A, Algorithm 2 (lines 9–10)] The pseudocode omits the random draw that determines E_i. It says 'Determine the operator E as shown in Eq. (3)' but does not specify the distribution. Please make the sampling step explicit and tie it to a corrected version of Eq. (4).
  2. [Fig. 4] Please add error bars or confidence intervals to the timing data, or state explicitly that points are single averages over 10 instances; currently the absence of variance information makes the scaling trend hard to assess.
  3. [§4.1, Fig. 2] The effective inverse temperatures β are fitted values, but the fitting procedure and uncertainties are not described. Please state how β was estimated and report standard errors.
  4. [Introduction] The sentence that rejection-sampling methods are 'effective only when output probabilities are nearly uniform' is too strong; rejection sampling can be effective in other regimes if a good proposal is available. Please soften or qualify.
  5. [General] No code or reproducibility package is provided. Given the numerical claims, even a minimal implementation or detailed algorithmic pseudocode for the amplitude oracle would significantly strengthen the paper.

Circularity Check

0 steps flagged

No significant circularity: the exact-sampling derivation is self-contained; the only overlapping-author citation ([15]) is a black-box amplitude backend, not the target result.

full rationale

The central derivation, Algorithm 1 and Proposition 1, does not reduce to its inputs by construction. The transition probabilities are defined directly from the Born probabilities of the prefix subcircuits and are shown to satisfy the marginal-matching condition (Eq. 1) using only unitarity of the gate blocks; the sample path is then a Markov chain with the required instantaneous distribution. Proposition 1 is a complexity accounting bound: it counts at most 2^ℓ(C) amplitude evaluations per non-monomial gate and multiplies by the single-amplitude cost Ĉ_amp(C). That is a conditional reduction to an oracle, not a circular prediction. The only self-citation with author overlap is [15] (Kalachev–Panteleev–Yung), cited as an example backend for amp(C_t,i); it is not invoked to justify correctness, to establish uniqueness, or to forbid alternatives, and the theorem holds for any exact amplitude oracle. Fitted quantities in the demonstration (effective temperatures β, QAOA parameter transfer) are experiment inputs and do not enter the correctness argument. The skeptical concern about Algorithm 2's Pauli-frame noise sampling (Eq. (4) using s rather than the physical state s⊕x) is a potential correctness bug, but it is not a circularity reduction and therefore does not raise this score.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

The central algorithm is a mathematical construction with no new physical entities. The free parameters are confined to the numerical demonstrations, not to the sampling theorem. The principal assumptions are the black-box exactness/efficiency of the tensor-network oracle and the restriction to local-submonomial noise; both are stated, but the scalability claim inherits their domain.

free parameters (3)
  • Noise model parameters = single-qubit error 0.005; two-qubit error 0.01; readout error 0.05; single-qubit gate 30 ns; two-qubit gate 80 ns; T1=10
    Chosen by hand as round numbers near Sycamore/Zuchongzhi values (Section 4.1); they drive all noisy numerical results and are not derived from a hardware model.
  • Hastings algorithm step size c_t = 0.3 for all t
    Used for all Hastings runs (Appendix C); no tuning procedure or sensitivity analysis is reported.
  • Effective inverse temperature β = Ideal: 0.273 (p=3), 0.224 (p=2), 0.143 (p=1); Noisy: 0.226 (p=3), 0.196 (p=2), 0.132 (p=1), as shown in Fig. 2
    These values are fitted to the sampled histograms to support the pseudo-Boltzmann claim; they are data fits, not predictions.
axioms (3)
  • standard math Quantum noise has an operator-sum (Kraus) representation with a unitary dilation using an ancilla qudit (Eq. 2).
    Used in Section 3.1 to replace a noise channel by an ancilla-assisted unitary followed by immediate measurement.
  • domain assumption The noise channels under consideration are local-submonomial (LSM): each Kraus operator is a tensor product of 2×2 matrices with at most one nonzero entry per row and column.
    Required for Proposition 2 and for evaluating noise probabilities with Eq. (4) without adding gates; holds for depolarizing, amplitude damping, and phase damping, but not for arbitrary noise.
  • domain assumption Tensor-network contraction [15] yields exact amplitudes of every prefix subcircuit and remains tractable for QAOA p<=3 on sparse graphs.
    This is the load-bearing premise for all numerical demonstrations; Proposition 1 retains exponential dependence on the per-amplitude cost, so the 'hundreds of qubits' demonstrations depend on the structure of the chosen circuits.

pith-pipeline@v1.3.0-alltime-deepseek · 16864 in / 16061 out tokens · 156348 ms · 2026-08-04T07:48:43.558440+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of Pilot-Wave Simulator: Exact Classical Sampling from Ideal and Noisy Quantum Circuits up to Hundreds of Qubits." pith.science (2026). https://pith.science/paper/WMBE3GHH

@misc{pith2026251024218,
  author       = {Pith},
  title        = {Pith review of: Pilot-Wave Simulator: Exact Classical Sampling from Ideal and Noisy Quantum Circuits up to Hundreds of Qubits},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WMBE3GHH}},
  note         = {Machine review of arXiv:2510.24218}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Quantum circuit simulators running on classical computers offer a vital platform for designing, testing, and optimizing quantum algorithms, driving innovation despite limited access to real quantum hardware. However, their scalability is inherently constrained by exponential memory and computational overhead, which restricts accurate simulation of large-scale quantum circuits and often results in approximate output distributions. Here, we propose an exact sampling algorithm that integrates tensor network contraction techniques with a Markov process, wherein a classical state evolves according to the local structure of the quantum circuit. As a demonstration, we target the challenge of generating samples from ideal and noisy QAOA circuits with up to 476 qubits, incorporating both depolarizing and amplitude damping noise models. These results enable further validation of several assumptions and conjectures at a scale previously out of reach, significantly expanding the scope of classical simulation in quantum algorithm research.

Figures

Figures reproduced from arXiv: 2510.24218 by Gleb Kalachev, Man-Hong Yung, Pavel Mosharev, Pavel Panteleev, Zuoheng Zou.

Figure 1
Figure 1. Figure 1: Panel (a) shows that Ut has the block structure of A ⊗ I2n−m (up to a basis permutation). Each block corresponds to the gate matrix A, which can be either (b) a monomial gate (i.e., exactly one nonzero per row and column) or (c) a general gate. Matrices of general gates are further partitioned into blocks using the BlockDecompose procedure from Algorithm 1. Proposition 1. For arbitrary quantum circuit C Cb… view at source ↗
Figure 2
Figure 2. Figure 2: Sampling results for QAOA circuits obtained with a simulated quantum device. Panels (a,b) correspond to the ideal [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Experimental results of sampling from QAOA circuit in comparison to classical Hastings algorithm. Each count on the [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Sampling times of QAOA circuits with p = 1, averaged over 10 independent problem instances for different graph topologies and problem sizes. rithm are close to depth-1 QAOA, and the results of the double-step Hastings are close to depth-2 QAOA in the ideal case. It is also clear that quantum noise deterio￾rates the performance of QAOA, making it inferior to the classical Hastings algorithm at a similar dep… view at source ↗
Figure 5
Figure 5. Figure 5: Fusion and propagation of an X gate in a quantum circuit. of TG will be identified with corresponding input legs. It is important to note that any one-qubit diagonal gate, such as Z, can be represented by a tensor with one leg. With this simplification, for example, a QAOA circuit of depth 1 for the Max-Cut problem on the graph Γ corresponds to the tensor network that is isomorphic to graph Γ where the edg… 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

49 extracted references · 8 canonical work pages

  1. [1]

    Xin-Chuan Wu, Sheng Di, Emma Maitreyee Das- gupta, Franck Cappello, Hal Finkel, Yuri Alexeev, and Frederic T. Chong. Full-state quantum cir- cuit simulation by using data compression. InPro- ceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC ’19, New York, NY, USA, 2019. As- sociation for Computin...

    arXiv 2019

  2. [2]

    Hq-sim: High-performance state vector simulation of quantum circuits on hetero- geneous HPC systems

    Boyuan Zhang, Bo Fang, Qiang Guan, Ang Li, and Dingwen Tao. Hq-sim: High-performance state vector simulation of quantum circuits on hetero- geneous HPC systems. InProceedings of the 2023 International Workshop on Quantum Classical Co- operative, QCCC ’23, page 1–4, New York, NY, USA, 2023. Association for Computing Machinery. doi:10.1145/3588983.3596679

    arXiv 2023

  3. [3]

    Benchmarking quantum computer simulation software packages: State vector simu- lators.SciPost Physics Core, 7, 11 2024.doi: 10.21468/SciPostPhysCore.7.4.075

    Amit Gangapuram, Andreas L¨ auchli, and Cor- nelius Hempel. Benchmarking quantum computer simulation software packages: State vector simu- lators.SciPost Physics Core, 7, 11 2024.doi: 10.21468/SciPostPhysCore.7.4.075

    work page doi:10.21468/scipostphyscore.7.4.075 2024

  4. [4]

    T. E. O’Brien, B. Tarasinski, and L. DiCarlo. Density-matrix simulation of small surface codes under current and projected experimental noise. npj Quantum Information, 3(1):39, Sep 2017.doi: 10.1038/s41534-017-0039-x

    work page doi:10.1038/s41534-017-0039-x 2017

  5. [5]

    Density matrix quantum circuit sim- ulation via the bsp machine on modern gpu clus- ters

    Ang Li, Omer Subasi, Xiu Yang, and Sriram Krish- namoorthy. Density matrix quantum circuit sim- ulation via the bsp machine on modern gpu clus- ters. InSC20: International Conference for High Performance Computing, Networking, Storage and Analysis, pages 1–15, 2020.doi:10.1109/SC4140 5.2020.00017

    arXiv 2020

  6. [6]

    Efficient classical simulation of slightly entangled quantum computations.Phys

    Guifr´ e Vidal. Efficient classical simulation of slightly entangled quantum computations.Phys. Rev. Lett., 91:147902, Oct 2003.doi:10.1103/Ph ysRevLett.91.147902

    doi:10.1103/ph 2003

  7. [7]

    Miles Stoudenmire, and Xavier Waintal

    Yiqing Zhou, E. Miles Stoudenmire, and Xavier Waintal. What limits the simulation of quantum computers?Phys. Rev. X, 10:041038, Nov 2020. doi:10.1103/PhysRevX.10.041038

    doi:10.1103/physrevx.10.041038 2020

  8. [8]

    Miles Stoudenmire, and Xavier Waintal

    Thomas Ayral, Thibaud Louvet, Yiqing Zhou, Cy- prien Lambert, E. Miles Stoudenmire, and Xavier Waintal. Density-matrix renormalization group al- gorithm for simulating quantum circuits with a fi- nite fidelity.PRX Quantum, 4:020304, Apr 2023. doi:10.1103/PRXQuantum.4.020304

    doi:10.1103/prxquantum.4.020304 2023

  9. [9]

    Simulating quantum circuits with tree tensor net- works using density-matrix renormalization group algorithm, 2025.arXiv:2504.16718

    Aditya Dubey, Zeki Zeybek, and Peter Schmelcher. Simulating quantum circuits with tree tensor net- works using density-matrix renormalization group algorithm, 2025.arXiv:2504.16718

    Pith/arXiv arXiv 2025

  10. [10]

    Jahromi, Sukhbinder Singh, and Rom´ an Or´ us

    Siddhartha Patra, Saeed S. Jahromi, Sukhbinder Singh, and Rom´ an Or´ us. Efficient tensor net- work simulation of ibm’s largest quantum proces- sors.Phys. Rev. Res., 6:013326, Mar 2024.doi: 10.1103/PhysRevResearch.6.013326

    doi:10.1103/physrevresearch.6.013326 2024

  11. [11]

    Ignacio Cirac, and Rahul Trivedi

    Guillermo Gonz´ alez-Garc ´ ıa, J. Ignacio Cirac, and Rahul Trivedi. Pauli path simulations of noisy quantum circuits beyond average case.Quantum, 10 9:1730, May 2025.doi:10.22331/q-2025-05-0 5-1730

    work page doi:10.22331/q-2025-05-0 2025

  12. [12]

    A flexible high-performance simulator for verifying and benchmarking quan- tum circuits implemented on real hardware.npj Quantum Information, 5(1):1–16, October 2019

    Benjamin Villalonga, Sergio Boixo, Bron Nelson, Christopher Henze, Eleanor Rieffel, Rupak Biswas, and Salvatore Mandr` a. A flexible high-performance simulator for verifying and benchmarking quan- tum circuits implemented on real hardware.npj Quantum Information, 5(1):1–16, October 2019. doi:10.1038/s41534-019-0196-1

    work page doi:10.1038/s41534-019-0196-1 2019

  13. [13]

    Frank, Liviu Oniciuc, Uwe H

    Michael P. Frank, Liviu Oniciuc, Uwe H. Meyer- Baese, and Irinel Chiorescu. A space-efficient quan- tum computer simulator suitable for high-speed FPGA implementation. In Eric J. Donkor, An- drew R. Pirich, and Howard E. Brandt, editors, Quantum Information and Computation VII, vol- ume 7342, page 734203. International Society for Optics and Photonics, SP...

    2009

  14. [14]

    How to simulate quantum measurement without computing marginals.Physical Review Letters, 128(22):220503, June 2022.arXiv:2112.08499, doi:10.1103/PhysRevLett.128.220503

    Sergey Bravyi, David Gosset, and Yinchen Liu. How to simulate quantum measurement without computing marginals.Physical Review Letters, 128(22):220503, June 2022.arXiv:2112.08499, doi:10.1103/PhysRevLett.128.220503

    Pith/arXiv arXiv 2022

  15. [15]

    Recursive multi-tensor contraction for xeb verification of quantum circuits, 2021.arXiv:21 08.05665

    Gleb Kalachev, Pavel Panteleev, and Man-Hong Yung. Recursive multi-tensor contraction for xeb verification of quantum circuits, 2021.arXiv:21 08.05665

    2021

  16. [16]

    A suggested interpretation of the quantum theory in terms of ”hidden” variables

    David Bohm. A suggested interpretation of the quantum theory in terms of ”hidden” variables. i. Phys. Rev., 85:166–179, Jan 1952.doi:10.1103/ PhysRev.85.166

    1952

  17. [17]

    Pilot-wave quan- tum theory in discrete space and time and the principle of least action.Foundations of Physics, 46:1502–1521, 2016.doi:10.1007/s10701-016-0 026-7

    Janusz Gluza and Jerzy Kosek. Pilot-wave quan- tum theory in discrete space and time and the principle of least action.Foundations of Physics, 46:1502–1521, 2016.doi:10.1007/s10701-016-0 026-7

    work page doi:10.1007/s10701-016-0 2016

  18. [18]

    Strategies for running the qaoa at hundreds of qubits, 2024

    Brandon Augustino, Madelyn Cain, Edward Farhi, Swati Gupta, Sam Gutmann, Daniel Ranard, Eu- gene Tang, and Katherine Van Kirk. Strategies for running the qaoa at hundreds of qubits, 2024. arXiv:2410.03015

    Pith/arXiv arXiv 2024

  19. [19]

    J. A. Monta˜ nez-Barrera and Kristel Michielsen. Toward a linear-ramp qaoa protocol: evidence of a scaling advantage in solving some combinatorial optimization problems.npj Quantum Information, 11(1):131, Aug 2025.doi:10.1038/s41534-025 -01082-1

    work page doi:10.1038/s41534-025 2025

  20. [20]

    Gpu- accelerated simulations of quantum annealing and the quantum approximate optimization algorithm

    Dennis Willsch, Madita Willsch, Fengping Jin, Kristel Michielsen, and Hans De Raedt. Gpu- accelerated simulations of quantum annealing and the quantum approximate optimization algorithm. Computer Physics Communications, 278:108411, 2022.doi:10.1016/j.cpc.2022.108411

    arXiv 2022

  21. [21]

    Quantum approximate optimization algorithm pseudo-boltzmann states.Phys

    Pablo D ´ ıez-Valle, Diego Porras, and Juan Jos´ e Garc ´ ıa-Ripoll. Quantum approximate optimization algorithm pseudo-boltzmann states.Phys. Rev. Lett., 130:050601, Feb 2023.doi:10.1103/PhysRe vLett.130.050601

    doi:10.1103/physre 2023

  22. [22]

    G´ omez-Ruiz, Diego Porras, and Juan Jos´ e Garc ´ ıa-Ripoll

    Pablo D ´ ıez-Valle, Fernando J. G´ omez-Ruiz, Diego Porras, and Juan Jos´ e Garc ´ ıa-Ripoll. Universal resources for qaoa and quantum annealing, 2025. arXiv:2506.03241

    arXiv 2025

  23. [23]

    Connection between single-layer quantum approximate optimization algorithm in- terferometry and thermal distribution sampling

    Pablo D ´ ıez-Valle, Diego Porras, and Juan Jos´ e Garc ´ ıa-Ripoll. Connection between single-layer quantum approximate optimization algorithm in- terferometry and thermal distribution sampling. Frontiers in Quantum Science and Technology, 3, 2024.doi:10.3389/frqst.2024.1321264

    arXiv 2024

  24. [24]

    Tensor net- work quantum simulator with step-dependent par- allelization

    Danylo Lykov, Roman Schutski, Alexey Galda, Valeri Vinokur, and Yuri Alexeev. Tensor net- work quantum simulator with step-dependent par- allelization. In2022 IEEE International Con- ference on Quantum Computing and Engineering (QCE), pages 582–593, 2022.doi:10.1109/QCE5 3715.2022.00081

    arXiv 2022

  25. [25]

    M. B. Hastings. Classical and quantum bounded depth approximation algorithms, 2019.arXiv:19 05.07047

    2019

  26. [26]

    Nielsen and Isaac L

    Michael A. Nielsen and Isaac L. Chuang.Quan- tum Computation and Quantum Information: 10th Anniversary Edition. Cambridge University Press, 2010.doi:10.1017/CBO9780511976667

    doi:10.1017/cbo9780511976667 2010

  27. [27]

    A quantum approximate optimization al- gorithm.arXiv preprint arXiv:1411.4028, 2014

    Edward Farhi, Jeffrey Goldstone, and Sam Gut- mann. A quantum approximate optimization al- gorithm.arXiv preprint arXiv:1411.4028, 2014. arXiv:1411.4028

    Pith/arXiv arXiv 2014

  28. [28]

    Egger, Jakub Mareˇ cek, and Stefan Woerner

    Daniel J. Egger, Jakub Mareˇ cek, and Stefan Woerner. Warm-starting quantum optimization. Quantum, 5:479, June 2021.doi:10.22331/q-2 021-06-17-479

    work page doi:10.22331/q-2 2021

  29. [29]

    Barron, F

    Linghua Zhu, Ho Lun Tang, George S. Barron, F. A. Calderon-Vargas, Nicholas J. Mayhall, Ed- win Barnes, and Sophia E. Economou. Adap- tive quantum approximate optimization algorithm for solving combinatorial problems on a quantum computer.Phys. Rev. Res., 4:033029, Jul 2022. doi:10.1103/PhysRevResearch.4.033029

    doi:10.1103/physrevresearch.4.033029 2022

  30. [30]

    Rieffel, Davide Venturelli, and Rupak Biswas

    Stuart Hadfield, Zhihui Wang, Bryan O’Gorman, Eleanor G. Rieffel, Davide Venturelli, and Rupak Biswas. From the quantum approximate optimiza- tion algorithm to a quantum alternating operator ansatz.Algorithms, 12(2), 2019.doi:10.3390/a1 2020034

    doi:10.3390/a1 2019

  31. [31]

    The quantum approximate optimization algorithm at high depth for maxcut on large-girth regular graphs and the sherrington-kirkpatrick model

    Joao Basso, Edward Farhi, Kunal Marwaha, Ben- jamin Villalonga, and Leo Zhou. The quantum approximate optimization algorithm at high depth for maxcut on large-girth regular graphs and the sherrington-kirkpatrick model. Schloss Dagstuhl 11 – Leibniz-Zentrum f¨ ur Informatik, 2022.doi: 10.4230/LIPICS.TQC.2022.7

    doi:10.4230/lipics.tqc.2022.7 2022

  32. [32]

    Parameter transfer for quantum approximate optimization of weighted maxcut.ACM Transactions on Quantum Computing, 4, 02 2023.doi:10.1145/3584706

    Ruslan Shaydulin, Phillip Lotshaw, Jeffrey Larson, James Ostrowski, and Travis Humble. Parameter transfer for quantum approximate optimization of weighted maxcut.ACM Transactions on Quantum Computing, 4, 02 2023.doi:10.1145/3584706

    doi:10.1145/3584706 2023

  33. [33]

    Parameter Setting in Quantum Approximate Optimization of Weighted Problems.Quantum, 8:1231, January 2024.doi: 10.22331/q-2024-01-18-1231

    Shree Hari Sureshbabu, Dylan Herman, Ruslan Shaydulin, Joao Basso, Shouvanik Chakrabarti, Yue Sun, and Marco Pistoia. Parameter Setting in Quantum Approximate Optimization of Weighted Problems.Quantum, 8:1231, January 2024.doi: 10.22331/q-2024-01-18-1231

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

  34. [34]

    Quantum supremacy using a programmable superconducting processor.Nature, 574(7779):505–510, Oct 2019.doi:10.1038/s415 86-019-1666-5

    Frank Arute, Kunal Arya, Ryan Babbush, Dave Bacon, and others. Quantum supremacy using a programmable superconducting processor.Nature, 574(7779):505–510, Oct 2019.doi:10.1038/s415 86-019-1666-5

    doi:10.1038/s415 2019

  35. [35]

    Quantum compu- tational advantage via 60-qubit 24-cycle random circuit sampling, 2021.arXiv:2109.03494

    Qingling Zhu, Sirui Cao, Fusheng Chen, Ming- Cheng Chen, Xiawei Chen, Tung-Hsun Chung, Hui Deng, Yajie Du, Daojin Fan, Ming Gong, Cheng Guo, Chu Guo, Shaojun Guo, Lianchen Han, Linyin Hong, and others. Quantum compu- tational advantage via 60-qubit 24-cycle random circuit sampling, 2021.arXiv:2109.03494

    Pith/arXiv arXiv 2021

  36. [36]

    Ocean software api refer- ence — samplers, 2025

    D–Wave Quantum Inc. Ocean software api refer- ence — samplers, 2025. General Ocean API ref- erence; see Simulated Annealing. URL:https: //docs.dwavequantum.com/en/latest/ocean/ api_ref_samplers/index.html

    2025

  37. [37]

    dwave–neal: Simulated an- nealing sampler (github repository), 2022

    D–Wave Quantum Inc. dwave–neal: Simulated an- nealing sampler (github repository), 2022. Ver- sion 0.6.0, released 2022-11-25. URL:https: //github.com/dwavesystems/dwave-neal

    2022

  38. [38]

    Abanin, Laleh Aghababaie-Beni, Igor Aleiner, others, Google Quantum AI, and Collaborators

    Rajeev Acharya, Dmitry A. Abanin, Laleh Aghababaie-Beni, Igor Aleiner, others, Google Quantum AI, and Collaborators. Quantum error correction below the sur- face code threshold.Nature, Dec 2024. doi:10.1038/s41586-024-08449-y

    doi:10.1038/s41586-024-08449-y 2024

  39. [39]

    Establishing a new benchmark in quantum computational advantage with 105-qubit Zuchongzhi 3.0 processor, 2024

    Dongxin Gao, Daojin Fan, Chen Zha, Jiahao Bei, Jian-Wei Pan, and others. Establishing a new benchmark in quantum computational advantage with 105-qubit Zuchongzhi 3.0 processor, 2024. arXiv:2412.11924

    Pith/arXiv arXiv 2024

  40. [40]

    Xuhao Zhu, Zuoheng Zou, Feitong Jin, Pavel Mosharev, Maolin Luo, Yaozu Wu, Jiachen Chen, Chuanyu Zhang, Yu Gao, Ning Wang, Yiren Zou, Aosai Zhang, Fanhao Shen, Zehang Bao, Zitian Zhu, Jiarun Zhong, Zhengyi Cui, Yihang Han, Yiyang He, Han Wang, Jia-Nan Yang, Yanzhe Wang, Jiayuan Shen, Gongyu Liu, Zixuan Song, Jinfeng Deng, Hang Dong, Pengfei Zhang, Chao So...

    arXiv 2025

  41. [41]

    The ibm quantum heavy hex lat- tice, 2021

    Paul Nation, Hanhee Paik, Andrew Cross, and Zaira Nazario. The ibm quantum heavy hex lat- tice, 2021. URL:https://www.ibm.com/quantu m/blog/heavy-hex-lattice

    2021

  42. [42]

    Quantum comput- ing dataset of maximum independent set problem on king lattice of over hundred rydberg atoms.Sci- entific Data, 11(1):111, Jan 2024.doi:10.1038/ s41597-024-02926-9

    Kangheun Kim, Minhyuk Kim, Juyoung Park, An- drew Byun, and Jaewook Ahn. Quantum comput- ing dataset of maximum independent set problem on king lattice of over hundred rydberg atoms.Sci- entific Data, 11(1):111, Jan 2024.doi:10.1038/ s41597-024-02926-9

    2024

  43. [43]

    Bunyk, Emile M

    Paul I. Bunyk, Emile M. Hoskinson, Mark W. Johnson, Elena Tolkacheva, Fabio Altomare, An- drew J. Berkley, Richard Harris, Jeremy P. Hilton, Trevor Lanting, Anthony J. Przybysz, and Jed Whittaker. Architectural considerations in the de- sign of a superconducting quantum annealing pro- cessor.IEEE Transactions on Applied Supercon- ductivity, 24(4):1–10, 20...

    work page doi:10.1109/tasc.2 2014

  44. [44]

    On the computational complexity of ising spin glass models.Journal of Physics A: Mathematical and General, 15(10):3241, oct 1982

    F Barahona. On the computational complexity of ising spin glass models.Journal of Physics A: Mathematical and General, 15(10):3241, oct 1982. doi:10.1088/0305-4470/15/10/028

    doi:10.1088/0305-4470/15/10/028 1982

  45. [45]

    force vector

    Kostas Blekos, Dean Brand, Andrea Ceschini, Chiao-Hui Chou, Rui-Hao Li, Komal Pandya, and Alessandro Summer. A review on quantum ap- proximate optimization algorithm and its variants. Physics Reports, 1068:1–66, 2024. A review on Quantum Approximate Optimization Algorithm and its variants.doi:10.1016/j.physrep.20 24.03.002. 12 ANoisy simulation details In...

    doi:10.1016/j.physrep.20 2024

  46. [46]

    Define cost and mixer HamiltoniansH C andH M that do not commute with each other. Cost Hamiltonian corresponds to Ising energy and is given by the following formula: HC = ∑ i,j Ji,jZiZj + ∑ i hiZi,(8) whereJ i,j andhi are Ising coupling matrix and linear shift fields,Zi are Pauli Z operators. Mixing Hamiltonian is a sum of Pauli X operators acting on each...

  47. [47]

    Initialize the circuit in equal superposition of all possible states by applying Hadamard gate to each qubit: |s⟩=|+⟩ ⊗n =H⊗n|0⟩⊗n.(9) 15

  48. [48]

    We use the rotation conventionR P (θ) :=e−iθP/2 for any Pauli productP

    Construct the circuit ansatz by defining the cost and mixer unitary operators: ˆUC(γ) =e−iγHC (10) ˆUM(β) =e−iβHM.(11) Here,γandβare variational parameters of the circuit. We use the rotation conventionR P (θ) :=e−iθP/2 for any Pauli productP. Each term in cost and mixer operator can be implemented with rotation gates as follows: e−iγJi,jZiZj =R ZiZj(2γJi...

  49. [49]

    With a given circuit depthp≥1, define parametersγ= (γ 1,...,γ p);β= (β 1,...,β p)and apply the full circuit to obtain the final quantum state: |Ψp(γ,β)⟩= ˆUM(βp) ˆUC(γp)... ˆUM(β1) ˆUC(γ1)|s⟩.(15) Original QAOA is a variational algorithm, which means that after construction of the quantum state|Ψ p(γ,β)⟩, the expectation value of the cost HamiltonianH C w...