betaFeedback

WorksPhysics and Astronomy

Trainability-Simulability Tradeoff in Parameterized IQP-based Reinforcement Learning Policies

Abstract

Parameterized instantaneous quantum polynomial-time (IQP) circuits generate distributions conjectured to be hard to sample classically. As reinforcement learning (RL) policies, their 2n2^n measurement outcomes are typically mapped to a smaller action space by a classical decoder. This aggregation can mitigate initialization concentration while making the policy classically simulable. We characterize this tradeoff through the decoder's Fourier–Pauli expansion. Under independent uniform initialization, we express policy variance as an exact sum over groups of nonconstant Pauli components with identical active-generator signatures. Each group contributes its squared aggregate coefficient times 2−r2^{-r}, where rr is the binary rank of its active generators. An inverse-polynomial aggregate coefficient with logarithmic rank prevents exponential policy concentration. Sparse Pauli expansions and contiguous action bins permit efficient classical additive estimation of action probabilities. We introduce a majority decoder combining protected low-weight components with exponentially large exact Fourier support, preserving initialization nonconcentration while excluding polynomial-size enumeration of the exact expansion. Numerical experiments corroborate the variance laws. In ten-qubit CartPole experiments, majority achieves higher returns than recursive parity and the tested local readouts, while higher-weight Pauli sectors contribute an increasing relative share of gradient magnitude during training. A complementary 21-qubit block-chain policy outperforms an extensive pool of fully connected classical neural networks in episode efficiency, even though its bounded-width structure permits efficient classical simulation. These results guide decoder design for quantum and quantum-inspired RL, distinguishing policy nonconcentration, learning performance, and classical simulability.

Introduction

Reinforcement learning (RL) provides a framework for sequential decision-making in which an agent learns through interaction with an environment. Rather than receiving a prescribed action for each observation, the agent seeks a policy that maximizes cumulative reward, accounting for the consequences of its decisions over time. This formulation makes RL relevant to problems in control, robotics, and resource allocation, where actions affect both future observations and the opportunities for obtaining reward [1]. The policy representation is therefore a central modeling choice: it determines which decision rules can be expressed and how they can be learned from experience.

Parameterized quantum circuits (PQCs) offer a class of trainable models for representing functions and probability distributions. Their adjustable gates transform an encoded input into a quantum state whose measurement statistics define the model output. Suitably constructed PQC families possess universal function-approximation or distribution-generation capabilities [2, 3], motivating their investigation as learning architectures. In RL, this construction yields a stochastic policy: the environment observation is encoded in a circuit, an action is obtained from its measurement outcomes, and the circuit parameters are optimized using the rewards collected by the agent [4, 5, 6, 7].

Instantaneous quantum polynomial-time (IQP) circuits provide a useful test case. Their full output distributions can be hard to sample classically under standard complexity assumptions [8, 9, 10], although selected expectation values admit efficient randomized classical estimation [11]. This distinction supports classically trained IQP generative models [12, 13]. It also creates a potential limitation for RL: coarse-graining a difficult full-register sampling problem can produce an easily simulated action distribution.

Connecting a quantum model to an environment requires specifying how its measurement outcomes represent actions. An nn-qubit circuit has 2n2^n computational-basis outcomes, whereas the environment may offer only A≪2nA \ll2^n discrete actions. Increasing the circuit width to enrich the model does not require a corresponding increase in the action space. A classical decoder therefore maps measured bit strings to actions, assigning multiple outcomes to the same action [6, 14]. The resulting policy probability is the sum of the Born probabilities of all strings assigned to that action.

This probability aggregation is more than an interface between the circuit and the environment: it changes the quantities that must be optimized and simulated. Each decoder defines action projectors with specific Pauli components. Their structure can mitigate initialization concentration, since observable locality plays a central role in barren-plateau mechanisms [15, 16], including those of quantum policies [17]. Yet structures that protect optimization signals can also make the model classically accessible [18]. This raises the central question of this work: how does the classical decoder jointly determine policy initialization variance and the cost of classical access to action probabilities?

The relevant access model must be specified. At deployment, sampling a measurement string and decoding it produces an action. Likelihood-based training additionally requires action probabilities: REINFORCE uses their logarithmic derivatives, and proximal policy optimization (PPO) uses probability ratios [19], [20]. Efficient additive estimation of all A=poly⁡(n)A = \operatorname{poly}(n) action probabilities therefore enables approximate classical action sampling, but does not by itself guarantee efficient relative-error or score-function estimation when an action probability is very small. Our simulation claims concern the stated additive accuracy and sampling task, not an unqualified efficient implementation of every policy-gradient update.

Our analysis uses the Walsh–Pauli expansion of each action projector. For an individual diagonal Pauli observable under independent uniform IQP angles, its variance is 2−ru2^{-r^{u}}, where rur^{u} is the binary rank of the active generator supports. This component identity also appears in Theorem 3.8 of Shen et al. [21], who study IQP Born-machine trainability; we give a self-contained, independently obtained proof. Their graph examples likewise connect rank to graph neighborhoods. The policy-level identity additionally requires vanishing covariances between distinct Pauli expectations. We establish this whenever the generator supports span F2n\mathbb{F}_{2}^{n}; independent local rotations on every qubit are sufficient but not necessary. For arbitrary generator sets, we derive an exact grouped variance law that accounts for coincident signatures and coefficient cancellation. We use these identities to relate decoder coefficients to policy concentration and classical estimation.

For repeated IQP-like circuits, the variance decomposition and propagated-Pauli bounds are applications of the framework of Letcher, Woerner, and Zoufal [22], not new general variance bounds. Their prescribed initial local layers and independent initialization are essential. Alternating commuting blocks remove global commutativity; their analysis nevertheless provides decoder-dependent nonconcentration certificates. The contribution here is to combine these ingredients with explicit action decoding, neural encoding, and simulation criteria, rather than to claim either the component rank identity or general Pauli-variance additivity as a wholly new principle.

Contributions.

  1. Decoder-dependent initialization certificates. We assemble exact IQP policy-variance laws and repeated-IQP lower bounds under explicit architecture and initialization assumptions. The individual-term IQP sum holds for full-rank generator support lists; otherwise, equal-signature terms must be grouped before squaring. A sufficiently weighted logarithmic-rank component, or a nonzero grouped coefficient, prevents exponential policy concentration. For graph IQPs with all independent local rotations, its rank is its structural backward light-cone size; other components may have extensive light cones. Neural encodings preserve these results at each fixed observation under the stated independent-offset assumptions.

  1. Action-level classical simulation. We specialize expectation-value estimation to efficiently accessible polynomial-size Pauli expansions and to contiguous action intervals, including non-power-of-two action counts. Together with sparse-postprocessing results [23], these criteria show why standard decoders can protect variance while making the coarse-grained policy classically accessible.

  1. Protected components with dense exact support. Majority–modulo decoding combines inverse-polynomial low-weight coefficients with exponentially large exact Pauli support in the proved block-size regime. With a suitable low-rank component, it retains nonconcentration while excluding direct polynomial-size enumeration of the exact expansion. This is a representation-level design principle, not a classical hardness result; aggregate and approximate simulation remain separate questions.

  1. Mechanism tests and a quantum-inspired policy. Initialization experiments through 16 qubits test the variance laws. Controlled CartPole studies compare majority with parity, valid local readouts, and MLPs; gradient-sector diagnostics measure relative participation rather than causal performance contributions. A connected, classically contractible 21-qubit policy achieves higher episode-based AUC than the tested MLPs under its original initialization, but this ordering reverses in the uniform-initialization follow-up. These results separate nonconcentration, initialization bias, learning performance, and classical runtime.

Organization. Section II defines the circuits, policy interface, and concentration criterion. Section III collects the main statements; Sections IV–VII develop the decoder-specific variance and simulation results. Section VIII presents the principal experiments and Section IX discusses their scope. Proofs, secondary CartPole studies, detailed gradient diagnostics, and circuit diagrams are provided in the appendices.

Background

Parameterized IQP circuits

An nn-qubit parameterized IQP circuit is a polynomial-size product of commuting XX-type Pauli rotations, also called an XX program [24]:

UX(θ)=∏ℓ=1me−iθℓPℓ/2,(1)U_X(\theta) = \prod_{\ell=1}^{m} e^{-i\theta_{\ell}P_{\ell}/2}, \tag*{(1)}

where mm is polynomial in nn and the Pℓ∈{I,X}⊗nP_{\ell} \in\{I,X\}^{\otimes n} commute. The equivalent diagonal ZZ program is DZ(θ):=H⊗nUX(θ)H⊗nD_Z(\theta) := H^{\otimes n}U_X(\theta)H^{\otimes n}, so UX=H⊗nDZH⊗nU_X = H^{\otimes n}D_ZH^{\otimes n}. Unless stated otherwise, the input is ∣0n⟩|0^n\rangle and the parameters are initialized

independently as θℓ∼Unif⁡[−π,π]\theta_\ell\sim\operatorname{Unif}[-\pi,\pi]. For the binary-rank analysis in Sec. IV, we write Pℓ=XgℓP_\ell= X^{g_\ell}, with gℓg_\ell its binary support vector. This algebraic restriction does not make the full IQP output distribution generically easy to sample classically [8], [9].

Most results use a graph G=(V,E)G=(V,E) to select one- and two-qubit generators:

UG(α,β)=∏i∈Ve−iαiXi/2∏(i,j)∈Ee−iβijXiXj/2=H⊗nDG(α,β)H⊗n,(2)U_G(\alpha,\beta)=\prod_{i\in V}e^{-i\alpha_iX_i/2}\prod_{(i,j)\in E}e^{-i\beta_{ij}X_iX_j/2} =H^{\otimes n}D_G(\alpha,\beta)H^{\otimes n}, \tag*{(2)}

where DGD_G is obtained by replacing every XX in the first line by ZZ. Figure 1 shows how vertices select one-qubit rotations and edges select two-qubit rotations.

IQP and repeated IQP-like Pauli architectures

Figure 1. IQP and repeated IQP-like Pauli architectures. (a) Vertices of an interaction graph select one-qubit ZZ phases and edges select two-qubit ZZZZ phases. (b) The resulting commuting graph-IQP block between Hadamards, equivalently the X/XXX/XX circuit of Eq. (2). (c) Native repeated form, with Hadamards interleaving polynomial-size diagonal ZZ programs. (d) Equivalent alternating XX- and ZZ-type Pauli layers; their generators may be higher-body and need not define a graph. The shaded DX(0)D_X^{(0)} auxiliary layer is used in the repeated trainability bounds, and setting its angles to zero recovers the standard repeated-IQP architecture. An IQP circuit obeys the exact rank law, which for a graph IQP reduces to backward-light-cone size. (e)–(f) illustrate this reduction for Zu=Z3Z^u = Z_3 on a five-qubit chain: propagation reaches qubits 2, 3, and 4, so ΔGθ(u)=3\Delta_G^\theta(u) = 3. In (b)–(d) and (f), the green meters denote computational-basis measurement; the single meter in (f) measures only qubit 3.

Repeated copies of the same XX program do not enlarge the circuit class: because all generators commute, UX(θ)L=UX(Lθ)U_X(\theta)^L=U_X(L\theta) (after collecting the angles), and the result remains an IQP circuit. The useful repeated construction instead uses the ZZ-program realization with Hadamards between diagonal layers. Those basis changes conjugate alternating layers into commuting XX-type and ZZ-type Pauli programs, which do not commute with one another and therefore increase expressivity. The generators within a layer need not be graph edges. For polynomial-size binary support lists srℓs_{r\ell} and trℓt_{r\ell}, write

DZ(r)=∏ℓe−iθrℓZsrℓ/2,DX(r)=∏ℓe−iφrℓXtrℓ/2.(3)D_Z^{(r)}=\prod_\ell e^{-i\theta_{r\ell}Z^{s_{r\ell}}/2},\qquad D_X^{(r)}=\prod_\ell e^{-i\varphi_{r\ell}X^{t_{r\ell}}/2}. \tag*{(3)}

Every factor inside either product commutes with the other factors in that layer. The supports may have arbitrary size, subject only to a polynomial total number of rotations. For example, a layer may contain all one-, two-, and three-qubit strings,

DZ(r)=∏iRZi∏i<jRZiZj∏i<j<kRZiZjZk,(4)D_Z^{(r)}=\prod_i R_{Z_i}\prod_{i<j}R_{Z_iZ_j}\prod_{i<j<k}R_{Z_iZ_jZ_k}, \tag*{(4)}

with independent angles suppressed for readability; this example has O(n3)O(n^3) rotations and is not specified by a pairwise interaction graph. In this work we consider the repeated IQP-like architecture

∣ψR⟩=DX(R)DZ(R)⋯DX(1)DZ(1)DX(0)∣0n⟩,(5)|\psi_R\rangle=D_X^{(R)}D_Z^{(R)}\cdots D_X^{(1)}D_Z^{(1)}D_X^{(0)}|0^n\rangle, \tag*{(5)}

where DX(0)=⨂ie−iϕiXi/2D_X^{(0)}=\bigotimes_i e^{-i\phi_iX_i/2} is an auxiliary local layer. Setting ϕ=0\phi=0 makes this layer the identity and recovers the standard repeated-IQP circuit, so the augmentation does not reduce expressivity. Its role is to supply, together with the local rotations in DZ(1)D_Z^{(1)}, the two orthogonal randomized layers required for the Pauli-light-cone bounds of Ref. [22]. We therefore assume that DZ(r)D_Z^{(r)} contains a trainable ZiZ_i rotation on every qubit; all remaining generators in every layer may be arbitrary Pauli strings of the indicated axis. Each DX(r)D_X^{(r)} or DZ(r)D_Z^{(r)} is internally commuting, but the complete Hadamard-interleaved alternating circuit is no longer an IQP circuit. The family is more expressive and, for R=poly⁡(n)R=\operatorname{poly}(n), contains universal alternating dynamics [25]. Figure 1 shows the diagonal and Pauli-rotation views used below.

Weak simulation of IQP output distributions

For a fixed circuit instance, let

pθ(x):=∣⟨x∣UX(θ)∣0n⟩∣2(6)p_\theta(x):=\left|\langle x|U_X(\theta)|0^n\rangle\right|^2 \tag*{(6)}

denote the complete nn-bit Born distribution. When the angles depend on an observation ss, every statement below applies pointwise at fixed ss. A classical weak simulator is a randomized classical algorithm that produces samples from some distribution qθq_\theta whose total-variation (TV) distance from pθp_\theta satisfies

dTV(pθ,qθ):=12∑x∈{0,1}n∣pθ(x)−qθ(x)∣≤ϵ(7)d_{\mathrm{TV}}(p_\theta,q_\theta):=\frac{1}{2}\sum_{x\in\{0,1\}^n}|p_\theta(x)-q_\theta(x)|\le\epsilon \tag*{(7)}

in time polynomial in nn and ϵ−1\epsilon^{-1}. The task of interest is sampling: generating outcomes, rather than calculating their individual probabilities. Multiplicative-error sampling instead requires pθ(x)/c≤qθ(x)≤cpθ(x)p_\theta(x)/c\le q_\theta(x)\le cp_\theta(x) for every outcome, for a fixed factor c≥1c\ge1. Efficient classical IQP sampling with sufficiently small multiplicative error is excluded unless the polynomial hierarchy (PH) collapses [8]. Such pointwise guarantees are also generally unrealistic for noisy quantum devices, since they constrain exponentially rare outcomes and require exact preservation of zeros; this is not an in-principle obstruction for an ideal quantum device. Additive-error sampling is therefore the physically relevant setting for a quantum–classical separation. Under the standard average-case hardness conjectures, an efficient classical sampler for general IQP circuits at sufficiently small constant TV error would likewise collapse PH [9, 26].

The policy studied in this work is the pushforward of pθp_\theta through a decoder ff. For Ba={x:f(x)=a}B_a=\{x:f(x)=a\},

πθ(a)=∑x∈Bapθ(x)(8)\pi_\theta(a)=\sum_{x\in B_a}p_\theta(x) \tag*{(8)}

A quantum device can sample this action distribution by drawing one bit string and returning f(x)f(x). We are interested in whether this decoded action-sampling task retains a classical simulation obstacle. Likelihood-based training additionally requires numerical access to the coarse-grained probability. Our classical-simulability results concern additive action-probability estimation: given a,ϵa,\epsilon, and δ\delta, the algorithm must output π^(a)\widehat{\pi}(a) such that

Pr⁡[∣π^(a)−πθ(a)∣≤ϵ]≥1−δ(9)\Pr[|\widehat{\pi}(a)-\pi_\theta(a)|\le\epsilon]\ge1-\delta \tag*{(9)}

using resources polynomial in nn, ϵ−1\epsilon^{-1}, and log⁡δ−1\log\delta^{-1}. This task is distinct from weak simulation of the full output distribution: full-distribution sampling hardness neither proves nor rules out efficient additive estimation after the decoder has aggregated many basis states.

Additive estimation of IQP observables and policies

The weak-sampling obstruction above does not imply that every statistic of an IQP output distribution is hard to compute. In particular, any fixed diagonal Pauli expectation admits an efficient randomized additive estimator implied by Theorem 3 of Ref. [11], stated below for completeness.

Proposition 1 (Van den Nest IQP expectation-value estimator [11]). Let UX(θ)U_X(\theta) be a polynomial-size parameterized IQP circuit and let ZuZ^u be any diagonal Pauli word. For every ϵ>0\epsilon> 0 with ϵ−1=poly⁡(n)\epsilon^{-1} = \operatorname{poly}(n), there exists a classical algorithm producing an unbiased estimator L^u\hat{L}_u of

Lu(θ)=⟨0n∣UX†(θ)ZuUX(θ)∣0n⟩(10)L_u(\theta) = \langle0^n | U_X^\dagger(\theta) Z^u U_X(\theta) | 0^n \rangle \tag*{(10)}

such that Std⁡[L^u]<ϵ\operatorname{Std}[\hat{L}_u] < \epsilon, using poly⁡(n)\operatorname{poly}(n) time and space.

A self-contained derivation is given in Appendix § 15-§ 1. The same expectation can be expressed in the Heisenberg picture by propagating ZuZ^u backwards through the circuit and evaluating UX†ZuUXU_X^\dagger Z^u U_X on ∣0n⟩|0^n\rangle. The support reached by this evolution defines the observable’s backward light cone. For IQP circuits, however, mutual commutativity allows every gate that commutes with the original ZuZ^u to cancel from the conjugation, irrespective of its position in the circuit. Thus newly reached qubits do not activate further interactions as they can in noncommuting circuits. For local-and-edge graph-IQP models, support can extend only across edges crossing the original observable’s support boundary, motivating the following definition.

References

  1. [1]R. S. Sutton and A. G. Barto, Reinforcement Learning: An Introduction, 2nd ed., ser. Adaptive Computation and Machine Learning series. Cambridge, Massachusetts: The MIT Press, 2018.
  2. [2]Díaz, and J. I. Latorre, “One qubit as a universal approximant,” Physical Review A, vol. 104, p. 012405, 2021.
  3. [3]A. Barthe, M. Grossi, S. Vallecorso, J. Tura, and V. Dunjko, “Parameterized quantum circuits as universal generative models for continuous multivariate distributions,” npj Quantum Information, vol. 11, p. 121, 2025. [Online]. Available: https://arxiv.org/abs/2402.09848
  4. [4]S. Y.-C. Chen, C.-H. H. Yang, J. Qi, P.-Y. Chen, X. Ma, and H.-S. Goan, “Variational quantum circuits for deep reinforcement learning,” IEEE Access, vol. 8, pp. 141 007–141 024, 2020.arxiv.org/abs/1907.00397
  5. [5]A. Skolik, S. Jerbi, and V. Dunjko, “Quantum agents in the gym: A variational quantum algorithm for deep Q-learning,” Quantum, vol. 6, p. 720, 2022.DOI
  6. [6]S. Jerbi, C. Gyurik, S. C. Marshall, H. J. Briegel, and V. Dunjko, “Parametrized quantum policies for reinforcement learning,” in Advances in Neural Information Processing Systems, vol. 34, 2021, pp. 28 362–28 375.arxiv.org/abs/2103.05577
  7. [7]A. Sequeira, L. P. Santos, and L. S. Barbosa, “Policy gradients using variational quantum circuits,” Quantum Machine Intelli- gence, vol. 5, p. 18, 2023.arxiv.org/abs/2203.10591
  8. [8]M. J. Bremner, R. Jozsa, and D. J. Shepherd, “Classical simulation of commuting quantum computations implies collapse of the polynomial hierarchy,” Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 467, no. 2126, pp. 459–472, 2011.DOI
  9. [9]M. J. Bremner, A. Montanaro, and D. J. Shepherd, “Average- case complexity versus approximate simulation of commuting quantum computations,” Physical Review Letters, vol. 117, p. 080501, 2016.DOI
  10. [10]——, “Achieving quantum supremacy with sparse and noisy commuting quantum computations,” Quantum, vol. 1, p. 8, 2017.DOI
  11. [11]M. Van den Nest, “Simulating quantum computers with prob- abilistic methods,” Quantum Information and Computation, vol. 11, no. 9–10, pp. 784–812, 2011.
  12. [12]E. Recio-Armengol and J. Bowles, “IQPot: Fast optimization of instantaneous quantum polynomial circuits in JAX,” 2025.
  13. [13]E. Recio-Armengol, S. Ahmed, and J. Bowles, “Train on classi- cal, deploy on quantum: Scaling generative quantum machine learning to a thousand qubits,” 2025.
  14. [14]N. Meyer, D. D. Scherer, A. Plinge, C. Mutschler, and M. J. Hartmann, “Quantum policy gradient algorithm with optimized action decoding,” in Proceedings of the 40th International Conference on Machine Learning, ser. Proceedings of Machine Learning Research, vol. 202. PMLR, 2023, pp. 24 592–24 613.arxiv.org/abs/2212.06663
  15. [15]J. R. McClean, S. Boixo, V. N. Smelyanskiy, R. Babbush, and H. Neven, “Barren plateaus in quantum neural network training landscapes,” Nature Communications, vol. 9, p. 4812, 2018.DOI
  16. [16]M. Cerezo, A. Sone, T. Volkoff, L. Cincio, and P. J. Coles, “Cost function dependent barren plateaus in shallow parametrized quantum circuits,” Nature Communications, vol. 12, p. 1791, 2021.DOI
  17. [17]A. Sequeira, L. P. Santos, and L. S. Barbosa, “Trainability issues in quantum policy gradients,” Machine Learning: Science and Technology, vol. 5, no. 3, p. 035037, 2024.arxiv.org/abs/2406.09614
  18. [18]M. Cerezo, M. Larocca, D. García-Martín, N. L. Díaz, P. Braccia, E. Fontana, M. S. Rudolph, H. Bermejo, A. Ijaz, S. Thanaslip, E. R. Anschuetz, and Z. Holmes, “Does provable absence of barren plateaus imply classical simulability?” Nature Communi- cations, vol. 16, p. 7907, 2025.DOI
  19. [19]R. J. Williams, “Simple statistical gradient-following algorithms for connectionist reinforcement learning,” Machine Learning, vol. 8, pp. 229–256, 1992.
  20. [20]J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, “Proximal policy optimization algorithms,” 2017.arxiv.org/abs/1707.06347
  21. [21]K. Shen, S. Pielawa, V. Dunjko, and H. Wang, “Characterizing trainability of instantaneous quantum polynomial circuit born machines,” 2026.arxiv.org/abs/2602.11042
  22. [22]A. Letcher, S. Woerner, and C. Zoufal, “Tight and efficient gradient bounds for parameterized quantum circuits,” Quantum, vol. 8, p. 1484, 2024.DOI
  23. [23]Y. Takahashi, M. Miyamoto, and N. Kunihiro, “Classical sim- ulability of quantum circuits followed by sparse classical post- processing,” 2026.arxiv.org/abs/2603.05920
  24. [24]D. Shepherd and M. J. Bremner, “Instantaneous quantum computation,” Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 465, no. 2105, pp. 1413– 1439, 2009.
  25. [25]S. Lloyd, “Quantum approximate optimization is computation- ally universal,” 2018.arxiv.org/abs/1812.11075
  26. [26]S. C. Marshall, S. Aaronson, and V. Dunjko, “Improved separation between quantum and classical computers for sampling and functional tasks,” arXiv preprint arXiv:2410.20935, 2024. [Online]. Available: https://arxiv.org/abs/2410.20935
  27. [27]I. L. Markov and Y. Shi, “Simulating quantum computation by contracting tensor networks,” SIAM Journal on Computing, vol. 38, no. 3, pp. 963–981, 2008.DOI
  28. [28]U. Schollwöck, “The density-matrix renormalization group in the age of matrix product states,” Annals of Physics, vol. 326, pp. 96–192, 2011.DOI
  29. [29]A. J. Ferris and G. Vidal, “Perfect sampling with unitary tensor networks,” Physical Review B, vol. 85, p. 165146, 2012.DOI
  30. [30]A. Arrasmith, Z. Holmes, M. Cerezo, and P. J. Coles, “Equivalence of quantum barren plateaus to cost concentration and narrow gorges,” Quantum Science and Technology, vol. 7, p. 045015, 2022.DOI
  31. [31]A. Pérez-Salinas, A. Cervera-Lierta, E. Gil-Fuster, and J. I. Latorre, “Data re-uploading for a universal quantum classifier,” Quantum, vol. 4, p. 226, 2020.
  32. [32]R. O’Donnell, Analysis of Boolean Functions. Cambridge University Press, 2014.
  33. [33]M. Schwarz and M. Van den Nest, “Simulating quantum circuits with sparse output distributions,” 2013.arxiv.org/abs/1310.6749
  34. [34]S. Bravyi, D. Browne, P. Calpin, E. Campbell, D. Gosset, and M. Howard, “Simulation of quantum circuits by low-rank stabilizer decompositions,” Quantum, vol. 3, p. 181, 2019.
  35. [35]J. Rajakumar, J. D. Watson, and Y.-K. Liu, “Polynomial-time classical simulation of noisy IQP circuits with constant depth,” in Proceedings of the 2025 Annual ACM-SIAM Symposium on Discrete Algorithms (SODA), 2025, pp. 1037–1056.DOI
  36. [36]E. R. Anschuetz, A. Bauer, B. T. Kiani, and S. Lloyd, “Efficient classical algorithms for simulating symmetric quantum systems,” 2023.

Paper details

vlnc:W_xdmbnr5z
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents