arxiv: 2605.03932 · v1 · submitted 2026-05-05 · 🪐 quant-ph · cs.AI

Recognition: unknown

Magic-Informed Quantum Architecture Search

Vincenzo Lipardi , Domenica Dibenedetto , Georgios Stamoulis , Mark H.M. Winands

Authors on Pith no claims yet

Pith reviewed 2026-05-07 16:29 UTC · model grok-4.3

classification 🪐 quant-ph cs.AI

keywords quantum architecture searchnonstabilizernessmagicgraph neural networkMonte Carlo tree searchquantum circuitsquantum resources

0 comments

The pith

A graph neural network can bias Monte Carlo tree search to control nonstabilizerness in quantum circuits while improving solution quality.

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

The paper proposes a way to steer the automated design of quantum circuits toward specific levels of nonstabilizerness, also called magic. It combines Monte Carlo tree search with a graph neural network that predicts how much magic a partial circuit contains. The network's estimates create a bias that pushes the search toward either high-magic or low-magic circuits depending on the goal. Tests on finding ground-state energies and approximating arbitrary quantum states show that the final circuits exhibit the intended magic levels and deliver better results than unbiased searches. The approach works even when the network encounters circuit sizes or structures it was not trained on.

Core claim

The magic-informed quantum architecture search technique uses a graph neural network to estimate the nonstabilizerness of candidate circuits and thereby biases Monte Carlo tree search toward high- or low-magic regimes; experiments demonstrate that this bias successfully modulates magic content throughout the search tree and in the final circuit, producing consistent gains in solution quality on both structured ground-state energy problems and general quantum state approximation tasks across different sizes and target magic values, including out-of-distribution cases.

What carries the argument

Graph Neural Network that estimates the magic of candidate quantum circuits and supplies a bias term to Monte Carlo Tree Search within the quantum architecture search loop.

If this is right

The search tree and final circuit can be steered to either high- or low-magic regimes as desired.
Solution quality improves on both ground-state energy minimization and quantum state approximation.
The bias remains effective even when the GNN processes out-of-distribution circuit instances.
The same framework applies across different problem sizes and target magic values.

Where Pith is reading between the lines

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

Resource-aware biases grounded in fundamental quantum properties may improve heuristic search in circuit design more broadly.
The method could be extended to bias searches according to other resources such as entanglement or coherence.
If the GNN estimates remain reliable at larger scales, the technique might help automate discovery of circuits that achieve quantum advantage with controlled resource costs.
Problem-agnostic resource estimates appear compatible with task-specific optimization rather than inherently limiting it.

Load-bearing premise

The graph neural network supplies sufficiently accurate magic estimates to guide the search productively without the added problem-agnostic bias restricting exploration enough to reduce solution quality on the target tasks.

What would settle it

If circuits found under high-magic bias do not show measurably higher nonstabilizerness than those found under low-magic bias, or if solution quality (energy error or approximation fidelity) is no better than standard unbiased search on the same problems, the central claim fails.

Figures

Figures reproduced from arXiv: 2605.03932 by Domenica Dibenedetto, Georgios Stamoulis, Mark H.M. Winands, Vincenzo Lipardi.

**Figure 2.** Figure 2: Generation procedure of the target states. For each view at source ↗

**Figure 3.** Figure 3: Average GNN estimations Mˆ 2 in the search tree. isolate the independent contributions of the two magic-based bias strategies. In the figure, the horizontal black lines in the boxplots indicate the median values, while the values in percentage correspond to the relative change in the mean Mˆ 2 with respect to the baseline. Overall, the magic-informed PWMCTS significantly shifts the distribution of Mˆ 2, wi… view at source ↗

**Figure 4.** Figure 4: H2. Energy achieved by each PWMCTS variant. The analysis of the performance based on the solution quality is illustrated in Figures 4, 5, and 6 for the molecules of H2, H2O, and LiH, respectively. Orange boxplots correspond to the energies achieved by the circuits designed by the gradient-free PWMCTS variants before the parameter view at source ↗

**Figure 5.** Figure 5: H2O. Energy achieved by each PWMCTS variant. The experiments on the simplest molecule H2 show that all variants manage to converge to the SCF solution. However, the magic-informed PWMCTS converge more reliably in the gradient-free phase to the solution compared to the baseline, and the combination of magic PW and magic UCT (all-in one) results in a systematic lower energy approaching to the FCI solution. T… view at source ↗

**Figure 7.** Figure 7: Exact M2 values and trace of the QFI matrix of the PQCs designed by each gradient-free PWMCTS variant. We further evaluate the magic-informed PWMCTS based on the solution quality, measured in terms of fidelity view at source ↗

**Figure 8.** Figure 8: Ground-state energy problem. Exact M2 values and trace of the QFI matrix of the PQCs designed by each gradient-free PWMCTS variant. result obtained by the gradient-free variants, while in dashed there is the improvement provided by finetuning the circuit parameters using the Adam optimizer. We observe that the M2 value of the target circuits significantly affects the difficulty of the PQC design problem. I… view at source ↗

**Figure 9.** Figure 9: Quantum state approximation. The bars report the median values of the fidelity achieved by each PWMCTS variant view at source ↗

read the original abstract

Nonstabilizerness, commonly referred to as magic, is a fundamental resource underpinning quantum advantage. In this paper, we propose a magic-informed quantum architecture search (QAS) technique that enables control over a quantum resource within the general framework of circuit design. Inspired by the AlphaGo approach, we tackle the problem with a Monte Carlo Tree Search technique equipped with a Graph Neural Network (GNN) that estimates the magic of candidate quantum circuits. The GNN model induces a magic-based bias that steers the search toward either high- or low-magic regimes, depending on the target objective. We benchmark the proposed magic-informed QAS technique on both the structured ground-state energy problem and on the more general quantum state approximation problem, spanning different sizes and target magic levels. Experimental results show that the proposed technique effectively influences the magic across the search tree and notably also on the resulting final circuit, even in regimes where the GNN operates on out-of-distribution instances. Although introducing a problem-agnostic magic bias could, in principle, constrain the search dynamics, we observe consistent improvements in solution quality across all problems tested.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper adds a GNN-based magic estimator to MCTS-driven quantum architecture search and shows it can steer final circuits toward target magic levels while improving solution quality on two benchmarks.

read the letter

The main thing here is a practical way to bake magic control into automated circuit design. They train a GNN to score the magic of candidate circuits during Monte Carlo tree search, then use that score to bias the search toward high-magic or low-magic regimes depending on the goal. On the ground-state energy and quantum state approximation tasks, the method shifts the magic distribution in the search tree and in the final circuits, and it does so even when the GNN sees out-of-distribution examples. Solution quality also improves across the board despite the added bias being problem-agnostic. That combination of MCTS with an explicit magic estimator is the genuinely new piece; most prior QAS work optimizes for depth or gate count, not nonstabilizerness as a resource. The experiments cover different system sizes and target magic values, which gives the results some breadth. The soft spot is the missing validation on the GNN itself. The abstract claims the bias works on OOD cases, but there are no reported error metrics, rank correlations, or comparisons against exact magic calculations on circuits sampled the same way the search produces them. Without that, it is possible the steering is driven by correlated features such as depth or specific gate patterns rather than true magic. The improvements are described as consistent, yet the abstract gives no baselines, effect sizes, or statistical tests, so the practical gain from the magic term is hard to quantify. This is aimed at people working on variational quantum algorithms and resource-aware circuit optimization. A reader who already uses MCTS or GNNs for QAS will see a clear extension they can test. The work is coherent enough and the claims are specific enough to deserve peer review; referees can ask for the GNN accuracy numbers and proper baselines, which would strengthen or qualify the results without changing the core idea.

Referee Report

3 major / 2 minor

Summary. The paper proposes a magic-informed quantum architecture search (QAS) technique that augments Monte Carlo Tree Search (MCTS) with a Graph Neural Network (GNN) to estimate nonstabilizerness (magic) of candidate circuits. The GNN induces a bias that steers the search toward user-specified high- or low-magic regimes. The method is benchmarked on structured ground-state energy minimization and general quantum state approximation tasks across varying system sizes and target magic levels, with claims that it successfully modulates magic both in the search tree and final circuits—even for out-of-distribution instances—while yielding consistent improvements in solution quality.

Significance. If the central claims are substantiated, the work would represent a meaningful step toward resource-aware automated quantum circuit design. By treating magic as an explicit, controllable bias within MCTS, it extends existing QAS frameworks to incorporate a key nonstabilizerness resource that underpins quantum advantage. The combination of GNN-based estimation with tree search is technically interesting and could generalize to other quantum resources if the accuracy and transferability issues are resolved.

major comments (3)

[Abstract and experimental results] Abstract and experimental results section: The central claim that the GNN supplies sufficiently accurate magic estimates to steer MCTS and produce better final circuits rests on unverified correlation between GNN outputs and true magic on circuits sampled from the MCTS search tree. No MAE, rank correlation, or other accuracy metrics are reported against exact magic values for held-out OOD nodes, leaving open the possibility that observed effects arise from proxy features (depth, gate count) rather than magic itself.
[Experimental results] Experimental results section: The assertion of 'consistent improvements in solution quality across all problems tested' is not supported by reported baselines (e.g., standard MCTS without magic bias), exact performance metrics, statistical tests, or variance across runs. Without these, it is impossible to determine whether the magic bias is load-bearing or incidental to the observed gains.
[Method] Method section on GNN training: The paper does not specify how the GNN training distribution relates to the distribution of circuits encountered during MCTS rollouts. If the training set does not cover the relevant circuit topologies and depths, the out-of-distribution performance claims cannot be rigorously evaluated.

minor comments (2)

[Background/Method] Notation for magic estimation and target levels should be introduced with explicit definitions and units in the background or method section to improve readability for readers outside the immediate subfield.
[Figures] Figure captions describing search-tree magic distributions should include the exact number of samples, error bars, and the precise definition of 'magic' used (e.g., which stabilizer Rényi entropy or other measure).

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback. We address each major comment below, providing clarifications based on the manuscript while agreeing to incorporate additional quantitative details and specifications in the revised version to strengthen the presentation.

read point-by-point responses

Referee: [Abstract and experimental results] Abstract and experimental results section: The central claim that the GNN supplies sufficiently accurate magic estimates to steer MCTS and produce better final circuits rests on unverified correlation between GNN outputs and true magic on circuits sampled from the MCTS search tree. No MAE, rank correlation, or other accuracy metrics are reported against exact magic values for held-out OOD nodes, leaving open the possibility that observed effects arise from proxy features (depth, gate count) rather than magic itself.

Authors: We appreciate this observation. The manuscript demonstrates the influence of the magic bias through direct measurements of nonstabilizerness in both the search tree nodes and the final selected circuits, including under out-of-distribution conditions. However, we agree that reporting explicit accuracy metrics, such as mean absolute error and Spearman rank correlation between GNN predictions and exact magic values on held-out circuits sampled from the MCTS process, would more rigorously substantiate that the steering effect is driven by magic estimation rather than proxy features. We will add these metrics, computed on a representative set of OOD circuits from the search, in the revised experimental results section. revision: yes
Referee: [Experimental results] Experimental results section: The assertion of 'consistent improvements in solution quality across all problems tested' is not supported by reported baselines (e.g., standard MCTS without magic bias), exact performance metrics, statistical tests, or variance across runs. Without these, it is impossible to determine whether the magic bias is load-bearing or incidental to the observed gains.

Authors: We acknowledge that the current presentation of results would benefit from more explicit baseline comparisons and statistical rigor. While the manuscript reports performance on the ground-state energy and state approximation tasks across system sizes and magic targets, it does not include a direct ablation against standard MCTS without the GNN-based magic bias, nor does it report run-to-run variance or statistical significance tests. We will revise the experimental results section to include these elements: direct comparisons to vanilla MCTS, mean and standard deviation over multiple independent runs, and appropriate statistical tests to confirm that the observed improvements are attributable to the magic-informed bias. revision: yes
Referee: [Method] Method section on GNN training: The paper does not specify how the GNN training distribution relates to the distribution of circuits encountered during MCTS rollouts. If the training set does not cover the relevant circuit topologies and depths, the out-of-distribution performance claims cannot be rigorously evaluated.

Authors: We agree that clarifying the training data distribution is necessary to support the OOD claims. The manuscript describes the GNN as trained on a dataset of quantum circuits but does not detail the generation procedure or its overlap with MCTS rollouts. In the revised Method section, we will specify the circuit sampling strategy used for training (including ranges of depths, gate sets, and topologies), the size of the training set, and a discussion of how this distribution relates to the circuits generated during MCTS search, thereby allowing readers to assess the degree of distribution shift. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical QAS method is self-contained

full rationale

The paper describes an empirical technique combining MCTS with a GNN for estimating circuit magic to bias architecture search. No derivation chain, equations, or uniqueness theorems are presented that reduce by construction to fitted inputs, self-definitions, or author-overlapping citations. The GNN is trained separately, benchmarks use independent problem instances, and claims of influence on magic and solution quality rest on experimental outcomes rather than tautological renaming or imported ansatzes. This is the standard case of a non-circular applied ML paper.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

Review based on abstract only; ledger reflects implied assumptions in the described method. No explicit free parameters, axioms, or invented entities are detailed in the abstract.

free parameters (2)

Target magic levels
Experiments span different target magic levels chosen as inputs to steer the search.
GNN weights
The graph neural network is trained to estimate magic, implying fitted parameters from training data.

axioms (2)

domain assumption Magic (nonstabilizerness) of a quantum circuit can be estimated from its graph representation by a trained GNN.
The bias mechanism depends on the GNN producing useful estimates.
domain assumption Biasing MCTS with these estimates improves or at least does not degrade solution quality on the tested problems.
The paper reports consistent improvements but this is an empirical assumption.

pith-pipeline@v0.9.0 · 5497 in / 1431 out tokens · 83403 ms · 2026-05-07T16:29:11.408442+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

54 extracted references · 6 canonical work pages · 2 internal anchors

[1]

Differentiable quantum architecture search,

S.-X. Zhang, C.-Y . Hsieh, S. Zhang, and H. Yao, “Differentiable quantum architecture search,”Quantum Science and Technology, vol. 7, no. 4, p. 045023, 2022

2022
[2]

Evolutionary-based searching method for quantum circuit architecture,

A. Zhang and S. Zhao, “Evolutionary-based searching method for quantum circuit architecture,”Quantum Information Processing, vol. 22, no. 7, p. 283, 2023

2023
[3]

Mog-vqe: Multiobjective genetic variational quantum eigensolver,

D. Chivilikhin, A. Y . Samarin, V . I. Ulyantsev, I. V . Iorsh, A. R. Oganov, and O. Kyriienko, “Mog-vqe: Multiobjective genetic variational quantum eigensolver,”arXiv preprint arXiv:2007.04424, 2020

work page arXiv 2007
[4]

Hierarchical quantum circuit representations for neural architecture search,

M. Lourens, I. Sinayskiy, D. K. Park, C. Blank, and F. Petruccione, “Hierarchical quantum circuit representations for neural architecture search,”npj Quantum Information, vol. 9, no. 1, p. 79, 2023

2023
[5]

Kuo, Y.-L

E.-J. Kuo, Y .-L. L. Fang, and S. Y .-C. Chen, “Quantum ar- chitecture search via deep reinforcement learning,”arXiv preprint arXiv:2104.07715, 2021

work page arXiv 2021
[6]

Reinforcement learning for optimization of variational quan- tum circuit architectures,

M. Ostaszewski, L. M. Trenkwalder, W. Masarczyk, E. Scerri, and V . Dunjko, “Reinforcement learning for optimization of variational quan- tum circuit architectures,”Advances in Neural Information Processing Systems, vol. 34, pp. 18182–18194, 2021

2021
[7]

Curriculum reinforcement learning for quantum architecture search under hardware errors,

Y . J. Patel, A. Kundu, M. Ostaszewski, X. Bonet-Monroig, V . Dun- jko, and O. Danaci, “Curriculum reinforcement learning for quantum architecture search under hardware errors,” inThe Twelfth International Conference on Learning Representations, 2024

2024
[8]

Automated quantum circuit design with Nested Monte Carlo Tree Search,

P. Wang, M. Usman, U. Parampalli, L. C. Hollenberg, and C. R. Myers, “Automated quantum circuit design with Nested Monte Carlo Tree Search,”IEEE Transactions on Quantum Engineering, 2023

2023
[9]

Quantum circuit design using a progressive widening enhanced monte carlo tree search,

V . Lipardi, D. Dibenedetto, G. Stamoulis, and M. H. M. Winands, “Quantum circuit design using a progressive widening enhanced monte carlo tree search,”Advanced Quantum Technologies, vol. 8, no. 10, p. e2500093, 2025

2025
[10]

Quantum architecture search with meta-learning,

Z. He, C. Chen, L. Li, S. Zheng, and H. Situ, “Quantum architecture search with meta-learning,”Advanced Quantum Technologies, vol. 5, no. 8, p. 2100134, 2022

2022
[11]

Self-supervised rep- resentation learning for bayesian quantum architecture search,

Z. He, H. Chen, Y . Zhou, H. Situ, Y . Li, and L. Li, “Self-supervised rep- resentation learning for bayesian quantum architecture search,”Physical Review A, vol. 111, no. 3, p. 032403, 2025

2025
[12]

Barren plateaus in variational quantum computing,

M. Larocca, S. Thanasilp, S. Wang, K. Sharma, J. Biamonte, P. J. Coles, L. Cincio, J. R. McClean, Z. Holmes, and M. Cerezo, “Barren plateaus in variational quantum computing,”Nature Reviews Physics, vol. 7, no. 4, pp. 174–189, 2025

2025
[13]

Variational quantum algorithms,

M. Cerezo, A. Arrasmith, R. Babbush, S. C. Benjamin, S. Endo, K. Fujii, J. R. McClean, K. Mitarai, X. Yuan, L. Cincio,et al., “Variational quantum algorithms,”Nature Reviews Physics, vol. 3, no. 9, pp. 625– 644, 2021

2021
[14]

Exponential con- centration in quantum kernel methods,

S. Thanasilp, S. Wang, M. Cerezo, and Z. Holmes, “Exponential con- centration in quantum kernel methods,”Nature communications, vol. 15, no. 1, p. 5200, 2024

2024
[15]

An adaptive variational algorithm for exact molecular simulations on a quantum computer,

H. R. Grimsley, S. E. Economou, E. Barnes, and N. J. Mayhall, “An adaptive variational algorithm for exact molecular simulations on a quantum computer,”Nature communications, vol. 10, no. 1, p. 3007, 2019

2019
[16]

Progressive strategies for Monte-Carlo Tree Search,

G. M. J.-B. Chaslot, M. H. M. Winands, H. J. van den Herik, J. W. H. M. Uiterwijk, and B. Bouzy, “Progressive strategies for Monte-Carlo Tree Search,”New Mathematics and Natural Computation, vol. 4, no. 3, pp. 343–357, 2008

2008
[17]

The Heisenberg Representation of Quantum Computers

D. Gottesman, “The heisenberg representation of quantum computers,” arXiv preprint quant-ph/9807006, 1998

work page internal anchor Pith review arXiv 1998
[18]

Stabilizer r ´enyi entropy,

L. Leone, S. F. Oliviero, and A. Hamma, “Stabilizer r ´enyi entropy,” Physical Review Letters, vol. 128, no. 5, p. 050402, 2022

2022
[19]

Nonstabilizerness estimation using graph neural networks,

V . Lipardi, D. Dibenedetto, G. Stamoulis, E. van Nieuwenburg, and M. H. M. Winands, “Nonstabilizerness estimation using graph neural networks,”Machine Learning: Science and Technology, 2026

2026
[20]

Mastering the game of Go with deep neural networks and tree search,

D. Silver, A. Huang, C. J. Maddison, A. Guez, L. Sifre, G. Van Den Driessche, J. Schrittwieser, I. Antonoglou, V . Panneershelvam, M. Lanctot,et al., “Mastering the game of Go with deep neural networks and tree search,”nature, vol. 529, no. 7587, pp. 484–489, 2016

2016
[21]

Many-body quantum magic,

Z.-W. Liu and A. Winter, “Many-body quantum magic,”PRX Quantum, vol. 3, no. 2, p. 020333, 2022

2022
[22]

Many-body magic via pauli-markov chains—from criticality to gauge theories,

P. S. Tarabunga, E. Tirrito, T. Chanda, and M. Dalmonte, “Many-body magic via pauli-markov chains—from criticality to gauge theories,”PRX Quantum, vol. 4, no. 4, p. 040317, 2023

2023
[23]

Quantum resources of quantum and classical variational methods,

T. Spriggs, A. Ahmadi, B. Chen, and E. Greplova, “Quantum resources of quantum and classical variational methods,”Machine Learning: Science and Technology, vol. 6, no. 1, p. 015042, 2025

2025
[24]

Capecci, G

C. Capecci, G. C. Santra, A. Bottarelli, E. Tirrito, and P. Hauke, “Role of nonstabilizerness in quantum optimization,”arXiv preprint arXiv:2505.17185, 2025

work page arXiv 2025
[25]

Improved simulation of stabilizer circuits,

S. Aaronson and D. Gottesman, “Improved simulation of stabilizer circuits,”Physical Review A—Atomic, Molecular, and Optical Physics, vol. 70, no. 5, p. 052328, 2004

2004
[26]

Robustness of magic and symmetries of the stabiliser polytope,

M. Heinrich and D. Gross, “Robustness of magic and symmetries of the stabiliser polytope,”Quantum, vol. 3, p. 132, 2019

2019
[27]

Stabilizer entropies and nonstabilizerness mono- tones,

T. Haug and L. Piroli, “Stabilizer entropies and nonstabilizerness mono- tones,”Quantum, vol. 7, p. 1092, 2023

2023
[28]

A nonstabilizerness monotone from stabilizerness asymmetry,

P. Sonya Tarabunga, M. Frau, T. Haug, E. Tirrito, and L. Piroli, “A nonstabilizerness monotone from stabilizerness asymmetry,”Quantum Science and Technology, vol. 10, no. 4, p. 045026, 2025

2025
[29]

Efficient witnessing and testing of magic in mixed quantum states,

T. Haug and P. S. Tarabunga, “Efficient witnessing and testing of magic in mixed quantum states,”npj Quantum Information, 2026

2026
[30]

Quantifying nonstabilizerness of matrix product states,

T. Haug and L. Piroli, “Quantifying nonstabilizerness of matrix product states,”Physical Review B, vol. 107, no. 3, p. 035148, 2023

2023
[31]

Nonstabilizerness via perfect pauli sampling of matrix product states,

G. Lami and M. Collura, “Nonstabilizerness via perfect pauli sampling of matrix product states,”Physical Review Letters, vol. 131, no. 18, p. 180401, 2023

2023
[32]

Nonstabi- lizerness via matrix product states in the pauli basis,

P. S. Tarabunga, E. Tirrito, M. C. Ba ˜nuls, and M. Dalmonte, “Nonstabi- lizerness via matrix product states in the pauli basis,”Physical Review Letters, vol. 133, no. 1, p. 010601, 2024

2024
[33]

Nonstabilizerness of neural quantum states,

A. Sinibaldi, A. F. Mello, M. Collura, and G. Carleo, “Nonstabilizerness of neural quantum states,”Physical Review Research, vol. 7, no. 4, p. 043289, 2025

2025
[34]

Magic-induced computational separation in entanglement theory,

A. Gu, S. F. Oliviero, and L. Leone, “Magic-induced computational separation in entanglement theory,”PRX Quantum, vol. 6, no. 2, p. 020324, 2025

2025
[35]

Entanglement – nonstabilizerness separation in hybrid quantum circuits,

G. E. Fux, E. Tirrito, M. Dalmonte, and R. Fazio, “Entanglement – nonstabilizerness separation in hybrid quantum circuits,”Phys. Rev. Res., vol. 6, p. L042030, Oct 2024

2024
[36]

Nonstabilizerness determining the hardness of direct fidelity estimation,

L. Leone, S. F. Oliviero, and A. Hamma, “Nonstabilizerness determining the hardness of direct fidelity estimation,”Physical Review A, vol. 107, no. 2, p. 022429, 2023

2023
[37]

Stabilizer entropies are monotones for magic- state resource theory,

L. Leone and L. Bittel, “Stabilizer entropies are monotones for magic- state resource theory,”Physical Review A, vol. 110, no. 4, p. L040403, 2024

2024
[38]

Unveiling the stabilizer group of a matrix product state,

G. Lami and M. Collura, “Unveiling the stabilizer group of a matrix product state,”Physical Review Letters, vol. 133, no. 1, p. 010602, 2024

2024
[39]

Retrieving nonstabilizerness with neural networks,

A. F. Mello, G. Lami, and M. Collura, “Retrieving nonstabilizerness with neural networks,”Physical Review A, vol. 111, no. 1, p. 012440, 2025

2025
[40]

A study on stabilizer r ´enyi entropy estimation using machine learning,

V . Lipardi, D. Dibenedetto, G. Stamoulis, and M. H. M. Winands, “A study on stabilizer r ´enyi entropy estimation using machine learning,” in 2025 IEEE International Conference on Quantum Artificial Intelligence (QAI), pp. 159–164, IEEE, 2025

2025
[41]

Efficient selectivity and backup operators in Monte-Carlo Tree Search,

R. Coulom, “Efficient selectivity and backup operators in Monte-Carlo Tree Search,” inInternational Conference on Computers and Games, pp. 72–83, Springer, 2007

2007
[42]

Finite-time analysis of the multiarmed bandit problem,

P. Auer, N. Cesa-Bianchi, and P. Fischer, “Finite-time analysis of the multiarmed bandit problem,”Machine learning, vol. 47, pp. 235–256, 2002

2002
[43]

Bandit based Monte-Carlo planning,

L. Kocsis and C. Szepesv ´ari, “Bandit based Monte-Carlo planning,” in Machine Learning: ECML 2006: 17th European Conference on Machine Learning Berlin, Germany, September 18-22, 2006 Proceedings 17, pp. 282–293, Springer, 2006

2006
[44]

Quantum circuit evolution on NISQ devices,

L. Franken, B. Georgiev, S. Mucke, M. Wolter, R. Heese, C. Bauckhage, and N. Piatkowski, “Quantum circuit evolution on NISQ devices,” in 2022 IEEE Congress on Evolutionary Computation (CEC), pp. 1–8, IEEE, 2022

2022
[45]

Nested Monte-Carlo Tree Search for Online Planning in Large MDPs,

H. Baier and M. H. M. Winands, “Nested Monte-Carlo Tree Search for Online Planning in Large MDPs,” inECAI, vol. 242, pp. 109–114, 2012

2012
[46]

Single-player monte-carlo tree search for SameGame,

M. P. D. Schadd, M. H. M. Winands, M. J. W. Tak, and J. W. H. M. Uiterwijk, “Single-player monte-carlo tree search for SameGame,” Knowledge-Based Systems, vol. 34, pp. 3–11, 2012

2012
[47]

Multi-armed bandits with episode context,

C. D. Rosin, “Multi-armed bandits with episode context,”Annals of Mathematics and Artificial Intelligence, vol. 61, no. 3, pp. 203–230, 2011

2011
[48]

The proof and measurement of association between two things.,

C. Spearman, “The proof and measurement of association between two things.,” 1961

1961
[49]

Adam: A Method for Stochastic Optimization

D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

work page internal anchor Pith review arXiv 2014
[50]

Statistical analysis of quantum state learning process in quantum neural networks,

H.-k. Zhang, C. Zhu, M. Jing, and X. Wang, “Statistical analysis of quantum state learning process in quantum neural networks,”Advances in Neural Information Processing Systems, vol. 36, pp. 33133–33160, 2023

2023
[51]

Does provable absence of barren plateaus imply classical simulability?,

M. Cerezo, M. Larocca, D. Garc ´ıa-Mart´ın, N. L. Diaz, P. Braccia, E. Fontana, M. S. Rudolph, P. Bermejo, A. Ijaz, S. Thanasilp,et al., “Does provable absence of barren plateaus imply classical simulability?,” Nature Communications, vol. 16, no. 1, p. 7907, 2025

2025
[52]

Quantum convolutional neural networks are effectively classically simulable,

P. Bermejo, P. Braccia, M. S. Rudolph, Z. Holmes, L. Cincio, and M. Cerezo, “Quantum convolutional neural networks are effectively classically simulable,”PRX Quantum, vol. 7, p. 020304, Apr 2026

2026
[53]

How Quantum Circuits Actually Learn: A Causal Identification of Genuine Quantum Contributions, March 2026

C. Y . Kesiku and B. Garcia-Zapirain, “How quantum circuits actually learn: A causal identification of genuine quantum contributions,”arXiv preprint arXiv:2603.16321, 2026

work page arXiv 2026
[54]

Generalized unitary coupled cluster wave functions for quantum computation,

J. Lee, W. J. Huggins, M. Head-Gordon, and K. B. Whaley, “Generalized unitary coupled cluster wave functions for quantum computation,” Journal of Chemical Theory and Computation, vol. 15, no. 1, pp. 311– 324, 2018

2018