Pith. sign in

REVIEW 3 major objections 4 minor 54 references

Special-Unitary Parameterization for Trainable Variational Quantum Circuits

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Restricting variational circuit layers to SU(2^k) blocks keeps gradient variance polynomial, making trainable variational circuits without barren plateaus.

desk verdict The paper's central claim that overlapping brick-wall SU(4) blocks yield a direct-sum DLA is false, so the barren-plateau argument collapses and the small numerical results don't rescue it. read the letter →

arxiv 2507.05535 v1 pith:IFYPWVVB submitted 2025-07-07 quant-ph cs.LG

classification quant-phcs.LG PACS 03.67.Lx03.67.-a
keywords variationalquantumcircuitsdynamicalLiealgebrabarrenplateausparameter-shiftruleSU(2^k)blocksLie-subalgebraengineeringautoencodingclassifier
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that a variational quantum circuit whose layers are single exponentials of a small special-unitary subgroup, SU(2^k) with k much smaller than n, avoids barren plateaus without sacrificing expressivity. The claim is that restricting every layer to k-qubit blocks reduces the dynamical Lie algebra dimension from exponential O(4^n) to linear O(n), and since gradient variance scales inversely with that dimension, gradients fade polynomially instead of exponentially. If true, this gives a principled symmetry-based recipe for trainable variational quantum algorithms on near-term hardware. The authors support the claim with exact generalized parameter-shift gradients and with numerical benchmarks on six-qubit quantum auto-encoding and a four-qubit classification task, where their circuits keep order-of-magnitude larger gradient signals, converge 2-3 times faster, and reach higher fidelities than Pauli-rotation or hardware-efficient baselines.

What carries the argument

The load-bearing object is the SUN-VQC block: a single exponential U(θ) = exp(i Σ_a θ_a Λ_a), where the Λ_a form a basis of su(2^k) acting on a k-qubit block, with k=2 in all benchmarks, giving su(4), the 15-dimensional Lie algebra of two-qubit unitaries. The exponential map is surjective onto SU(4), so each block can realize any two-qubit gate in one geodesic step. The argument then leans on the claimed direct-sum structure of the global dynamical Lie algebra over the blocks, g = ⊕_blocks su(2^k), whose dimension grows only linearly in the number of blocks, plus the variance bound Var = O(dim $g^{{-1}}$). The generalized parameter-shift rule supplies exact, hardware-compatible derivatives by writing each partial derivative as a finite combination of shifted circuit evaluations.

What would settle it

Run the iterative Lie closure of the 15 generators of each SU(4) block in the six-qubit depth-two brick-wall circuit by repeated commutators, and compare the resulting dimension with 12 * 15 = 180. If the closure spans operators on three or more qubits and closes to a space larger than the direct sum, the central claim fails at the benchmark itself; the paper does not report the computed DLA dimension from its numerical experiment.

Watch

Extended reading notes

Core claim

The central discovery, on the paper's own terms, is that the exponential map of a compact Lie subalgebra su(2^k) gives a variational layer whose reachable manifold is both broad and low-dimensional. For k=2, each block W(ϑ) = exp(i Σ_{a=1}^{15} ϑ_a Λ_a) is a single geodesic step over all of SU(4). The paper claims that a depth-two brick-wall of such blocks on six qubits has dynamical Lie algebra g(12) = ⊕_{j=1}^{12} su(4) with dimension 180, and more generally that k-qubit block layers give dim g = O(n). Combining this with the bound Var(∂_θ C) = O(dim $g^{{-1}}$) yields only polynomial suppression of gradient variance, which the paper identifies as circumventing barren plateaus. Exact gradients are then obtained from the generalized parameter-shift rule, which expresses each derivative as a finite linear combination of shifted expectation values with coefficients derived from the generator spectrum.

Load-bearing premise

The load-bearing premise is that a brick-wall stack of k-qubit block exponentials generates a dynamical Lie algebra that stays a direct sum of independent su(2^k) factors; if adjacent blocks share qubits, their nested commutators can generate operators on larger supports and the algebra can grow toward su(2^n), which would erase the polynomial-gradient guarantee.

Editorial extensions

If this is right

  • A k-qubit block circuit has dynamical Lie algebra dimension O(n), so gradient variance decays as O(1/n) and optimization signal survives as the qubit count grows.
  • Exact gradients are available from a fixed parameter-shift stencil, so no ancillary qubits or finite-difference bias are needed.
  • On the six-qubit auto-encoding task, the SUN-VQC approach reaches the 10^-2 infidelity threshold within O(10^3) iterations and approaches the Schmidt-decomposition bound, whereas the hardware-efficient baseline stagnates roughly two orders of magnitude higher.
  • On the two-moons classification task, the SUN-VQC reaches about 98% accuracy and crosses 95% within roughly 50 epochs, outperforming the depth-matched hardware-efficient circuit by about 3 percentage points.
  • Since the reachable manifold has polynomial volume, classical optimizers search a dramatically smaller parameter space at fixed depth.

Reading between the lines

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

  • The paper leaves implicit that if the direct-sum DLA claim is verified for larger brick-wall layouts, the polynomial gradient scaling should extend beyond the six-qubit benchmark, making a 50+ qubit simulation of gradient variance versus depth a natural next test.
  • A single-exponential SU(4) block could be implemented as one calibrated two-qubit pulse on tunable-coupler or cross-resonance hardware, and the paper's own caveat about compilation overhead makes this a concrete hardware experiment.
  • The same Lie-subalgebra restriction generalizes to non-compact or projective symmetry groups such as Sp(2N) for fermionic parity, which would carry the barren-plateau argument into quantum chemistry and lattice-gauge ansätze.
  • The parameter-shift machinery is not tied to Pauli generators, so it applies to any structured circuit whose generators have discrete spectra, not only to the SUN-VQC construction.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes SUN-VQC, a variational quantum circuit architecture in which each elementary layer is a single exponential of an embedded su(2^k) subalgebra, with the intention of keeping the dynamical Lie algebra (DLA) small and thereby avoiding barren plateaus. The theoretical argument is that, for a circuit composed of two-qubit SU(4) blocks in a brick-wall layout, the DLA is a direct sum of local su(4) algebras, so dim g = O(n) and gradient variance is only polynomially suppressed. The authors derive a generalized parameter-shift rule, and support the proposal with numerical experiments on a six-qubit quantum autoencoding task and a four-qubit classification task, comparing SUN-VQC against Cartan-decomposition, Pauli-rotation, and hardware-efficient ansatze.

Significance. If the DLA claim were correct, the paper would offer a principled and practically relevant route to barren-plateau-resilient variational circuits. The paper does present a clear statement of the generalized parameter-shift framework and a useful finite-size comparison of several SU(4) parameterizations, which are strengths. However, the central theoretical premise is false for the implemented brick-wall layout: overlapping two-qubit blocks generate nested commutators that close to the full su(2^n), not to a direct sum of su(4) algebras. Since every scaling claim in the abstract and Section III depends on this premise, the main contribution does not stand as written.

major comments (3)
  1. [III.B, after Eq. (5)] The central claim g(12) = ⊕_{j=1}^{12} su(4) is not correct for the implemented circuit. A depth-two brick-wall layout on six qubits applies SU(4) blocks on overlapping nearest-neighbour pairs: for example, layer one contains pair (1,2) and layer two contains pair (2,3). The generators of the su(4) algebra on (1,2) and those on (2,3) do not mutually commute; nested commutators produce operators supported on three qubits, and repeated commutation along the connected brick-wall graph generates the full su(64), of dimension 4^6 - 1 = 4095, not 180. The sentence 'Because the block acts on disjoint qubit pairs' is inconsistent with the brick-wall geometry, and the paper's own Sec. II.A warns that when the control set is not a basis of a single embedded su(2^k), the DLA can grow up to the full su(2^n). Since the gradient-variance bound in Sec. II.A is Var = O(1/dim g), the claimed polynomial suppression and barren-plateau circumvention rest entirely on this false premise.
  2. [III.B, '12 blocks' paragraph] The circuit description is internally inconsistent: a depth-two brick-wall on six qubits contains five nearest-neighbour blocks in total (three in the first layer and two in the second), so it cannot yield 12 two-qubit blocks. If the authors instead intend multiple repetitions or a different arrangement, this must be specified, because the DLA calculation depends on the exact set of pairs and their overlaps. As written, the dimension count 180 = 12 × 15 does not match the stated depth-two brick-wall layout.
  3. [III.D and IV] The numerical evidence does not test the scaling claim. The benchmarks are performed at n = 6 (autoencoding) and n = 4 (classification), and no gradient-variance statistics are reported at any system size. The convergence curves in Figs. 2 and 3 could be explained by the larger per-block expressiveness or by small-system finite-size effects; they do not verify the predicted O(n^{-1}) gradient-variance scaling. To support the abstract's scalability claim, the paper would need to measure Var(∂_θ C) as a function of n over a meaningful range and compare it with dim g.
minor comments (4)
  1. [II.C] The generalized parameter-shift formula contains an unexplained weight matrix ω_ℓm; please define it explicitly and state the conditions under which the formula holds, since the reference to Wierichs et al. does not by itself justify the displayed expression.
  2. [Abstract and II.A] The statement 'SU(2^k) ⊂ SU(2^n) with k ≪ n' is imprecise: the embedding is a diagonal block embedding (Eq. 2), and the dimension reduction is O(4^k) per block only when the blocks are mutually non-overlapping; the text should say this explicitly to avoid confusion.
  3. [References] Reference [37] duplicates reference [10]; please remove the duplicate and renumber.
  4. [Figures] In Fig. 2, the vertical axis label is not visible in the text; please ensure all axis labels and legends are legible in the final version.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the barren-plateau argument rests on an external variance bound and an (unproven, likely incorrect) DLA assumption, not on fitted data or self-citation.

full rationale

The paper's load-bearing chain is: (i) gradient variance scales as O(1/dim g), cited to external prior work; (ii) local SU(4) blocks are asserted to give dim g = O(n); (iii) hence polynomial gradient suppression. No step reuses the paper's numerical results as input: the benchmarks report fidelity and accuracy curves, not gradient variances, and no parameter is fitted to training data and then called a prediction. The DLA claim in Sec. III.B ('Because the block acts on disjoint qubit pairs... the direct sum g(12) = ⊕_{j=1}^{12} su(4)') is asserted rather than derived; for the described depth-two brick-wall layout with overlapping nearest-neighbour pairs the claim appears mathematically incorrect, and if false the barren-plateau conclusion collapses. That is a correctness/mathematical-error concern, not circularity: the conclusion is not equivalent to the premise by definition, and the premise is not justified by a self-citation. The parameter-shift rule is cited to Wierichs et al. as independent external work, and the many self-citations in the introduction are background references rather than load-bearing premises. No circularity score above 0 is warranted.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The theoretical derivation introduces no fitted numerical constants for the barren-plateau claim. The argument rests on two external literature results (gradient-variance bound and parameter-shift rule) and one incorrect structural assumption about the DLA of the brick-wall circuit. The numerical speedup claims depend on manually chosen optimization hyperparameters.

free parameters (2)
  • learning rate eta = 2e-2
    Manually chosen for all ansatze; affects the convergence-speed claim but not the theoretical barren-plateau argument.
  • number of gradient iterations T = 30000
    Chosen by hand to guarantee monotone convergence; supports the numerical results only.
assumptions (5)
  • domain assumption Var[∂_θ C] = O(dim g^{-1}) is a universal bound.
    Invoked in Sec. II.A with citation to Refs. [41] and [51], but the result holds only under specific conditions (e.g., particular cost functions, approximate t-designs, or deep circuits). Treating it as universal overstates the theoretical support.
  • ad hoc to paper The depth-two brick-wall circuit on six qubits has DLA g(12) = ⊕_{j=1}^{12} su(4).
    Stated in Sec. III.B without proof. For overlapping nearest-neighbor brick-wall layers, the Lie closure of su(4) algebras on adjacent qubit pairs generates the full su(2^6), so this assumption is false and load-bearing.
  • standard math The exponential map is surjective for SU(4), so every two-qubit unitary is reachable in a single step.
    Standard Lie group fact, correctly cited in Sec. III.B.
  • standard math The generalized parameter-shift rule of Wierichs et al. is valid for SU(N) generators.
    External published result (Ref. [42]), used correctly in Sec. II.C.
  • domain assumption The fidelity F = ⟨000|ρ_trash|000⟩ is an appropriate cost function for auto-encoding.
    Standard choice for quantum autoencoders, not justified beyond the reference in Sec. III.A.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Special-Unitary Parameterization for Trainable Variational Quantum Circuits." pith.science (2026). https://pith.science/paper/IFYPWVVB

@misc{pith2026250705535,
  author       = {Pith},
  title        = {Pith review of: Special-Unitary Parameterization for Trainable Variational Quantum Circuits},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IFYPWVVB}},
  note         = {Machine review of arXiv:2507.05535}
}
abstract

We propose SUN-VQC, a variational-circuit architecture whose elementary layers are single exponentials of a symmetry-restricted Lie subgroup, $\mathrm{SU}(2^{k}) \subset \mathrm{SU}(2^{n})$ with $k \ll n$. Confining the evolution to this compact subspace reduces the dynamical Lie-algebra dimension from $\mathcal{O}(4^{n})$ to $\mathcal{O}(4^{k})$, ensuring only polynomial suppression of gradient variance and circumventing barren plateaus that plague hardware-efficient ans\"atze. Exact, hardware-compatible gradients are obtained using a generalized parameter-shift rule, avoiding ancillary qubits and finite-difference bias. Numerical experiments on quantum auto-encoding and classification show that SUN-VQCs sustain order-of-magnitude larger gradient signals, converge 2--3$\times$ faster, and reach higher final fidelities than depth-matched Pauli-rotation or hardware-efficient circuits. These results demonstrate that Lie-subalgebra engineering provides a principled, scalable route to barren-plateau-resilient VQAs compatible with near-term quantum processors.

Figures

Figures reproduced from arXiv: 2507.05535 by the authors.

Figure 1
Figure 1. Brick-wall SUN-VQC on four qubits. Blue blocks are parametrised [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. Learning dynamics of the SUN-VQC (blue) versus an equally deep [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 2
Figure 2. Mean infidelity L¯(t) for the four ansatze described in Sec. ¨ III, aver￾aged over 20 Haar-random input states. Shaded regions indicate ±1σ across the ensemble. The SUN-VQC ansatz (solid blue) attains the lowest steady￾state loss and exhibits the steepest initial descent, reflecting the favourable trade-off between expressivity and barren-plateau mitigation achieved by locality-preserving su(4) blocks. 0.825 0.850 0… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

54 extracted references · 27 canonical work pages

  1. [1]

    Quantum computing in the nisq era and beyond,

    J. Preskill, “Quantum computing in the nisq era and beyond,” Quantum, vol. 2, p. 79, 2018

  2. [2]

    Noisy intermediate-scale quantum algorithms,

    K. Bharti, A. Cervera-Lierta, T. H. Kyaw, T. Haug, S. Alperin-Lea, A. Anand, M. Degroote, H. Heimonen, J. S. Kottmann, T. Menke, et al., “Noisy intermediate-scale quantum algorithms,” Reviews of Modern Physics, vol. 94, no. 1, p. 015004, 2022

  3. [3]

    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

  4. [4]

    Training variational quantum algorithms is np-hard,

    L. Bittel and M. Kliesch, “Training variational quantum algorithms is np-hard,” Physical review letters, vol. 127, no. 12, p. 120502, 2021

  5. [5]

    Varia- tional quantum algorithms for nonlinear problems,

    M. Lubasch, J. Joo, P. Moinier, M. Kiffner, and D. Jaksch, “Varia- tional quantum algorithms for nonlinear problems,” Physical Review A , vol. 101, no. 1, p. 010301, 2020

  6. [6]

    Performance comparison of optimization methods on variational quantum algorithms,

    X. Bonet-Monroig, H. Wang, D. Vermetten, B. Senjean, C. Moussa, T. B ¨ack, V . Dunjko, and T. E. O’Brien, “Performance comparison of optimization methods on variational quantum algorithms,” Physical Review A, vol. 107, no. 3, p. 032407, 2023

  7. [7]

    Schr¨odinger-heisenberg variational quantum algorithms,

    Z.-X. Shang, M.-C. Chen, X. Yuan, C.-Y . Lu, and J.-W. Pan, “Schr¨odinger-heisenberg variational quantum algorithms,” Physical Re- view Letters, vol. 131, no. 6, p. 060406, 2023

  8. [8]

    Exploiting symmetry in variational quantum machine learning,

    J. J. Meyer, M. Mularski, E. Gil-Fuster, A. A. Mele, F. Arzani, A. Wilms, and J. Eisert, “Exploiting symmetry in variational quantum machine learning,” PRX quantum, vol. 4, no. 1, p. 010328, 2023

Show all 54 references
  1. [9]

    Variational quantum circuits for deep reinforcement learning,

    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. 141007–141024, 2020

  2. [11]

    Quantum agents in the gym: a variational quantum algorithm for deep q-learning,

    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

  3. [12]

    A distributed learning scheme for variational quantum algorithms,

    Y . Du, Y . Qian, X. Wu, and D. Tao, “A distributed learning scheme for variational quantum algorithms,” IEEE Transactions on Quantum Engineering, vol. 3, pp. 1–16, 2022

  4. [13]

    Learning quantum phase estimation by variational quantum circuits,

    C.-Y . Liu, K.-C. Chen, and C.-H. A. Lin, “Learning quantum phase estimation by variational quantum circuits,” in 2024 International Joint Conference on Neural Networks (IJCNN) , pp. 1–6, IEEE, 2024

  5. [14]

    Compressedmediq: Hybrid quantum machine learning pipeline for high-dimentional neu- roimaging data,

    K.-C. Chen, Y .-T. Li, T.-Y . Li, and C.-Y . Liu, “Compressedmediq: Hybrid quantum machine learning pipeline for high-dimentional neu- roimaging data,” arXiv preprint arXiv:2409.08584 , 2024

  6. [15]

    Quantum op- timization using variational algorithms on near-term quantum devices,

    N. Moll, P. Barkoutsos, L. S. Bishop, J. M. Chow, A. Cross, D. J. Egger, S. Filipp, A. Fuhrer, J. M. Gambetta, M. Ganzhorn, et al., “Quantum op- timization using variational algorithms on near-term quantum devices,” Quantum Science and Technology , vol. 3, no. 3, p. 030503, 2018

  7. [16]

    Improving variational quantum optimization using cvar,

    P. K. Barkoutsos, G. Nannicini, A. Robert, I. Tavernelli, and S. Woerner, “Improving variational quantum optimization using cvar,” Quantum, vol. 4, p. 256, 2020

  8. [17]

    Noise- aware distributed quantum approximate optimization algorithm on near- term quantum hardware,

    K.-C. Chen, X. Xu, F. Burt, C.-Y . Liu, S. Yu, and K. K. Leung, “Noise- aware distributed quantum approximate optimization algorithm on near- term quantum hardware,” in 2024 IEEE International Conference on Quantum Computing and Engineering (QCE) , vol. 2, pp. 144–149, IEEE, 2024

  9. [18]

    Learning to learn with quantum optimization via quantum neural networks,

    K.-C. Chen, H. Matsuyama, and W.-H. Huang, “Learning to learn with quantum optimization via quantum neural networks,” arXiv preprint arXiv:2505.00561, 2025

  10. [19]

    Quantum variational autoencoder,

    A. Khoshaman, W. Vinci, B. Denis, E. Andriyash, H. Sadeghi, and M. H. Amin, “Quantum variational autoencoder,” Quantum Science and Technology, vol. 4, no. 1, p. 014001, 2018

  11. [20]

    Effect of data encoding on the expressive power of variational quantum-machine-learning models,

    M. Schuld, R. Sweke, and J. J. Meyer, “Effect of data encoding on the expressive power of variational quantum-machine-learning models,” Physical Review A , vol. 103, no. 3, p. 032430, 2021

  12. [21]

    Shedding light on the future: Exploring quantum neural networks through optics,

    S. Yu, Z. Jia, A. Zhang, E. Mer, Z. Li, V . Crescimanna, K.-C. Chen, R. B. Patel, I. A. Walmsley, and D. Kaszlikowski, “Shedding light on the future: Exploring quantum neural networks through optics,” Advanced Quantum Technologies, p. 2400074, 2024

  13. [22]

    Quantum machine learning for particle physics using a variational quantum classifier,

    A. Blance and M. Spannowsky, “Quantum machine learning for particle physics using a variational quantum classifier,” Journal of High Energy Physics, vol. 2021, no. 2, pp. 1–20, 2021

  14. [23]

    Quantum-enhanced support vector machine for large-scale multi-class stellar classification,

    K.-C. Chen, X. Xu, H. Makhanov, H.-H. Chung, and C.-Y . Liu, “Quantum-enhanced support vector machine for large-scale multi-class stellar classification,” in International Conference on Intelligent Com- puting, pp. 155–168, Springer, 2024

  15. [24]

    Quantum- classical-quantum workflow in quantum-hpc middleware with gpu ac- celeration,

    K.-C. Chen, X. Li, X. Xu, Y .-Y . Wang, and C.-Y . Liu, “Quantum- classical-quantum workflow in quantum-hpc middleware with gpu ac- celeration,” in 2024 International Conference on Quantum Communica- tions, Networking, and Computing (QCNC) , pp. 304–311, IEEE, 2024

  16. [25]

    Consensus-based distributed quantum kernel learning for speech recognition,

    K.-C. Chen, W. Ma, and X. Xu, “Consensus-based distributed quantum kernel learning for speech recognition,” arXiv preprint arXiv:2409.05770, 2024

  17. [26]

    Robust decentralized quantum kernel learning for noisy and adversarial environment,

    W. Ma, K.-C. Chen, S. Yu, M. Liu, and R. Deng, “Robust decentralized quantum kernel learning for noisy and adversarial environment,” arXiv preprint arXiv:2504.13782, 2025

  18. [27]

    Quantum-train long short- term memory: Application on flood prediction problem,

    C.-H. A. Lin, C.-Y . Liu, and K.-C. Chen, “Quantum-train long short- term memory: Application on flood prediction problem,” in 2024 IEEE International Conference on Quantum Computing and Engineering (QCE), vol. 2, pp. 268–273, IEEE, 2024

  19. [28]

    Quantum kernel-based long short-term memory for climate time-series forecasting,

    Y .-C. Hsu, N.-Y . Chen, T.-Y . Li, P.-H. H. Lee, and K.-C. Chen, “Quantum kernel-based long short-term memory for climate time-series forecasting,” in 2025 International Conference on Quantum Commu- nications, Networking, and Computing (QCNC) , pp. 421–426, IEEE, 2025

  20. [29]

    Programming variational quantum circuits with quantum-train agent,

    C.-Y . Liu, S. Y .-C. Chen, K.-C. Chen, W.-J. Huang, and Y .-J. Chang, “Programming variational quantum circuits with quantum-train agent,” in 2025 International Conference on Quantum Communications, Net- working, and Computing (QCNC) , pp. 544–548, IEEE, 2025

  21. [30]

    Toward large-scale distributed quantum long short-term memory with modular quantum computers,

    K.-C. Chen, S. Y .-C. Chen, C.-Y . Liu, and K. K. Leung, “Toward large-scale distributed quantum long short-term memory with modular quantum computers,” arXiv preprint arXiv:2503.14088 , 2025

  22. [31]

    Differentiable quantum architecture search in quantum-enhanced neural network parameter generation,

    S. Y .-C. Chen, C.-Y . Liu, K.-C. Chen, W.-J. Huang, Y .-J. Chang, and W.-H. Huang, “Differentiable quantum architecture search in quantum-enhanced neural network parameter generation,” arXiv preprint arXiv:2505.09653, 2025

  23. [32]

    Quantum-enhanced parameter- efficient learning for typhoon trajectory forecasting,

    C.-Y . Liu, K.-C. Chen, Y .-C. Chen, S. Y .-C. Chen, W.-H. Huang, W.-J. Huang, and Y .-J. Chang, “Quantum-enhanced parameter- efficient learning for typhoon trajectory forecasting,” arXiv preprint arXiv:2505.09395, 2025

  24. [33]

    Distributed quantum neural networks on distributed photonic quantum computing,

    K.-C. Chen, C.-Y . Liu, Y . Shang, F. Burt, and K. K. Leung, “Distributed quantum neural networks on distributed photonic quantum computing,” arXiv preprint arXiv:2505.08474 , 2025

  25. [34]

    Federated quantum-train with batched parameter generation,

    C.-Y . Liu and S. Y .-C. Chen, “Federated quantum-train with batched parameter generation,” arXiv preprint arXiv:2409.02763 , 2024

  26. [35]

    Qtrl: Toward practical quantum reinforcement learning via quantum- train,

    C.-Y . Liu, C.-H. A. Lin, C.-H. H. Yang, K.-C. Chen, and M.-H. Hsieh, “Qtrl: Toward practical quantum reinforcement learning via quantum- train,” arXiv preprint arXiv:2407.06103 , 2024

  27. [36]

    Differentiable quantum architecture search in asynchronous quantum reinforcement learning,

    S. Y .-C. Chen, “Differentiable quantum architecture search in asynchronous quantum reinforcement learning,” arXiv preprint arXiv:2407.18202, 2024

  28. [37]

    Variational quantum reinforcement learning via evolutionary optimization,

    S. Y .-C. Chen, C.-M. Huang, C.-W. Hsing, H.-S. Goan, and Y .- J. Kao, “Variational quantum reinforcement learning via evolutionary optimization,” Machine Learning: Science and Technology, vol. 3, no. 1, p. 015025, 2022

  29. [38]

    Quantum-train- based distributed multi-agent reinforcement learning,

    K.-C. Chen, S. Y .-C. Chen, C.-Y . Liu, and K. K. Leung, “Quantum-train- based distributed multi-agent reinforcement learning,” in 2025 IEEE Symposium for Multidisciplinary Computational Intelligence Incubators (MCII Companion), pp. 1–5, IEEE, 2025

  30. [39]

    Parameterized quantum circuits as machine learning models,

    M. Benedetti, E. Lloyd, S. Sack, and M. Fiorentini, “Parameterized quantum circuits as machine learning models,” Quantum science and technology, vol. 4, no. 4, p. 043001, 2019

  31. [40]

    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 , pp. 1–16, 2025

  32. [41]

    Cost function dependent barren plateaus in shallow parametrized quantum circuits,

    M. Cerezo, A. Sone, T. V olkoff, L. Cincio, and P. J. Coles, “Cost function dependent barren plateaus in shallow parametrized quantum circuits,” Nature communications, vol. 12, no. 1, p. 1791, 2021

  33. [42]

    General parameter- shift rules for quantum gradients,

    D. Wierichs, J. Izaac, C. Wang, and C. Y .-Y . Lin, “General parameter- shift rules for quantum gradients,” Quantum, vol. 6, p. 677, 2022

  34. [43]

    Classifica- tion of dynamical lie algebras of 2-local spin systems on linear, circular and fully connected topologies,

    R. Wiersema, E. K ¨okc¨u, A. F. Kemper, and B. N. Bakalov, “Classifica- tion of dynamical lie algebras of 2-local spin systems on linear, circular and fully connected topologies,” npj Quantum Information , vol. 10, no. 1, p. 110, 2024

  35. [44]

    Lie algebraic analysis and control of quantum dy- namics,

    D. D’Alessandro, “Lie algebraic analysis and control of quantum dy- namics,” arXiv preprint arXiv:0803.1193 , 2008

  36. [45]

    Barren plateaus in quantum neural network training landscapes,

    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, no. 1, p. 4812, 2018

  37. [46]

    Monotone metrics on matrix spaces,

    D. Petz, “Monotone metrics on matrix spaces,” Linear algebra and its applications, vol. 244, pp. 81–96, 1996

  38. [47]

    Geometric theory of nonlocal two-qubit operations,

    J. Zhang, J. Vala, S. Sastry, and K. B. Whaley, “Geometric theory of nonlocal two-qubit operations,” Physical Review A , vol. 67, no. 4, p. 042313, 2003

  39. [48]

    Constructive control of quantum systems using factorization of unitary operators,

    S. G. Schirmer, A. D. Greentree, V . Ramakrishna, and H. Rabitz, “Constructive control of quantum systems using factorization of unitary operators,” Journal of Physics A: Mathematical and General , vol. 35, no. 39, p. 8315, 2002

  40. [49]

    Eval- uating analytic gradients on quantum hardware,

    M. Schuld, V . Bergholm, C. Gogolin, J. Izaac, and N. Killoran, “Eval- uating analytic gradients on quantum hardware,” Physical Review A , vol. 99, no. 3, p. 032331, 2019

  41. [50]

    Here comes the su (n): multivariate quantum gates and gradients,

    R. Wiersema, D. Lewis, D. Wierichs, J. Carrasquilla, and N. Killoran, “Here comes the su (n): multivariate quantum gates and gradients,” Quantum, vol. 8, p. 1275, 2024

  42. [51]

    A lie algebraic theory of barren plateaus for deep parameterized quantum circuits,

    M. Ragone, B. N. Bakalov, F. Sauvage, A. F. Kemper, C. Ortiz Marrero, M. Larocca, and M. Cerezo, “A lie algebraic theory of barren plateaus for deep parameterized quantum circuits,” Nature Communications, vol. 15, no. 1, p. 7172, 2024

  43. [52]

    Cartan subalgebra approach to efficient measurements of quantum observables,

    T.-C. Yen and A. F. Izmaylov, “Cartan subalgebra approach to efficient measurements of quantum observables,” PRX quantum , vol. 2, no. 4, p. 040320, 2021

  44. [53]

    Time optimal control in spin systems,

    N. Khaneja, R. Brockett, and S. J. Glaser, “Time optimal control in spin systems,” Physical Review A , vol. 63, no. 3, p. 032308, 2001

  45. [54]

    Elementary gates for quantum computation,

    A. Barenco, C. H. Bennett, R. Cleve, D. P. DiVincenzo, N. Margolus, P. Shor, T. Sleator, J. A. Smolin, and H. Weinfurter, “Elementary gates for quantum computation,” Physical review A , vol. 52, no. 5, p. 3457, 1995

  46. [55]

    Efficient discrete feature encoding for variational quantum classifier,

    H. Yano, Y . Suzuki, K. M. Itoh, R. Raymond, and N. Yamamoto, “Efficient discrete feature encoding for variational quantum classifier,” IEEE Transactions on Quantum Engineering , vol. 2, pp. 1–14, 2021

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.