pith. machine review for the scientific record. sign in

arxiv: 2605.04604 · v1 · submitted 2026-05-06 · 🪐 quant-ph · cs.LG

Recognition: 3 theorem links

· Lean Theorem

Generative Quantum-inspired Kolmogorov-Arnold Eigensolver

Yu-Cheng Lin , Yu-Chao Hsu , I-Shan Tsai , Chun-Hua Lin , Kuo-Chung Peng , Jiun-Cheng Jiang , Yun-Yuan Wang , Tzung-Chi Huang
show 4 more authors
Tai-Yue Li Kuan-Cheng Chen Samuel Yen-Chi Chen Nan-Yow Chen
Authors on Pith no claims yet

Pith reviewed 2026-05-08 18:07 UTC · model grok-4.3

classification 🪐 quant-ph cs.LG
keywords generative quantum eigensolverKolmogorov-Arnold networksquantum chemistryparameter efficiencymolecular energiesstrongly correlated systemshybrid quantum-classical
0
0 comments X

The pith

A parameter-efficient Kolmogorov-Arnold version of the generative quantum eigensolver matches chemical accuracy while cutting trainable parameters and memory by about two thirds.

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

The paper introduces the generative quantum-inspired Kolmogorov-Arnold eigensolver as a lighter alternative to the earlier GPT-style generative quantum eigensolver for computing molecular ground-state energies. It keeps the autoregressive operator selection and quantum-selected configuration interaction steps but replaces the dense feed-forward networks with hybrid Kolmogorov-Arnold modules that rely on single-qubit data re-uploading activations for the required nonlinear mappings. Benchmarks on H4, N2, LiH, ethane, water, and the water dimer show the new version reaches the same chemical accuracy threshold, uses roughly 66 percent fewer parameters, runs faster in wall time, and converges better on strongly correlated cases. A reader would care because the reduced classical overhead makes hybrid quantum-classical energy calculations more feasible on current high-performance computers without changing the quantum part of the workflow.

Core claim

GQKAE replaces the parameter-heavy feed-forward network components in GPT-style generative eigensolvers with hybrid quantum-inspired Kolmogorov-Arnold network modules, forming a compact HQKANsformer backbone. The method preserves autoregressive operator selection and the quantum-selected configuration interaction evaluation pipeline, while using single-qubit DatA Re-Uploading ActivatioN modules to provide expressive nonlinear mappings. Numerical benchmarks on H4, N2, LiH, C2H6, H2O, and the H2O dimer show that GQKAE achieves chemical accuracy comparable to the GPT-based GQE architecture, while reducing trainable parameters and memory by approximately 66% and improving wall-time performance.

What carries the argument

Hybrid quantum-inspired Kolmogorov-Arnold networks with single-qubit data re-uploading activation modules that replace dense feed-forward layers to supply the nonlinear transformations inside the generative backbone.

If this is right

  • The full autoregressive operator selection and post-processing pipeline remains unchanged while classical memory and training cost drop.
  • Convergence behavior and final energy accuracy improve on strongly correlated molecules such as N2 and LiH.
  • Wall-clock time decreases, supporting longer runs or larger active spaces on the same classical hardware.
  • The approach preserves compatibility with existing quantum circuit simulation and selected configuration interaction stages.

Where Pith is reading between the lines

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

  • Kolmogorov-Arnold replacements could be tried in other generative or autoregressive components of quantum-chemistry pipelines to achieve similar resource savings.
  • Lower memory footprints may allow the same workflow to handle larger basis sets or more electrons without upgrading classical hardware.
  • Direct tests on actual quantum hardware rather than classical simulators would show whether the reduced parameter count also lowers circuit depth or noise sensitivity.
  • The method might enable faster retraining when the molecular Hamiltonian changes, supporting on-the-fly adaptation in screening workflows.

Load-bearing premise

Single-qubit data re-uploading activation modules inside the Kolmogorov-Arnold networks can generate nonlinear mappings expressive enough to fully replace the original feed-forward networks across the tested molecular systems without any loss of accuracy.

What would settle it

If GQKAE on a new strongly correlated molecule such as the chromium dimer produces final energy errors larger than chemical accuracy while the original GQE reaches chemical accuracy with the same or fewer total parameters, the efficiency claim would not hold.

Figures

Figures reproduced from arXiv: 2605.04604 by Chun-Hua Lin, I-Shan Tsai, Jiun-Cheng Jiang, Kuan-Cheng Chen, Kuo-Chung Peng, Nan-Yow Chen, Samuel Yen-Chi Chen, Tai-Yue Li, Tzung-Chi Huang, Yu-Chao Hsu, Yu-Cheng Lin, Yun-Yuan Wang.

Figure 1
Figure 1. Figure 1: Architecture of the Jiang–Huang–Chen–Goan Net￾work (JHCG Net) [44]. The JHCG Net uses an encoder– processor–decoder structure, where a QKAN-based latent processor forms the Hybrid QKAN (HQKAN) module and introduces quantum-inspired nonlinear transformations in the latent space. where ϕl,j,i(·) denotes the learnable univariate function associ￾ated with the edge from input coordinate i to output coordinate j… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the GQKAE framework. The HQKANsformer defines an autoregressive distribution over operator sequences, which are used to construct candidate quantum circuits. The generated circuits prepare trial states that are measured to obtain sampled Slater determinants. These determinants span a truncated subspace of dimension Ddmax , where the Hamiltonian is classically diagonalized to compute the QSCI en… view at source ↗
Figure 3
Figure 3. Figure 3: Optimization history showing the best-so-far energy error relative to the CASCI baseline up to each iteration. Solid view at source ↗
Figure 4
Figure 4. Figure 4: Potential energy surface of the six target molecules. Subfigures present the energy variations with respect to bond view at source ↗
Figure 5
Figure 5. Figure 5: Absolute error from the CASCI energy on a logarithmic scale for the six molecular systems. The blue dashed line view at source ↗
Figure 6
Figure 6. Figure 6: Absolute energy error as a function of measurement view at source ↗
Figure 7
Figure 7. Figure 7: Absolute energy error as a function of the view at source ↗
read the original abstract

High-performance computing (HPC) is increasingly important for scalable quantum chemistry workflows that couple classical generative models, quantum circuit simulation, and selected configuration interaction postprocessing. We present the generative quantum-inspired Kolmogorov-Arnold eigensolver (GQKAE), a parameter-efficient extension of the generative quantum eigensolver (GQE) for quantum chemistry. GQKAE replaces the parameter-heavy feed-forward network components in GPT-style generative eigensolvers with hybrid quantum-inspired Kolmogorov-Arnold network modules, forming a compact HQKANsformer backbone. The method preserves autoregressive operator selection and the quantum-selected configuration interaction evaluation pipeline, while using single-qubit DatA Re-Uploading ActivatioN modules to provide expressive nonlinear mappings. Numerical benchmarks on H4, N2, LiH, C2H6, H2O, and the H2O dimer show that GQKAE achieves chemical accuracy comparable to the GPT-based GQE architecture, while reducing trainable parameters and memory by approximately 66% and improving wall-time performance. For strongly correlated systems such as N2 and LiH, GQKAE also improves convergence behavior and final energy errors. These results indicate that quantum-inspired Kolmogorov-Arnold networks can reduce classical-side overhead while preserving circuit-generation quality, offering a scalable route for HPC-quantum co-design on near-term quantum platforms.

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.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces the Generative Quantum-inspired Kolmogorov-Arnold Eigensolver (GQKAE) as a parameter-efficient variant of the generative quantum eigensolver (GQE). It replaces the feed-forward network components of the GPT-style backbone with hybrid quantum-inspired Kolmogorov-Arnold network (HQKAN) modules that employ single-qubit DatA Re-Uploading Activation (DRA) modules to supply nonlinear mappings, while retaining the autoregressive operator selection and quantum-selected configuration interaction pipeline. Numerical experiments on H4, N2, LiH, C2H6, H2O and the H2O dimer are reported to show chemical accuracy comparable to GQE, with roughly 66% fewer trainable parameters, reduced memory, faster wall-clock time, and improved convergence on strongly correlated cases.

Significance. If the reported numerical results prove robust, the work would demonstrate that quantum-inspired Kolmogorov-Arnold networks can materially reduce classical-side overhead in generative quantum-chemistry pipelines without sacrificing circuit-generation quality, thereby supporting more scalable HPC-quantum co-design on near-term hardware.

major comments (2)
  1. [Abstract / Numerical benchmarks] Abstract and numerical benchmarks section: the central claims of comparable chemical accuracy and 66% parameter reduction rest on empirical results, yet no error bars, exact baseline definitions, data-exclusion criteria, or statistical significance tests are supplied. This absence prevents assessment of whether the reported gains for N2 and LiH are statistically meaningful or artifacts of the chosen test set.
  2. [HQKANsformer backbone / single-qubit DRA modules] Architecture description of the HQKANsformer backbone: the claim that single-qubit DRA modules furnish sufficiently expressive univariate functions to replace the original feed-forward layers is not accompanied by any expressivity analysis or comparison of the spanned function space. Given the Kolmogorov-Arnold theorem's dependence on the richness of the univariate approximants, the limited rotation-angle and measurement basis of a single qubit may restrict representational power for high-dimensional configuration-interaction spaces; this is load-bearing for the parameter-reduction claim.
minor comments (2)
  1. [Abstract] The acronym 'DatA Re-Uploading ActivatioN' contains inconsistent capitalization; standardize to 'Data Re-Uploading Activation (DRA)'.
  2. [Numerical benchmarks] The manuscript should include a table or explicit list of the exact number of trainable parameters for both GQE and GQKAE on each benchmark molecule to substantiate the 'approximately 66%' reduction.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review of our manuscript. We have carefully considered each major comment and provide point-by-point responses below. Revisions have been made to strengthen the presentation of results and clarify architectural choices.

read point-by-point responses

  1. Referee: [Abstract / Numerical benchmarks] Abstract and numerical benchmarks section: the central claims of comparable chemical accuracy and 66% parameter reduction rest on empirical results, yet no error bars, exact baseline definitions, data-exclusion criteria, or statistical significance tests are supplied. This absence prevents assessment of whether the reported gains for N2 and LiH are statistically meaningful or artifacts of the chosen test set.

    Authors: We agree that the absence of error bars, explicit baseline definitions, and statistical tests limits the ability to fully assess robustness. In the revised manuscript we have added error bars (standard deviation over five independent training runs) to all reported energies, parameter counts, and wall-clock times. We have clarified that the GQE baseline uses the original GPT-style feed-forward networks with identical autoregressive operator selection and quantum-selected CI pipeline. No data points were excluded; all six molecular systems were evaluated using standard geometries and basis sets from the literature. We have also included paired t-test p-values confirming that the convergence improvements on N2 and LiH are statistically significant at the 95% level. These additions directly address the concern and demonstrate that the reported gains are not artifacts. revision: yes

  2. Referee: [HQKANsformer backbone / single-qubit DRA modules] Architecture description of the HQKANsformer backbone: the claim that single-qubit DRA modules furnish sufficiently expressive univariate functions to replace the original feed-forward layers is not accompanied by any expressivity analysis or comparison of the spanned function space. Given the Kolmogorov-Arnold theorem's dependence on the richness of the univariate approximants, the limited rotation-angle and measurement basis of a single qubit may restrict representational power for high-dimensional configuration-interaction spaces; this is load-bearing for the parameter-reduction claim.

    Authors: We acknowledge that a formal expressivity analysis comparing the function space of single-qubit DRA modules to standard feed-forward layers is not provided in the original submission. While a complete theoretical characterization of the spanned function space lies beyond the scope of this applied work, the Kolmogorov-Arnold theorem guarantees universal approximation when the univariate functions are sufficiently rich; our single-qubit DRA modules achieve nonlinearity through data re-uploading and tunable rotation angles, which prior literature has shown to be expressive for regression tasks. The empirical results across H4, N2, LiH, C2H6, H2O and the H2O dimer—including strongly correlated cases—demonstrate that chemical accuracy is retained with the 66% parameter reduction. In revision we have added a short paragraph in the methods section discussing the design rationale for single-qubit modules, citing relevant quantum-inspired activation studies, and noting that the observed performance validates practical sufficiency even if tighter theoretical bounds remain an open question for future investigation. revision: partial

Circularity Check

0 steps flagged

No significant circularity in empirical architecture comparison

full rationale

The paper introduces GQKAE as an empirical extension of GQE, replacing feed-forward components with hybrid quantum-inspired Kolmogorov-Arnold networks using single-qubit DRA modules. All performance claims (chemical accuracy, 66% parameter reduction, improved convergence on N2/LiH) are presented as outcomes of numerical benchmarks on fixed molecular systems (H4, N2, LiH, C2H6, H2O, H2O dimer) rather than any mathematical derivation, first-principles prediction, or self-referential fit. No equations or steps reduce by construction to inputs; results are externally falsifiable via the reported experiments. Self-citation of GQE (if present) is not load-bearing for any derivation.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The central claim rests on empirical validation of a new network architecture; free parameters are the trainable weights in the HQKANsformer; axioms include standard assumptions of quantum circuit simulation and selected configuration interaction; the DatA Re-Uploading modules are an invented entity without independent evidence outside the benchmarks.

free parameters (1)
  • trainable parameters in HQKANsformer backbone
    Network weights fitted during training on molecular benchmarks to achieve the reported accuracy and parameter reduction.
axioms (1)
  • domain assumption Quantum circuit simulation and selected configuration interaction postprocessing accurately evaluate the generated operators
    The method preserves the GQE evaluation pipeline and assumes its correctness for the reported energies.
invented entities (1)
  • single-qubit DatA Re-Uploading ActivatioN modules no independent evidence
    purpose: Provide expressive nonlinear mappings in the Kolmogorov-Arnold network components
    New module type introduced to replace feed-forward activations; no independent falsifiable evidence provided beyond the molecular benchmarks.

pith-pipeline@v0.9.0 · 5586 in / 1485 out tokens · 53262 ms · 2026-05-08T18:07:50.391005+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

  • Cost.FunctionalEquation (J-cost uniqueness) washburn_uniqueness_aczel unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    QKAN realizes learnable edge functions through DatA Re-Uploading ActivatioN (DARUAN) modules, inspired by single-qubit data re-uploading circuits whose Fourier spectrum expands with repeated encoding.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

99 extracted references · 16 canonical work pages · 3 internal anchors

  1. [1]

    Quantum-centric high performance computing for quantum chemistry,

    J. Liu, H. Ma, H. Shang, Z. Li, and J. Yang, “Quantum-centric high performance computing for quantum chemistry,”Physical Chemistry Chemical Physics, vol. 26, pp. 15 831–15 843, 2024

    2024

  2. [2]

    Massively scalable workflows for quantum chemistry: Bigchem and chemcloud,

    C. B. Hicks and T. J. Martinez, “Massively scalable workflows for quantum chemistry: Bigchem and chemcloud,”The Journal of Chemical Physics, vol. 160, no. 14, p. 142501, 2024

    2024

  3. [3]

    Chemistry beyond the scale of exact di- agonalization on a quantum-centric supercomputer,

    J. Robledo-Morenoet al., “Chemistry beyond the scale of exact di- agonalization on a quantum-centric supercomputer,”Science Advances, vol. 11, no. 25, p. eadu9991, 2025

    2025

  4. [4]

    Efficient and noise resilient measurements for quantum chemistry on near-term quantum computers,

    W. J. Huggins, J. R. McClean, N. C. Rubin, Z. Jiang, N. Wiebe, K. B. Whaley, and R. Babbush, “Efficient and noise resilient measurements for quantum chemistry on near-term quantum computers,”npj Quantum Information, vol. 7, no. 1, p. 23, 2021

    2021

  5. [5]

    Kanno, M

    K. Kannoet al., “Quantum-selected configuration interaction: Classi- cal diagonalization of hamiltonians in subspaces selected by quantum computers,”arXiv preprint arXiv:2302.11320, 2023

    work page arXiv 2023

  6. [6]

    Generative Circuit Design for Quantum-Selected Configuration Interaction

    R. Kemmokuet al., “Generative circuit design for quantum-selected configuration interaction,”arXiv preprint arXiv:2604.09756, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  7. [7]

    Babbush, R

    R. Babbushet al., “The grand challenge of quantum applications,”arXiv preprint arXiv:2511.09124, 2025

    work page arXiv 2025

  8. [8]

    Quantum circuit learning,

    K. Mitarai, M. Negoro, M. Kitagawa, and K. Fujii, “Quantum circuit learning,”Phys. Rev. A, vol. 98, p. 032309, Sep 2018

    2018

  9. [9]

    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

  10. [10]

    Quantum machine learning beyond kernel methods,

    S. Jerbi, L. J. Fiderer, H. Poulsen Nautrup, J. M. K ¨ubler, H. J. Briegel, and V . Dunjko, “Quantum machine learning beyond kernel methods,” Nature Communications, vol. 14, no. 1, p. 517, 2023

    2023

  11. [11]

    Quantum adaptive excitation network with variational quantum circuits for channel atten- tion,

    Y .-C. Hsu, K.-C. Chen, T.-Y . Li, and N.-Y . Chen, “Quantum adaptive excitation network with variational quantum circuits for channel atten- tion,” in2025 IEEE International Conference on Quantum Computing and Engineering (QCE), vol. 2. IEEE, 2025, pp. 344–349

    2025

  12. [12]

    A quantum circuit-based compression perspective for parameter-efficient learning,

    C.-Y . Liu, C.-H. H. Yang, H.-S. Goan, and M.-H. Hsieh, “A quantum circuit-based compression perspective for parameter-efficient learning,” inThe Thirteenth International Conference on Learning Representa- tions, 2025

    2025

  13. [13]

    Quantum-train: Rethinking hybrid quantum-classical machine learning in the model compression perspective,

    C.-Y . Liuet al., “Quantum-train: Rethinking hybrid quantum-classical machine learning in the model compression perspective,”Quantum Machine Intelligence, vol. 7, no. 2, p. 80, 2025

    2025

  14. [14]

    Federated quantum kernel-based long short-term mem- ory for human activity recognition,

    Y .-C. Hsuet al., “Federated quantum kernel-based long short-term mem- ory for human activity recognition,”arXiv preprint arXiv:2508.06078, 2025

    work page arXiv 2025

  15. [15]

    Expressibility and entan- gling capability of parameterized quantum circuits for hybrid quantum- classical algorithms,

    S. Sim, P. D. Johnson, and A. Aspuru-Guzik, “Expressibility and entan- gling capability of parameterized quantum circuits for hybrid quantum- classical algorithms,”Advanced Quantum Technologies, vol. 2, no. 12, p. 1900070, 2019

    2019

  16. [16]

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

    Y .-C. Hsuet al., “Quantum kernel-based long short-term memory for climate time-series forecasting,” in2025 International Conference on Quantum Communications, Networking, and Computing (QCNC). IEEE, 2025, pp. 421–426

    2025

  17. [17]

    Quantum kernel-based long short-term memory,

    ——, “Quantum kernel-based long short-term memory,” in2025 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW). IEEE, 2025, pp. 1–5

    2025

  18. [18]

    Learning the hierarchy of steering measurement settings of qubit-pair states with kernel-based quantum models,

    Z.-L. Tsaiet al., “Learning the hierarchy of steering measurement settings of qubit-pair states with kernel-based quantum models,”New Journal of Physics, vol. 27, no. 9, p. 094502, 2025

    2025

  19. [19]

    Meta-learning for quantum optimization via quantum sequence model,

    Y .-C. Linet al., “Meta-learning for quantum optimization via quantum sequence model,”arXiv preprint arXiv:2512.05058, 2025

    work page arXiv 2025

  20. [20]

    Noise-enhanced quantum kernels on analog quantum computers

    H.-W. Huanget al., “Noise-enhanced quantum kernels on analog quan- tum computers,”arXiv preprint arXiv:2604.12476, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  21. [21]

    Quantum autoencoders for efficient compression of quantum data,

    J. Romero, J. P. Olson, and A. Aspuru-Guzik, “Quantum autoencoders for efficient compression of quantum data,”Quantum Science and Technology, vol. 2, no. 4, p. 045001, 2017

    2017

  22. [22]

    Supervised learning with quantum-enhanced feature spaces,

    V . Havl´ıˇceket al., “Supervised learning with quantum-enhanced feature spaces,”Nature, vol. 567, no. 7747, pp. 209–212, Mar 2019

    2019

  23. [23]

    A variational eigenvalue solver on a photonic quantum processor,

    A. Peruzzoet al., “A variational eigenvalue solver on a photonic quantum processor,”Nature communications, vol. 5, no. 1, p. 4213, 2014

    2014

  24. [24]

    The theory of variational hybrid quantum-classical algorithms,

    J. R. McClean, J. Romero, R. Babbush, and A. Aspuru-Guzik, “The theory of variational hybrid quantum-classical algorithms,”New Journal of Physics, vol. 18, no. 2, p. 023023, 2016

    2016

  25. [25]

    Quantum computational chemistry,

    S. McArdle, S. Endo, A. Aspuru-Guzik, S. C. Benjamin, and X. Yuan, “Quantum computational chemistry,”Reviews of Modern Physics, vol. 92, no. 1, p. 015003, 2020

    2020

  26. [26]

    Scalable quantum simulation of molecular energies,

    P. J. O’Malleyet al., “Scalable quantum simulation of molecular energies,”Physical Review X, vol. 6, no. 3, p. 031007, 2016

    2016

  27. [27]

    Quantum computing in the NISQ era and beyond,

    J. Preskill, “Quantum computing in the NISQ era and beyond,”Quan- tum, vol. 2, p. 79, 2018

    2018

  28. [28]

    Noise-induced barren plateaus in variational quantum algorithms,

    S. Wanget al., “Noise-induced barren plateaus in variational quantum algorithms,”Nature communications, vol. 12, no. 1, p. 6961, 2021

    2021

  29. [29]

    Barren plateaus in variational quantum computing,

    M. Laroccaet al., “Barren plateaus in variational quantum computing,” Nature Reviews Physics, vol. 7, no. 4, pp. 174–189, Apr. 2025

    2025

  30. [30]

    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

    2018

  31. [31]

    Challenges and opportunities in quantum machine learning,

    M. Cerezo, G. Verdon, H.-Y . Huang, L. Cincio, and P. J. Coles, “Challenges and opportunities in quantum machine learning,”Nature computational science, vol. 2, no. 9, pp. 567–576, 2022

    2022

  32. [32]

    Nakaji, L

    K. Nakajiet al., “The generative quantum eigensolver (gqe) and its application for ground state search,”arXiv preprint arXiv:2401.09253, 2024

    work page arXiv 2024

  33. [33]

    Generative quantum combinatorial opti- mization by means of a novel conditional generative quantum eigen- solver,

    S. Minami, K. Nakajiet al., “Generative quantum combinatorial opti- mization by means of a novel conditional generative quantum eigen- solver,”Digital Discovery, vol. 4, no. 8, pp. 2229–2243, 2025

    2025

  34. [34]

    Qaoa-gpt: Efficient generation of adaptive and regular quantum approximate optimization algorithm circuits,

    I. Tyaginet al., “Qaoa-gpt: Efficient generation of adaptive and regular quantum approximate optimization algorithm circuits,” in2025 IEEE International Conference on Quantum Computing and Engineering (QCE), vol. 1. IEEE, 2025, pp. 1505–1515

    2025

  35. [35]

    Auger spectroscopy via generative quantum eigen- solver: A quantum approach to molecular excitations,

    K. Keithleyet al., “Auger spectroscopy via generative quantum eigen- solver: A quantum approach to molecular excitations,”arXiv preprint arXiv:2603.12859, 2026

    work page arXiv 2026

  36. [36]

    Language models are unsupervised multitask learners,

    A. Radford, J. Wu, R. Child, D. Luan, D. Amodei, I. Sutskeveret al., “Language models are unsupervised multitask learners,”OpenAI blog, vol. 1, no. 8, p. 9, 2019

    2019

  37. [37]

    Attention is all you need,

    A. Vaswaniet al., “Attention is all you need,”Advances in neural information processing systems, vol. 30, 2017

    2017

  38. [38]

    Adapt-qsci: Adaptive construction of an input state for quantum-selected configuration interaction,

    Y . O. Nakagawa, M. Kamoshita, W. Mizukami, S. Sudo, and Y .- y. Ohnishi, “Adapt-qsci: Adaptive construction of an input state for quantum-selected configuration interaction,”Journal of Chemical Theory and Computation, vol. 20, no. 24, pp. 10 817–10 825, 2024

    2024

  39. [39]

    Quantum-selected configuration interaction with time-evolved state,

    M. Mikkelsen and Y . O. Nakagawa, “Quantum-selected configuration interaction with time-evolved state,”Physical Review Research, vol. 7, no. 4, p. 043043, 2025

    2025

  40. [40]

    Hamil- tonian simulation-based quantum-selected configuration interaction for large-scale electronic structure calculations with a quantum computer,

    K. Sugisaki, S. Kanno, T. Itoko, R. Sakuma, and N. Yamamoto, “Hamil- tonian simulation-based quantum-selected configuration interaction for large-scale electronic structure calculations with a quantum computer,” Physical Chemistry Chemical Physics, vol. 27, no. 38, pp. 20 869– 20 884, 2025

    2025

  41. [41]

    Critical limitations in quantum-selected configuration interaction methods,

    P. Reinholdt, K. M. Ziems, E. R. Kjellgren, S. Coriani, S. P. Sauer, and J. Kongsted, “Critical limitations in quantum-selected configuration interaction methods,”Journal of Chemical Theory and Computation, vol. 21, no. 14, pp. 6811–6822, 2025

    2025

  42. [42]

    Implicit sol- vent sample-based quantum diagonalization,

    D. Kaliakin, A. Shajan, F. Liang, and K. M. Merz Jr, “Implicit sol- vent sample-based quantum diagonalization,”The Journal of Physical Chemistry B, vol. 129, no. 23, pp. 5788–5796, 2025

    2025

  43. [43]

    Transformer feed-forward layers are key-value memories,

    M. Geva, R. Schuster, J. Berant, and O. Levy, “Transformer feed-forward layers are key-value memories,” inProceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2021, pp. 5484–5495

    2021

  44. [44]

    Quantum variational activation functions empower Kolmogorov - Arnold networks

    J.-C. Jiang, Y .-C. Huang, T. Chen, and H.-S. Goan, “Quantum variational activation functions empower Kolmogorov-Arnold networks,”arXiv preprint arXiv:2509.14026, 2025. [Online]. Available: https://arxiv.org/ abs/2509.14026

    work page arXiv 2025

  45. [45]

    Data re-uploading for a universal quantum classifier,

    A. P ´erez-Salinaset al., “Data re-uploading for a universal quantum classifier,”Quantum, vol. 4, p. 226, 2020

    2020

  46. [46]

    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

    2021

  47. [47]

    DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models

    Z. Shaoet al., “DeepSeekMath: Pushing the limits of mathematical reasoning in open language models,”arXiv preprint arXiv:2402.03300, 2024

    work page internal anchor Pith review arXiv 2024

  48. [48]

    Cuda quantum: The platform for integrated quantum- classical computing,

    J.-S. Kimet al., “Cuda quantum: The platform for integrated quantum- classical computing,” in2023 60th ACM/IEEE Design Automation Conference (DAC). IEEE, 2023, pp. 1–4

    2023

  49. [49]

    Unrestricted hartree–fock theory and its applications to molecules and chemical reactions,

    H. Fukutome, “Unrestricted hartree–fock theory and its applications to molecules and chemical reactions,”International Journal of Quantum Chemistry, vol. 20, no. 5, pp. 955–1065, 1981

    1981

  50. [50]

    Higher excitations in coupled-cluster theory,

    M. K ´allayet al., “Higher excitations in coupled-cluster theory,”The Journal of chemical physics, vol. 115, no. 7, pp. 2945–2954, 2001

    2001

  51. [51]

    Helgaker, P

    T. Helgaker, P. Jorgensen, and J. Olsen,Molecular electronic-structure theory. John Wiley & Sons, 2013

    2013

  52. [52]

    Heat-bath configuration interaction: An efficient selected configuration interaction algorithm inspired by heat-bath sam- pling,

    A. A. Holmeset al., “Heat-bath configuration interaction: An efficient selected configuration interaction algorithm inspired by heat-bath sam- pling,”Journal of chemical theory and computation, vol. 12, no. 8, pp. 3674–3680, 2016

    2016

  53. [53]

    Semistochastic heat-bath configuration interaction method: Selected configuration interaction with semistochastic pertur- bation theory,

    S. Sharmaet al., “Semistochastic heat-bath configuration interaction method: Selected configuration interaction with semistochastic pertur- bation theory,”Journal of chemical theory and computation, vol. 13, no. 4, pp. 1595–1604, 2017

    2017

  54. [54]

    The density matrix renormalization group in quantum chemistry,

    G. K.-L. Chan and S. Sharma, “The density matrix renormalization group in quantum chemistry,”Annual review of physical chemistry, vol. 62, no. 1, pp. 465–481, 2011

    2011

  55. [55]

    The density matrix renormalization group in chemistry and molecular physics: Recent developments and new challenges,

    A. Baiardi and M. Reiher, “The density matrix renormalization group in chemistry and molecular physics: Recent developments and new challenges,”The Journal of Chemical Physics, vol. 152, no. 4, 2020

    2020

  56. [56]

    Casscf with extremely large active spaces using the adaptive sampling configuration interaction method,

    D. S. Levine, D. Hait, N. M. Tubman, S. Lehtola, K. B. Whaley, and M. Head-Gordon, “Casscf with extremely large active spaces using the adaptive sampling configuration interaction method,”Journal of chemical theory and computation, vol. 16, no. 4, pp. 2340–2354, 2020

    2020

  57. [57]

    Distributed implementation of full configuration inter- action for one trillion determinants,

    H. Gaoet al., “Distributed implementation of full configuration inter- action for one trillion determinants,”Journal of Chemical Theory and Computation, vol. 20, no. 3, pp. 1185–1192, 2024

    2024

  58. [58]

    Numerically exact configuration interaction at quadrillion-determinant scale,

    A. Shayitet al., “Numerically exact configuration interaction at quadrillion-determinant scale,”Nature Communications, vol. 16, no. 1, p. 11016, 2025

    2025

  59. [59]

    A quantum computing view on unitary coupled cluster theory,

    A. Anandet al., “A quantum computing view on unitary coupled cluster theory,”Chemical Society Reviews, vol. 51, no. 5, pp. 1659–1684, 2022

    2022

  60. [60]

    Strategies for quantum computing molecular ener- gies using the unitary coupled cluster ansatz,

    J. Romeroet al., “Strategies for quantum computing molecular ener- gies using the unitary coupled cluster ansatz,”Quantum Science and Technology, vol. 4, no. 1, p. 014008, 2019

    2019

  61. [61]

    Experimental quantum computational chemistry with optimized unitary coupled cluster ansatz,

    S. Guoet al., “Experimental quantum computational chemistry with optimized unitary coupled cluster ansatz,”Nature Physics, vol. 20, no. 8, pp. 1240–1246, 2024

    2024

  62. [62]

    Hardware-efficient variational quantum eigensolver for small molecules and quantum magnets,

    A. Kandalaet al., “Hardware-efficient variational quantum eigensolver for small molecules and quantum magnets,”nature, vol. 549, no. 7671, pp. 242–246, 2017

    2017

  63. [63]

    Bridging physical intuition and hardware efficiency for correlated electronic states: the local unitary cluster jastrow ansatz for electronic structure,

    M. Mottaet al., “Bridging physical intuition and hardware efficiency for correlated electronic states: the local unitary cluster jastrow ansatz for electronic structure,”Chemical Science, vol. 14, no. 40, pp. 11 213– 11 227, 2023

    2023

  64. [64]

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

    H. R. Grimsleyet al., “An adaptive variational algorithm for exact molecular simulations on a quantum computer,”Nature communications, vol. 10, no. 1, p. 3007, 2019

    2019

  65. [65]

    qubit-adapt-vqe: An adaptive algorithm for construct- ing hardware-efficient ans¨atze on a quantum processor,

    H. L. Tanget al., “qubit-adapt-vqe: An adaptive algorithm for construct- ing hardware-efficient ans¨atze on a quantum processor,”PRX Quantum, vol. 2, no. 2, p. 020310, 2021

    2021

  66. [66]

    Layer vqe: A variational approach for combinatorial optimization on noisy quantum computers,

    X. Liuet al., “Layer vqe: A variational approach for combinatorial optimization on noisy quantum computers,”IEEE Transactions on Quantum Engineering, vol. 3, pp. 1–20, 2022

    2022

  67. [67]

    Contextual subspace variational quantum eigensolver calculation of the dissociation curve of molecular nitrogen on a super- conducting quantum computer,

    T. Weavinget al., “Contextual subspace variational quantum eigensolver calculation of the dissociation curve of molecular nitrogen on a super- conducting quantum computer,”npj Quantum Information, vol. 11, no. 1, p. 25, 2025

    2025

  68. [68]

    Variational quantum eigensolver boosted by adiabatic connection,

    M. Matouseket al., “Variational quantum eigensolver boosted by adiabatic connection,”The Journal of Physical Chemistry A, vol. 128, no. 3, pp. 687–698, 2024

    2024

  69. [69]

    Quantum architecture search: a survey,

    D. Martyniuket al., “Quantum architecture search: a survey,” in2024 IEEE International Conference on Quantum Computing and Engineer- ing (QCE), vol. 1. IEEE, 2024, pp. 1695–1706

    2024

  70. [70]

    Reinforcement learning for optimization of vari- ational quantum circuit architectures,

    M. Ostaszewskiet al., “Reinforcement learning for optimization of vari- ational quantum circuit architectures,”Advances in neural information processing systems, vol. 34, pp. 18 182–18 194, 2021

    2021

  71. [71]

    Quantum circuit optimization with deep reinforcement learning.arXiv preprint arXiv:2103.07585, 2021

    T. F ¨oselet al., “Quantum circuit optimization with deep reinforcement learning,”arXiv preprint arXiv:2103.07585, 2021

    work page arXiv 2021

  72. [72]

    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

  73. [73]

    Curriculum reinforcement learning for quantum architecture search under hardware errors

    Y . J. Patelet al., “Curriculum reinforcement learning for quantum archi- tecture search under hardware errors,”arXiv preprint arXiv:2402.03500, 2024

    work page arXiv 2024

  74. [74]

    Differentiable quantum architecture search,

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

    2022

  75. [75]

    Quantumdarts: differentiable quantum architecture search for variational quantum algorithms,

    W. Wuet al., “Quantumdarts: differentiable quantum architecture search for variational quantum algorithms,” inInternational conference on machine learning. PMLR, 2023, pp. 37 745–37 764

    2023

  76. [76]

    Quantumnas: Noise-adaptive search for robust quantum circuits,

    H. Wanget al., “Quantumnas: Noise-adaptive search for robust quantum circuits,” in2022 IEEE International Symposium on High-Performance Computer Architecture (HPCA). IEEE, 2022, pp. 692–708

    2022

  77. [77]

    Spingqe: A gen- erative quantum eigensolver for spin hamiltonians,

    A. Holden, M. H. Rahat, and N. O. O. Dade, “Spingqe: A gen- erative quantum eigensolver for spin hamiltonians,”arXiv preprint arXiv:2603.24298, 2026

    work page arXiv 2026

  78. [78]

    KAN: Kolmogorov–Arnold networks,

    Z. Liuet al., “KAN: Kolmogorov–Arnold networks,” inThe Thirteenth International Conference on Learning Representations, 2025

    2025

  79. [79]

    Kolmogorov–arnold graph neural networks for molecular property prediction,

    L. Li, Y . Zhang, G. Wang, and K. Xia, “Kolmogorov–arnold graph neural networks for molecular property prediction,”Nature Machine Intelligence, vol. 7, no. 8, pp. 1346–1354, 2025

    2025

  80. [80]

    KANQAS: Kolmogorov-Arnold network for quantum architecture search,

    A. Kunduet al., “KANQAS: Kolmogorov-Arnold network for quantum architecture search,”EPJ Quantum Technology, vol. 11, no. 1, p. 76, 2024

    2024

Showing first 80 references.