arxiv: 2605.04945 · v1 · submitted 2026-05-06 · 🪐 quant-ph

Recognition: unknown

Beyond Gates: Pulse Level Quantum Fourier Models

Achim Streit, Eileen Kuehn, Lukas Scheller, Maja Franz, Melvin Strobl, Wolfgang Mauerer

Pith reviewed 2026-05-08 16:16 UTC · model grok-4.3

classification 🪐 quant-ph

keywords quantum Fourier modelspulse levelvariational quantum algorithmsquantum machine learningoptimization landscapemonomial couplingsexpressibilitytrainability

0 comments

The pith

Pulse-level parameters in quantum Fourier models replace single gate angles with multiple independent sub-angles, relaxing rigid monomial couplings and improving training.

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

The paper examines quantum Fourier models at the pulse level instead of the conventional gate level. It shows that while overall expressibility and Fourier coefficient correlations stay similar, the local optimization landscape changes because composite gates now admit several independently tunable pulse scalings. These scalings break the fixed linkages among monomial terms that gate-level angles enforce. Gradient descent therefore gains additional dimensions for navigating the loss surface, which decouples local constraints and produces markedly better training performance. A sympathetic reader would care because improved trainability could make variational quantum machine learning more feasible on present-day hardware.

Core claim

Control over pulse shapes does not significantly alter the global expressibility or structural correlations of the Ansatz, but it fundamentally alters the local optimisation landscape. For composite gates, independent pulse scalings replace a single logical angle by multiple independently tunable sub-angles. This relaxes the rigid monomial couplings induced by the gate-level parameterisation, and provides gradient descent with higher-dimensional escape routes, decoupling local parameter constraints and significantly boosting performance during training. An analytical proof is followed by numerical validation on a QFM with an exponential (ternary) feature map trained on a Fourier series with

What carries the argument

Independent pulse scalings applied to the microwave parameters of composite gates, which replace one logical angle with multiple tunable sub-angles and thereby relax the monomial couplings in the quantum Fourier expansion.

If this is right

Gradient descent obtains higher-dimensional escape routes from local minima during training.
Local parameter constraints become decoupled, allowing the model to reach better minima.
Global expressibility and Fourier coefficient correlations remain essentially unchanged from gate-level versions.
The performance boost is observed when fitting a Fourier series whose frequencies match those of an exponential ternary feature map.

Where Pith is reading between the lines

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

Pulse-level control might improve trainability in other variational quantum circuits that are not restricted to Fourier models.
Hardware-native implementations could avoid some compilation overhead that gate abstractions introduce in quantum machine learning.
The decoupling mechanism suggests testing whether similar gains appear when noise models or different pulse shapes are introduced.

Load-bearing premise

The analytical mapping from pulse scalings to relaxed monomial couplings continues to hold when realistic hardware constraints such as noise and pulse distortions are present, and that the observed gains on the ternary feature map extend to other Fourier series and feature maps.

What would settle it

Training the same QFM on the identical Fourier series with gate-level parameters versus pulse-level scalings and finding that final loss values and convergence rates show no advantage for the pulse version would falsify the claim that the sub-angle relaxation improves the optimization landscape.

Figures

Figures reproduced from arXiv: 2605.04945 by Achim Streit, Eileen Kuehn, Lukas Scheller, Maja Franz, Melvin Strobl, Wolfgang Mauerer.

**Figure 1.** Figure 1: Illustration of our contribution. The optimisation view at source ↗

**Figure 3.** Figure 3: Count of active frequencies, calculated by the number view at source ↗

**Figure 2.** Figure 2: MSE for all ansätze examined in this work, trained either only using unitary gate parameters (without and with a decomposition into basis gates), or in combination with pulse parameters. Results represent the average over ten different seeds used for both the model and data initialisation. Standard deviation is reported via error bars. The results confirm our theory from Sec. II that access to the pulse pa… view at source ↗

**Figure 4.** Figure 4: Fidelity between gate- and pulse-level simulations for view at source ↗

**Figure 5.** Figure 5: MSE over the course of training for different ansätze view at source ↗

**Figure 6.** Figure 6: FCC and KL-divergence DKL (Eq. 20, i.e., inverse of the expressibility) over pulse parameter variance σ 2 λ (coloured) for different ansätze view at source ↗

read the original abstract

In the domain of variational quantum algorithms, quantum Fourier models (QFMs) provide a mathematically well defined structure for quantum machine learning (QML). There has been a substantial amount of work on the scalability and trainability of such models showcasing the potential but also the limitations for the prospective application of QFMs. However, much less is known in the context of pulse-level quantum computing, where the microwave parameters that implement unitary operations on the hardware are used to perform computations directly instead of through the interface of quantum circuits. In this work, we evaluate QFMs through the lens of pulse parameters and link metrics such as expressibility and Fourier coefficient correlation (FCC) to this extended set of variational parameters. We show that while control over pulse shapes does not significantly alter the global expressibility or structural correlations of the Ansatz, it fundamentally alters the local optimisation landscape. For composite gates, independent pulse scalings replace a single logical angle by multiple independently tunable sub-angles. This relaxes the rigid monomial couplings induced by the gate-level parameterisation, and provides gradient descent with higher-dimensional escape routes, decoupling local parameter constraints and significantly boosting performance during training. Following an analytical proof, we show numerical results validating our theory on training a QFM with an exponential (ternary) feature map on a Fourier series with the same frequencies.

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

Pulse-level parameterization relaxes monomial couplings on composite gates and improves local optimization for QFMs, but the numerical checks are narrow and skip realistic hardware constraints.

read the letter

The main point is that this paper shows pulse-level scalings on composite gates replace one logical angle with multiple independent sub-angles. That change loosens the rigid monomial couplings built into gate-level QFMs and gives gradient descent higher-dimensional paths to escape poor local minima. They lay this out with an analytical mapping and test it on a ternary exponential feature map for a Fourier series with matching frequencies. They also connect the pulse parameters to expressibility and Fourier coefficient correlation without seeing big shifts in those global measures. The hardware-native angle is the useful step here. It moves the discussion from abstract circuits toward actual control parameters, which matters for anyone trying to run these models on real devices. The analytical link between pulse scalings and relaxed couplings is clear enough on paper and directly targets the trainability problems that have been discussed in the gate-level QFM work. The soft spots sit in the validation. The numerics stay with one feature map and one series, with no ablations or reported error bars. The abstract and available details do not address how amplitude limits, fixed pulse durations, or crosstalk would affect the claimed independence of the sub-angles. Those constraints could easily re-couple the effective parameters and shrink the escape routes the authors highlight. A fair parameter-count comparison between the gate and pulse versions is also missing. This work is for people in quantum machine learning who focus on variational models and their optimization landscapes. A reader already following QFM trainability issues would pick up the core idea quickly. I would send it to peer review. The direction is honest and builds on existing limitations in the literature, even though the experiments need more breadth and hardware realism to carry the performance claims.

Referee Report

3 major / 3 minor

Summary. The paper claims that in variational quantum algorithms, pulse-level control of quantum Fourier models (QFMs) improves trainability over gate-level approaches. For composite gates, independent scalings of pulse parameters replace a single logical angle with multiple tunable sub-angles. This relaxes rigid monomial couplings in the Fourier expansion, provides higher-dimensional escape routes for gradient descent, and boosts performance. The claim is supported by an analytical proof followed by numerical validation on training a QFM using an exponential (ternary) feature map for a Fourier series with matching frequencies. The work also links pulse parameters to metrics such as expressibility and Fourier coefficient correlation (FCC).

Significance. If the analytical mapping and numerical results hold under realistic conditions, the work identifies a concrete mechanism by which pulse-level parameterization can decouple local constraints in QFMs without altering global expressibility or structural correlations. This could inform hardware-aware variational quantum machine learning designs. The explicit connection between pulse scalings and relaxed monomial couplings in the Fourier series is a useful conceptual contribution, though the narrow scope of the numerical experiments limits broader applicability.

major comments (3)

[Analytical proof (referenced in abstract and introduction)] The central claim rests on an analytical proof that independent pulse scalings on composite gates replace one logical angle by multiple independently tunable sub-angles, thereby relaxing monomial couplings. No equations, derivation steps, or explicit mapping from pulse parameters to the Fourier coefficients are provided, making it impossible to verify the claimed relaxation of couplings or the higher-dimensional escape routes for gradient descent.
[Numerical results section] Numerical validation is performed only on an exponential (ternary) feature map for one specific Fourier series. No error bars, details on the optimization algorithm, learning rates, data exclusion criteria, or ablations that enforce hardware constraints (amplitude/duration limits, crosstalk) are reported. This leaves open whether the reported performance boost is statistically significant or an artifact of unconstrained parameters.
[Discussion of pulse-level parameterization] The analysis treats pulse parameters as fully independent and unconstrained. Realistic hardware constraints (fixed pulse durations, amplitude bounds, channel crosstalk) are not modeled; these could re-couple the effective sub-angles and reduce the dimensionality of the optimization landscape, directly affecting the load-bearing claim about decoupled local parameter constraints.

minor comments (3)

[Metrics and methods] Clarify the precise definition and computation of the Fourier coefficient correlation (FCC) metric when applied to pulse parameters rather than gate angles.
[Abstract and results] The abstract states that control over pulse shapes 'does not significantly alter the global expressibility'; provide quantitative comparison (e.g., expressibility values or plots) between gate-level and pulse-level models to support this.
[Figures] Ensure all figures include axis labels, legends, and error bars where applicable; the current description of numerical results lacks these details.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thorough review and valuable comments on our manuscript. We address each of the major comments below and outline the revisions we will make to strengthen the paper. We believe these changes will clarify the analytical foundations and improve the presentation of the numerical results while maintaining the core contributions regarding pulse-level parameterization in quantum Fourier models.

read point-by-point responses

Referee: [Analytical proof (referenced in abstract and introduction)] The central claim rests on an analytical proof that independent pulse scalings on composite gates replace one logical angle by multiple independently tunable sub-angles, thereby relaxing monomial couplings. No equations, derivation steps, or explicit mapping from pulse parameters to the Fourier coefficients are provided, making it impossible to verify the claimed relaxation of couplings or the higher-dimensional escape routes for gradient descent.

Authors: We agree that the submitted manuscript lacks sufficient detail in presenting the analytical proof. The derivation begins from the pulse-level implementation of a composite gate, where the effective unitary is parameterized by independent scaling factors on the pulse amplitudes or durations, replacing the single angle theta in the gate-level model with a set of sub-angles {alpha_k * theta}. This leads to the Fourier series expansion where the monomial terms (e.g., products of cos(theta) and sin(theta)) are relaxed to independent products involving the sub-angles. We will include the full step-by-step derivation, including the explicit mapping to the Fourier coefficients and how this provides additional degrees of freedom for gradient descent, in the revised version of the paper. revision: yes
Referee: [Numerical results section] Numerical validation is performed only on an exponential (ternary) feature map for one specific Fourier series. No error bars, details on the optimization algorithm, learning rates, data exclusion criteria, or ablations that enforce hardware constraints (amplitude/duration limits, crosstalk) are reported. This leaves open whether the reported performance boost is statistically significant or an artifact of unconstrained parameters.

Authors: We acknowledge the need for more rigorous reporting in the numerical section. In the revision, we will add error bars from multiple independent runs with different random seeds, specify the optimization algorithm (gradient descent with a fixed learning rate), and include details on the training procedure. While we will not perform full hardware-constrained ablations in this work as it focuses on the idealized case, we will note this as a limitation and suggest it for future studies. These additions will help establish the statistical significance of the observed performance improvements. revision: partial
Referee: [Discussion of pulse-level parameterization] The analysis treats pulse parameters as fully independent and unconstrained. Realistic hardware constraints (fixed pulse durations, amplitude bounds, channel crosstalk) are not modeled; these could re-couple the effective sub-angles and reduce the dimensionality of the optimization landscape, directly affecting the load-bearing claim about decoupled local parameter constraints.

Authors: The referee correctly identifies that our analysis assumes unconstrained pulse parameters to isolate the effect of independent sub-angles. In realistic hardware, constraints could indeed reintroduce couplings. However, the primary contribution is to demonstrate the mechanism by which pulse-level control can relax monomial couplings in principle. We will expand the discussion to explicitly state this assumption and its implications, and add a paragraph on how hardware constraints might affect the results, without altering the core claim for the idealized setting. revision: partial

Circularity Check

0 steps flagged

No significant circularity; analytical derivation and numerical validation are independent

full rationale

The paper derives the relaxation of monomial couplings via an explicit analytical mapping from independent pulse scalings to multiple sub-angles on composite gates, then validates the resulting optimization benefit with separate numerical experiments on a ternary feature map. No step reduces a claimed prediction or uniqueness result to a fitted parameter, self-citation chain, or definitional renaming; the central claim is supported by a self-contained proof whose assumptions are stated separately from the numerical outcomes.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review performed on abstract only; full paper may contain additional parameters or assumptions not visible here. No explicit free parameters, axioms, or invented entities are stated in the provided text.

axioms (1)

domain assumption Standard assumptions of variational quantum algorithms and pulse-level control models hold without additional hardware noise or calibration errors.
Implicit in the transition from gate-level to pulse-level analysis.

pith-pipeline@v0.9.0 · 5545 in / 1245 out tokens · 54695 ms · 2026-05-08T16:16:14.001558+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

44 extracted references · 35 canonical work pages · 1 internal anchor

[2]

George Arfken.Mathematical Methods for Physicists. Third. San Diego: Academic Press, Inc., 1985

1985
[3]

Bayerstadler, G

Andreas and Bayerstadler, Guillaume Becquin, Julia Binder, et al. “Industry quantum computing applications”. In:EPJ Quantum Technology8.1 (Nov. 2021).DOI: 10.1140/epjqt/s40507-021-00114-x

work page doi:10.1140/epjqt/s40507-021-00114-x 2021
[4]

Parameterized quantum circuits as machine learning models

Marcello Benedetti, Erika Lloyd, Stefan Sack, et al. “Parameterized quantum circuits as machine learning models”. In:Quantum science and technology4.4 (2019), p. 043001

2019
[5]

Challenges for Quantum Software Engineering: An Industrial Application Scenario Perspective

Cecilia Carbonelli, Michael Felderer, Matthias Jung, et al. “Challenges for Quantum Software Engineering: An Industrial Application Scenario Perspective”. In: Quantum Software: Aspects of Theory and System Design. Ed. by Iaakov Exman, Ricardo Perez-Castillo, Mario Piattini, et al. Springer Nature, 2024, pp. 311–335. DOI: 10.1007/978-3-031-64136-7_12

work page doi:10.1007/978-3-031-64136-7_12 2024
[6]

doi:10.1038/s41467-021- 27138-2 (2021)

M. Cerezo, Akira Sone, Tyler V olkoff, et al. “Cost Func- tion Dependent Barren Plateaus in Shallow Parametrized Quantum Circuits”. In:Nature Communications12.1 (Mar. 19, 2021), p. 1791.DOI: 10.1038/s41467-021- 21728-w

work page doi:10.1038/s41467-021- 2021
[7]

Spectral bias in variational quantum machine learning,

Callum Duffy and Marcin Jastrzebski. “Spectral Bias in Variational Quantum Machine Learning”. In:arXiv preprint arXiv:2506.22555(2025)

work page arXiv 2025
[8]

Qibo: A Framework for Quantum Simula- tion with Hardware Acceleration

Stavros Efthymiou, Sergi Ramos-Calderer, Carlos Bravo- Prieto, et al. “Qibo: A Framework for Quantum Simula- tion with Hardware Acceleration”. In:Quantum Science and Technology7.1 (Jan. 2022), p. 015018.DOI: 10. 1088/2058-9565/ac39f5

2022
[9]

Spectral analysis for noise diagnostics and filter-based digital error mitigation

Enrico Fontana, Ivan Rungger, Ross Duncan, et al. Spectral analysis for noise diagnostics and filter-based digital error mitigation. Tech. rep. arXiv, Nov. 2022. DOI: 10.48550/arXiv.2206.08811

work page doi:10.48550/arxiv.2206.08811 2022
[10]

Out of Tune: Demystifying Noise-Effects on Quantum Fourier Models

Maja Franz, Melvin Strobl, Leonid Chaichenets, et al. Out of Tune: Demystifying Noise-Effects on Quantum Fourier Models. 2025.DOI: 10.48550/arxiv.2506.09527

work page doi:10.48550/arxiv.2506.09527 2025
[11]

Software Between Quantum and Machine Learning - And Down to Pulses

Maja Franz, Melvin Strobl, Jonathan Hunz, Lucas van der Horst, Lukas Scheller, Eileen Kuehn, Achim Streit, and Wolfgang Mauerer. “Software Between Quantum and Machine Learning - And Down to Pulses”. In:To be published(2026)

2026
[12]

Statistics (international student edition)

David Freedman, Robert Pisani, and Roger Purves. “Statistics (international student edition)”. In:Pisani, R. Purves, 4th edn. WW Norton & Company, New York (2007)

2007
[13]

Quantum Data Encoding Patterns and their Conse- quences

Martin Gogeissl, Hila Safi, and Wolfgang Mauerer. “Quantum Data Encoding Patterns and their Conse- quences”. In:Proceedings of the 1st Workshop on Quantum Computing and Quantum-Inspired Technology for Data-Intensive Systems and Applications. Q-Data ’24. Santiago, AA, Chile: Association for Computing Machinery, 2024, pp. 27–37.DOI: 10.1145/3665225. 3665446

work page doi:10.1145/3665225 2024
[14]

Effects of Imperfections on Quantum Algorithms: A Software Engineering Perspective

Felix Greiwe, Tom Krüger, and Wolfgang Mauerer. “Effects of Imperfections on Quantum Algorithms: A Software Engineering Perspective”. In:IEEE Interna- tional Conference on Quantum Software (QSW). IEEE, 2023, pp. 31–42.DOI: 10.1109/QSW59989.2023.00014

work page doi:10.1109/qsw59989.2023.00014 2023
[15]

2018 , journal =

Jeff Heaton. “Ian Goodfellow, Yoshua Bengio, and Aaron Courville: Deep Learning”. In:Genetic Programming and Evolvable Machines19.1 (June 2018), pp. 305–307. DOI: 10.1007/s10710-017-9314-z

work page doi:10.1007/s10710-017-9314-z 2018
[16]

Gentile, Youssef Achari Berrada, et al.Let Quantum Neural Networks Choose Their Own Frequencies

Ben Jaderberg, Antonio A. Gentile, Youssef Achari Berrada, et al.Let Quantum Neural Networks Choose Their Own Frequencies. Tech. rep. 4. arXiv, Sept. 2023, p. 042421.DOI: 10.48550/arXiv.2309.03279

work page doi:10.48550/arxiv.2309.03279 2023
[17]

A quantum engineer’s guide to superconducting qubits

P. Krantz, M. Kjaergaard, F. Yan, et al. “A quantum engineer’s guide to superconducting qubits”. In:Applied Physics Reviews6.2 (June 2019).DOI: 10 . 1063 / 1 . 5089550

2019
[18]

The Annals of Mathematical Statistics , author =

S. Kullback and R. A. Leibler. “On Information and Sufficiency”. In:The Annals of Mathematical Statistics 22.1 (Mar. 1951). Publisher: Institute of Mathematical Statistics, pp. 79–86.DOI: 10.1214/aoms/1177729694

work page doi:10.1214/aoms/1177729694 1951
[19]

Jonas Landman, Slimane Thabet, Constantin Dalyac, et al.Classically Approximating Variational Quantum Machine Learning with Random Fourier Features. Tech. rep. arXiv, Oct. 2022.DOI: 10.48550/arXiv.2210.13200

work page doi:10.48550/arxiv.2210.13200 2022
[20]

Nature Computational Science , volume=

Martin Larocca, Nathan Ju, Diego García-Martín, et al. “Theory of Overparametrization in Quantum Neural Networks”. In:Nature Computational Science3.6 (June 2023), pp. 542–551.DOI: 10.1038/s43588-023-00467-6

work page doi:10.1038/s43588-023-00467-6 2023
[21]

Jiaqi Leng, Yuxiang Peng, Yi-Ling Qiao, et al.Differen- tiable Analog Quantum Computing for Optimization and Control. Oct. 2022.DOI: 10.48550/arXiv.2210.15812

work page doi:10.48550/arxiv.2210.15812 2022
[22]

Zhiding Liang, Jinglei Cheng, Hang Ren, et al.NAPA: Intermediate-level Variational Native-pulse Ansatz for Variational Quantum Algorithms. Jan. 2024.DOI: 10. 48550/arXiv.2208.01215

work page arXiv 2024
[23]

From Pulses to Circuits and Back Again: A Quantum Optimal Control Perspective on Variational Quantum Algorithms

Alicia B. Magann, Christian Arenz, Matthew D. Grace, et al. “From Pulses to Circuits and Back Again: A Quantum Optimal Control Perspective on Variational Quantum Algorithms”. In:PRX Quantum2.1 (Jan. 2021), p. 010101.DOI: 10.1103/PRXQuantum.2.010101

work page doi:10.1103/prxquantum.2.010101 2021
[24]

Make Some Noise! Measuring Noise Model Quality in Real-World Quantum Software

Stefan Raimund Maschek, Jürgen Schwittalla, Maja Franz, et al. “Make Some Noise! Measuring Noise Model Quality in Real-World Quantum Software”. In: Proceedings of the IEEE International Conference on Quantum Software (QSW). IEEE, 2025, pp. 1–11.DOI: 10.1109/QSW67625.2025.00010

work page doi:10.1109/qsw67625.2025.00010 2025
[25]

1-2-3 Reproducibility for Quantum Software Experiments

Wolfgang Mauerer and Stefanie Scherzinger. “1-2-3 Reproducibility for Quantum Software Experiments”. In: IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER). 2022, pp. 1247– 1248.DOI: 10.1109/SANER53432.2022.00148

work page doi:10.1109/saner53432.2022.00148 2022
[26]

Hela Mhiri, Leo Monbroussou, Mario Herrero-Gonzalez, et al.Constrained and Vanishing Expressivity of Quan- tum Fourier Models. Mar. 2024.DOI: 10.48550/arXiv. 2403.09417

work page internal anchor Pith review doi:10.48550/arxiv 2024
[27]

Quantum circuit learning

Kosuke Mitarai, Makoto Negoro, Masahiro Kitagawa, et al. “Quantum circuit learning”. In:Physical Review A98.3 (2018), p. 032309

2018
[28]

Motzoi, J

F. Motzoi, J. M. Gambetta, P. Rebentrost, et al. “Simple Pulses for Elimination of Leakage in Weakly Nonlinear Qubits”. In:Physical Review Letters103.11 (Sept. 2009), p. 110501.DOI: 10.1103/PhysRevLett.103.110501

work page doi:10.1103/physrevlett.103.110501 2009
[29]

Data re-uploading for a universal quantum classifier,

Adrián Pérez-Salinas, Alba Cervera-Lierta, Elies Gil- Fuster, et al. “Data re-uploading for a universal quantum classifier”. In:Quantum4 (Feb. 2020), p. 226.DOI: 10.22331/q-2020-02-06-226

work page doi:10.22331/q-2020-02-06-226 2020
[30]

Guided-SPSA: Simultaneous Perturba- tion Stochastic Approximation Assisted by the Parameter Shift Rule

Maniraman Periyasamy, Axel Plinge, Christopher Mutschler, et al. “Guided-SPSA: Simultaneous Perturba- tion Stochastic Approximation Assisted by the Parameter Shift Rule”. In:IEEE International Conference on Quantum Computing and Engineering (QCE). IEEE, 2024, pp. 1504–1515.DOI: 10.1109/QCE60285.2024. 00177

work page doi:10.1109/qce60285.2024 2024
[31]

Evan Peters and Maria Schuld.Generalization despite overfitting in quantum machine learning models. Tech. rep. arXiv, Sept. 2022.DOI: 10.48550/arXiv.2209.05523

work page doi:10.48550/arxiv.2209.05523 2022
[32]

A Lie Algebraic Theory of Barren Plateaus for Deep Parameterized Quantum Circuits

Michael Ragone, Bojko N. Bakalov, Frédéric Sauvage, et al. “A Lie Algebraic Theory of Barren Plateaus for Deep Parameterized Quantum Circuits”. In:Nature Communications15.1 (Aug. 2024), p. 7172.DOI: 10. 1038/s41467-024-49909-3

2024
[33]

Evaluation of Pulse Level Quan- tum Fourier Models

Tilmann Rothe Santos. “Evaluation of Pulse Level Quan- tum Fourier Models”. 46.21.02; LK 01. Abschlussarbeit - Bachelor. Karlsruher Institut für Technologie (KIT), 2025.DOI: 10.5445/IR/1000184129

work page doi:10.5445/ir/1000184129 2025
[34]

Evaluating analytic gradients on quantum hardware,

Maria Schuld, Ville Bergholm, Christian Gogolin, et al. “Evaluating Analytic Gradients on Quantum Hardware”. In:Physical Review A99.3 (Mar. 21, 2019), p. 032331. DOI: 10.1103/PhysRevA.99.032331

work page doi:10.1103/physreva.99.032331 2019
[35]

doi:10.1103/physreva.103.032430

Maria Schuld, Ryan Sweke, and Johannes Jakob Meyer. “The effect of data encoding on the expressive power of variational quantum machine learning models”. In: Physical Review A103.3 (Mar. 2021), p. 032430.DOI: 10.1103/PhysRevA.103.032430

work page doi:10.1103/physreva.103.032430 2021
[36]

Communication in the Presence of Noise

C.E. Shannon. “Communication in the Presence of Noise”. In:Proceedings of the IRE37.1 (1949), pp. 10– 21.DOI: 10.1109/JRPROC.1949.232969

work page doi:10.1109/jrproc.1949.232969 1949
[37]

Expressibility and entangling capability of parameter- ized quantum circuits for hybrid quantum-classical al- gorithms,

Sukin Sim, Peter D. Johnson, and Alán Aspuru-Guzik. “Expressibility and Entangling Capability of Parameter- ized Quantum Circuits for Hybrid Quantum-Classical Algorithms”. en. In:Advanced Quantum Technologies 2.12 (2019), p. 1900070.DOI: 10.1002/qute.201900070

work page doi:10.1002/qute.201900070 2019
[38]

Version v1.0.0

Melvin Strobl and Maja Franz.cirKITers/pulse-level- quantum-fourier-models: Initial Release. Version v1.0.0. Apr. 2026.DOI: 10.5281/zenodo.19847940

work page doi:10.5281/zenodo.19847940 2026
[39]

2025.DOI: 10.48550/arXiv.2506.06695

Melvin Strobl, Maja Franz, Eileen Kuehn, et al.QML Essentials – A framework for working with Quantum Fourier Models. 2025.DOI: 10.48550/arXiv.2506.06695

work page doi:10.48550/arxiv.2506.06695 2025
[40]

Emre Sahin, Lucas van der Horst, et al

Melvin Strobl, M. Emre Sahin, Lucas van der Horst, et al. Fourier Fingerprints of Ansatzes in Quantum Machine Learning. 2025.DOI: 10.48550/arxiv.2508.20868

work page doi:10.48550/arxiv.2508.20868 2025
[41]

Potential and limitations of random Fourier features for dequantizing quantum machine learning

Ryan Sweke, Erik Recio-Armengol, Sofiene Jerbi, et al. “Potential and limitations of random Fourier features for dequantizing quantum machine learning”. In:Quantum9 (Feb. 2025), p. 1640.DOI: 10.22331/q-2025-02-20-1640

work page doi:10.22331/q-2025-02-20-1640 2025
[42]

Predict and Con- quer: Navigating Algorithm Trade-Offs with Quantum Design Automation

Simon Thelen and Wolfgang Mauerer. “Predict and Con- quer: Navigating Algorithm Trade-Offs with Quantum Design Automation”. In:IEEE International Conference on Quantum Computing and Engineering (QCE). Los Alamitos, CA, USA: IEEE Computer Society, 2025, pp. 591–602.DOI: 10.1109/QCE65121.2025.00071

work page doi:10.1109/qce65121.2025.00071 2025
[43]

Fourier anal- ysis of variational quantum circuits for supervised learning,

Marco Wiedmann, Maniraman Periyasamy, and Daniel D. Scherer.Fourier Analysis of Variational Quantum Circuits for Supervised Learning. Nov. 2024.DOI: 10. 48550/arXiv.2411.03450

work page arXiv 2024
[44]

General Parameter-Shift Rules for Quantum Gradients

David Wierichs, Josh Izaac, Cody Wang, et al. “General Parameter-Shift Rules for Quantum Gradients”. In: Quantum6 (Mar. 2022), p. 677.DOI: 10 . 22331 / q - 2022-03-30-677

2022
[45]

Connectivity for Quantum Graphs via Quantum Adjacency Operators

Tao Yue, Wolfgang Mauerer, Shaukat Ali, et al. “Chal- lenges and Opportunities in Quantum Software Archi- tecture”. In:Software Architecture: Research Roadmaps from the Community. Ed. by Patrizio Pelliccione, Rick Kazman, Ingo Weber, et al. Cham: Springer Nature Switzerland, 2023, pp. 1–23.DOI: 10.1007/978-3-031- 36847-9_1

work page doi:10.1007/978-3-031- 2023