Pith. sign in

REVIEW 4 major objections 5 minor 66 references

Unwrapping photonic reservoirs: enhanced expressivity via random Fourier encoding over stretched domains

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Phase wrapping beyond 2π boosts photonic reservoir accuracy

desk verdict A simple, real idea with a clean mechanism and honest math—but the paper oversells an experiment that isn't there and leans on single-run simulations. read the letter →

arxiv 2506.01410 v1 pith:RB6MVW6T submitted 2025-06-02 physics.optics quant-ph

Gerard McCaul , Girish Tripathy , Giulia Marcucci , Juan Sebastian Totero Gongora This is my paper
classification physics.opticsquant-ph
keywords photonicreservoircomputingphasewrappingrandomFourierfeaturesscattering-assistedopticalnonlinearfrequencymixingextremelearningmachineneuromorphicexpressivepower
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

Photonic reservoir computers encode data into the phase of a light field, and standard practice restricts that phase to a single $[0,2\pi)$ cycle to preserve a one-to-one map from input to output. This paper argues that the opposite is better: deliberately wrapping the phase many times around the circle, by a factor $\alpha>1$, makes the reservoir substantially more expressive. In the paper's model, the output intensity becomes a weighted sum of random synthetic frequencies whose spacing is multiplied by $\alpha$, and this richer frequency spectrum lets a linear readout fit nonlinear targets that are unreachable at $\alpha=1$. The paper demonstrates the effect on a sinc-function regression (error dropping from NMSE $\approx 0.48$ to $\approx 2\times 10^{-6}$ at $\alpha\approx 3.9$) and on a two-spiral classification task, and shows it survives bit-depth-limited SLMs, CCDs, and shot noise. If true, the result says that improving a photonic reservoir can be as simple as turning up an encoding parameter, with no change to the scattering hardware.

What carries the argument

The load-bearing object is the intensity-detection identity of Eq. (18), $I_m(u_k)=\sum_{np} T_{mn}T_{mp}^* \exp[2\pi i\alpha(G_n-G_p)u_k]$, which converts a linear phase encoding into a nonlinear set of difference frequencies. The wrapping factor $\alpha$ acts as a global bandwidth control on that frequency set, and the random mask $G_n$ determines which difference frequencies are present and whether collisions in individual pixel phases can synchronize across the full output vector. The paper also uses the Fourier spectrum of the target function as the reference against which the reservoir's available modes are compared, and the Shannon entropy of the encoded phase distribution to quantify separability loss.

What would settle it

Measure the complex field after the scatterer with a phase-sensitive heterodyne detector instead of the intensity-only CCD, and sweep $\alpha$ on the sinc task; if the error still falls to $\sim 10^{-6}$ at $\alpha\approx 3.9$, then intensity-difference mixing (Eq. 18) is not the cause. Alternatively, numerically check whether two distinct inputs in the training set at the optimal $\alpha$ produce bit-identical quantized $N$-pixel intensity vectors; if they do, the separability premise is violated.

Watch

Extended reading notes

Core claim

The central discovery is that the canonical restriction of phase encoding to $[0,2\pi)$ is not a computational necessity but a discarded resource. Writing the input phase as $\varphi_n(u_k)=2\pi\alpha G_n u_k \mod 2\pi$, the intensity recorded by the $m$-th output pixel expands as $I_m(u_k)=\sum_{np} T_{mn}T_{mp}^* \exp[2\pi i\alpha(G_n-G_p)u_k]$. This identity shows that scattering plus intensity detection performs quadratic frequency-difference mixing: even if the mask $G_n$ supplies only a fixed set of encoding frequencies, the detection step generates the difference frequencies $\alpha(G_n-G_p)$. Increasing $\alpha$ stretches all these synthetic frequencies, giving the linear readout a wider and denser spectral palette with which to reconstruct the target function. The per-pixel phase map becomes non-bijective, but with a random continuous mask exact collisions are measure-zero, so the full output vector remains separable; only at very large $\alpha$ does dense mode aliasing make all outputs statistically identical. The paper supports this mechanism by Fourier-analyzing the encoding, transmitted, and detected fields at different $\alpha$, and by showing that the reservoir matrix rank does not track the performance gain.

Load-bearing premise

The argument depends on the assumption that with a random mask $G_n$, the wrapped encoding maps distinct inputs to distinct full intensity patterns even for $\alpha>1$, so that the extra frequencies are not cancelled by collisions in the output vector.

Editorial extensions

If this is right

  • Any phase-encoded scattering reservoir can in principle be improved on nonlinear regression and classification tasks by sweeping $\alpha$ upward to a task-specific optimum, with no hardware modification.
  • The optimum is finite: for very large $\alpha$ the interference sum self-averages and all readouts converge to the mean, so performance peaks and then degrades (the paper identifies this with exponential concentration).
  • The effect persists under 8-bit SLM phase quantization, 16-bit CCD detection, and shot noise, indicating compatibility with realistic experimental hardware.
  • Because the feature-space expansion is controlled by the mask's difference spectrum, the same mechanism maps onto Hamiltonian-encoded quantum reservoirs, where $\alpha$ plays the role of evolution time.
  • Reservoir matrix rank is not what changes with $\alpha$; the gain comes from the spectral content of the generated features, so the effect is expressivity rather than conditioning.

Reading between the lines

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

  • The same 'wrap beyond the natural period' principle should apply to any wave-based reservoir with phase encoding and square-law detection—acoustic, radio-frequency, or nonlinear—suggesting a general design rule for physical feature maps.
  • If the mechanism is purely spectral, the optimal $\alpha$ for a given task should be predictable from the mask's pairwise difference distribution, e.g., by maximizing the coverage of the target's Fourier support; a testable extension would be to optimize the mask jointly with $\alpha$.
  • The analogy with quantum Hamiltonian encoding implies that quantum reservoirs with qudit or oscillator evolution may show analogous non-monotonic expressivity as their evolution time passes the period, which could guide experiments beyond the usual short-time operating point.
  • For finite bit-depth masks, exact collisions are no longer measure-zero, so the optimal $\alpha$ and the achievable gain will depend on the specific mask realization; averaging over masks may be needed to guarantee the effect.
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

4 major / 5 minor

Summary. The paper proposes a scattering-assisted photonic reservoir encoding in which the input phase is deliberately wrapped by a factor α>1, i.e., φ_n(u_k)=2παG_nu_k mod 2π, instead of being restricted to the canonical [0,2π) interval. The central claim is that, despite the loss of single-pixel phase bijectivity, phase wrapping enhances reservoir expressivity and improves regression and classification performance. The mechanism is analyzed through Eq. (18), an exact expansion of the detected intensity as a sum of Fourier modes at frequency differences α(G_n−G_p), and is supported by numerical experiments on sinc regression and a two-spiral classification task, including bit-depth and shot-noise robustness checks.

Significance. If the result is robust, the paper identifies a simple, hardware-free resource: increasing the wrapping factor α can substantially improve the accuracy of phase-encoded scattering-assisted reservoirs. The analytic expansion in Eq. (18) is an exact and useful characterization of the feature space, and the paper makes an interesting connection to random Fourier features and quantum extreme learning machines. The bit-depth and shot-noise simulations are a valuable robustness check, and the central idea is falsifiable and easy to test in existing setups. However, the paper's current evidence is largely based on single realizations and lacks a rigorous treatment of collisions under quantization, so the significance is conditional on additional validation.

major comments (4)
  1. [Section III and Eqs. (5)–(8)] The load-bearing separability claim for α>1 is not established at the level of the implemented model. The argument that I_m(u_k)=I_m(u_k') has 'no deterministic solutions' for random continuous G_n is a probabilistic heuristic, while the simulations use a finite N and quantized SLM/CCD described by Eqs. (5)–(8); the entropy analysis in Fig. 2 concerns only the marginal phase distribution and does not bound collisions in the joint N-pixel output vector. Please provide either a finite-N bound or a quantitative collision analysis, or repeated-mask experiments showing that no two inputs map to identical or near-identical reservoir vectors across realistic N, b_SLM, and b_CCD.
  2. [Sections IV and VI, Figs. 3 and 6] All reported performance figures appear to come from a single realization of the random mask G and scattering matrix T, with no repeated-seed statistics. The optimal α values (e.g., α=3.9 for sinc regression and the near-100% spiral accuracy) could therefore be realization-dependent. Please report the distribution of NMSE and F1 over multiple random T and G realizations, with medians and interquartile ranges, to support the claim that the wrapping enhancement is a generic property of the encoding rather than a sample-specific artifact.
  3. [Section VII, Discussion] The discussion states that the wrapping enhancement 'was demonstrated both numerically and experimentally,' but the manuscript contains no experimental setup, experimental data, or experimental figures. This statement is unsupported and should either be removed or replaced by an actual experimental section; as written, it overstates the evidence contained in the paper.
  4. [Section V, Eq. (18)] The proposed mechanism of 'new synthetic frequencies' does not explain the improvement observed for the equispaced mask. For equispaced G_n, the differences G_n−G_p are integer multiples of the same fundamental spacing, so Eq. (18) does not generate any frequency values beyond those already present in the encoding; nevertheless Fig. 3(e) shows a clear enhancement. Please clarify whether the beneficial effect for equispaced masks is instead due to redistribution of amplitudes and phases onto existing frequency lines, and state this distinction explicitly alongside the random-mask case.
minor comments (5)
  1. [Section II, Eq. (6)] The notation I_M and I_SAT is inconsistent: Eq. (6) introduces I_SAT as the saturation intensity, while the text refers to I_M; Eq. (8) then uses I_MAX. Please define all quantities consistently.
  2. [Section II, Eq. (9)] The mod 2π in Eq. (9) is redundant for the physical field exp(iφ_n), but it matters for the phase PDF analysis in Fig. 2; please clarify whether φ_n denotes the actual SLM phase or a data-encoding observable.
  3. [Section IV, Fig. 3] The text gives 'α3.7' (apparently missing an equals sign) while the Fig. 3 caption and panel (b) state α=3.9; the optimal value and the MSE/NMSE nomenclature should be unified.
  4. [Section V, Eq. (18)] The mask entries G_n are real, so the conjugate G_p^* in Eq. (18) should be G_p; as written, the complex conjugate on the mask is a typo.
  5. [Section VI, Fig. 6] The y-axis of Fig. 6 is labeled F1 score while the text discusses classification accuracy; please clarify whether the reported quantity is accuracy or weighted F1 and use the terminology consistently.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central claims are supported by exact model equations and independent numerical experiments, with self-citations only contextual.

full rationale

The paper does not fit its target results into its parameters. The central expression, Eq. (18), is an exact algebraic expansion of the intensity readout derived from Eqs. (1)-(4): I_m(u_k) = sum_{np} T_{mn} T_{mp}^* exp[2pi i alpha (G_n - G_p^*) u_k]. This expansion is not an assumption equivalent to the claimed enhancement; it is a faithful consequence of the stated model. The reported improvements in sinc regression and spiral classification are obtained by sweeping the wrapping factor alpha in numerical simulations, with the ridge readout trained on the reservoir activations, rather than by tuning alpha to reproduce the target outputs. The optimal alpha values differ across tasks (3.9 for the sinc regression in Fig. 3, 7.8 in Fig. 4, 3.9 for the spiral classification in Fig. 6), which indicates the observed behavior is not a construction forcing the conclusion. The interpretation in terms of random Fourier features and frequency-difference mixing is post-hoc but grounded in Eq. (18), and the paper explicitly acknowledges the prior Random Fourier Features framework [50]. Self-citations such as [22] and [58] are used for conceptual analogy and dataset provenance, not as load-bearing uniqueness theorems or as substitutes for the model derivation. The Section III separability argument for random continuous masks is a probabilistic heuristic that is not fully proven for finite bit-depth, but this is a robustness gap, not circular reasoning: nothing in the argument is defined in terms of the claim it supports. The paper also states a limitation of the approach (exponential concentration for sufficiently large alpha), further showing that the derivation is not structured to guarantee the reported improvement by definition. Overall, no circular step is exhibited. The manuscript is self-contained in its analytical model and numerical demonstrations, so the appropriate circularity score is 0.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central claim rests on the standard scattering model, quadratic intensity detection, and a probabilistic injectivity assumption for the wrapped encoding. No new physical entities are introduced; the 'synthetic frequencies' are mathematical consequences of Eq. (18).

free parameters (1)
  • alpha (wrapping factor) = 3.9 for sinc, 3.9 for spiral (optimal values from sweep over [0.1, 20])
    The entire effect is controlled by α. The paper selects the value with best test performance on the sweep, making it a free parameter for each task.
assumptions (3)
  • domain assumption The scattering matrix T is a complex Gaussian random matrix with i.i.d. entries (mean 0, variance 1/2 per real/imaginary part).
    Invoked in Section II.A, Eq. (3); this is the standard model for a random scattering medium.
  • domain assumption The intensity detection is in the linear regime, I = |E|^2, before optional bit-depth/noise effects (Eq. 4).
    Assumed in the derivation of Eq. (18); a nonlinear camera response would alter the frequency mixing structure.
  • domain assumption For a random continuous mask G_n, the map from data u to the output intensity vector is injective with probability 1 even for α>1.
    Stated in Section III without a formal proof; this assumption underlies the claim that wrapping does not destroy separability. It is only a heuristic for finite bit depth.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unwrapping photonic reservoirs: enhanced expressivity via random Fourier encoding over stretched domains." pith.science (2026). https://pith.science/paper/RB6MVW6T

@misc{pith2026250601410,
  author       = {Pith},
  title        = {Pith review of: Unwrapping photonic reservoirs: enhanced expressivity via random Fourier encoding over stretched domains},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RB6MVW6T}},
  note         = {Machine review of arXiv:2506.01410}
}
abstract

Photonic Reservoir Computing (RC) systems leverage the complex propagation and nonlinear interaction of optical waves to perform information processing tasks. These systems employ a combination of optical data encoding (in the field amplitude and/or phase), random scattering, and nonlinear detection to generate nonlinear features that can be processed via a linear readout layer. In this work, we propose a novel scattering-assisted photonic reservoir encoding scheme where the input phase is deliberately wrapped multiple times beyond the natural period of the optical waves $[0,2\pi)$. We demonstrate that, rather than hindering nonlinear separability through loss of bijectivity, wrapping significantly improves the reservoir's prediction performance across regression and classification tasks that are unattainable within the canonical $2\pi$ period. We demonstrate that this counterintuitive effect stems from the nonlinear interference between sets of random synthetic frequencies introduced by the encoding, which generates a rich feature space spanning both the feature and sample dimensions of the data. Our results highlight the potential of engineered phase wrapping as a computational resource in RC systems based on phase encoding, paving the way for novel approaches to designing and optimizing physical computing platforms based on topological and geometric stretching.

Figures

Figures reproduced from arXiv: 2506.01410 by the authors.

Figure 1
Figure 1. FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2 [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. (a) illustrates the results for our regression task when considering this scenario with N = 100. As evident from the figure, the regression fails, resulting in an NMSE = 0.48. Quite interestingly, the prediction is not entirely random, but the predicted function resembles a sinc function defined on a smaller restricted domain. An unexpected result, however, oc￾curs when α is increased beyond the natural value, as sh… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6 [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

66 extracted references · 35 canonical work pages

  1. [1]

    author author D. J. \ Gauthier , author E. Bollt , author A. Griffith , \ and\ author J.-N. \ Barbosa ,\ title title Next generation reservoir computing , \ @noop journal journal Nature Communications \ volume 12 ,\ pages 5564 ( year 2021 ) NoStop

    work page 2021

  2. [2]

    Jaeger \ and\ author H

    author author H. Jaeger \ and\ author H. Haas ,\ title title Harnessing Nonlinearity : Predicting Chaotic Systems and Saving Energy in Wireless Communication , \ 10.1126/science.1091277 journal journal Science \ volume 304 ,\ pages 78--80 ( year 2004 ) ,\ note publisher: American Association for the Advancement of Science NoStop

    doi:10.1126/science.1091277

  3. [3]

    Maass , author T

    author author W. Maass , author T. Natschl \"a ger , \ and\ author H. Markram ,\ title title Real- Time Computing Without Stable States : A New Framework for Neural Computation Based on Perturbations , \ 10.1162/089976602760407955 journal journal Neural Computation \ volume 14 ,\ pages 2531--2560 ( year 2002 ) NoStop

    doi:10.1162/089976602760407955

  4. [4]

    Dale , author J

    author author M. Dale , author J. Dewhirst , author S. O'Keefe , author A. Sebald , author S. Stepney , \ and\ author M. A. \ Trefzer ,\ title title The Role of Structure and Complexity on Reservoir Computing Quality , \ in\ 10.1007/978-3-030-19311-9_6 booktitle Unconventional Computation and Natural Computation ,\ Vol.\ volume 11493 ,\ editor edited by\ ...

    work page doi:10.1007/978-3-030-19311-9_6

  5. [5]

    Marcucci , author D

    author author G. Marcucci , author D. Pierangeli , \ and\ author C. Conti ,\ title title Theory of Neuromorphic Computing by Waves : Machine Learning by Rogue Waves , Dispersive Shocks , and Solitons , \ 10.1103/PhysRevLett.125.093901 journal journal Physical Review Letters \ volume 125 ,\ pages 093901 ( year 2020 ) NoStop

    doi:10.1103/physrevlett.125.093901

  6. [6]

    Dong , author S

    author author J. Dong , author S. Gigan , author F. Krzakala , \ and\ author G. Wainrib ,\ title title Scaling up echo-state networks with multiple light scattering , \ in\ 10.1109/SSP.2018.8450698 booktitle 2018 IEEE Statistical Signal Processing Workshop (SSP) \ ( publisher IEEE ,\ address Freiburg im Breisgau, Germany ,\ year 2018 )\ pp.\ pages 448--452 NoStop

    work page arXiv 2018

  7. [7]

    author author M. Lukoševičius ,\ title title A Practical Guide to Applying Echo State Networks , \ in\ 10.1007/978-3-642-35289-8_36 booktitle Neural Networks : Tricks of the Trade ,\ Vol.\ volume 7700 ,\ editor edited by\ editor G. Montavon , editor G. B. \ Orr , \ and\ editor K.-R. \ Müller \ ( publisher Springer Berlin Heidelberg ,\ address Berlin, Heid...

    work page doi:10.1007/978-3-642-35289-8_36

  8. [8]

    Tanaka , author T

    author author G. Tanaka , author T. Yamane , author J. B. \ H \'e roux , author R. Nakane , author N. Kanazawa , author S. Takeda , author H. Numata , author D. Nakano , \ and\ author A. Hirose ,\ title title Recent advances in physical reservoir computing: A review , \ 10.1016/j.neunet.2019.03.005 journal journal Neural Networks \ volume 115 ,\ pages 100...

    doi:10.1016/j.neunet.2019.03.005

Show all 66 references
  1. [9]

    Kudithipudi , author C

    author author D. Kudithipudi , author C. Schuman , author C. M. \ Vineyard , author T. Pandit , author C. Merkel , author R. Kubendran , author J. B. \ Aimone , author G. Orchard , author C. Mayr , author R. Benosman , author J. Hays , author C. Young , author C. Bartolozzi , ...

  2. [10]

    Markovi \'c , author A

    author author D. Markovi \'c , author A. Mizrahi , author D. Querlioz , \ and\ author J. Grollier ,\ title title Physics for neuromorphic computing , \ 10.1038/s42254-020-0208-2 journal journal Nature Reviews Physics \ volume 2 ,\ pages 499--510 ( year 2020 ) NoStop

    doi

  3. [11]

    Roy , author A

    author author K. Roy , author A. Jaiswal , \ and\ author P. Panda ,\ title title Towards spike-based machine intelligence with neuromorphic computing , \ 10.1038/s41586-019-1677-2 journal journal Nature \ volume 575 ,\ pages 607--617 ( year 2019 ) NoStop

    doi

  4. [12]

    Xu , author T

    author author Z. Xu , author T. Zhou , author M. Ma , author C. Deng , author Q. Dai , \ and\ author L. Fang ,\ title title Large-scale photonic chiplet Taichi empowers 160- TOPS / W artificial general intelligence , \ 10.1126/science.adl1203 journal journal Science \ volume 3...

    doi

  5. [13]

    Fujii \ and\ author K

    author author K. Fujii \ and\ author K. Nakajima ,\ title title Harnessing disordered-ensemble quantum dynamics for machine learning , \ 10.1103/PhysRevApplied.8.024030 journal journal Physical Review Applied \ volume 8 ,\ pages 024030 ( year 2017 ) NoStop

    doi

  6. [14]

    Ghosh , author A

    author author S. Ghosh , author A. Opala , author M. Matuszewski , author T. Paterek , \ and\ author T. C. H. \ Liew ,\ title title Quantum reservoir processing , \ 10.1038/s41534-019-0149-8 journal journal npj Quantum Information \ volume 5 ,\ pages 35 ( year 2019 ) NoStop

    doi

  7. [15]

    Ballarini , author A

    author author D. Ballarini , author A. Gianfrate , author R. Panico , author A. Opala , author S. Ghosh , author L. Dominici , author V. Ardizzone , author M. De Giorgi , author G. Lerario , author G. Gigli , author T. C. H. \ Liew , author M. Matuszewski , \ and\ author D. Sa...

    doi

  8. [16]

    Nokkala , author R

    author author J. Nokkala , author R. Mart \' nez-Pe \ n a , author G. L. \ Giorgi , author V. Parigi , author M. C. \ Soriano , \ and\ author R. Zambrini ,\ title title Gaussian states of continuous-variable quantum systems provide universal and versatile reservoir computing ,...

    doi

  9. [17]

    author author J. Nokkala ,\ title title Online quantum time series processing with random oscillator networks , \ 10.1038/s41598-023-34811-7 journal journal Scientific Reports \ volume 13 ,\ pages 7694 ( year 2023 ) NoStop

    2023 doi

  10. [18]

    author author L. C. G. \ Govia , author S. M. \ Szo ke , author T. J. \ Elliott , author D. K. \ O’Keeffe , author B. R. \ Patton , author P. R. \ Weightman , author F. M. \ Miatto , \ and\ author P.-L. \ Garc \'i a ,\ title title Quantum reservoir computing with a single nonl...

    doi

  11. [19]

    Angelatos , author S

    author author G. Angelatos , author S. A. \ Khan , \ and\ author H. E. \ T \"u reci ,\ title title Reservoir computing approach to quantum state measurement , \ 10.1103/PhysRevX.11.041062 journal journal Physical Review X \ volume 11 ,\ pages 041062 ( year 2021 ) NoStop

    doi

  12. [20]

    McCaul , author K

    author author G. McCaul , author K. Jacobs , \ and\ author D. I. \ Bondar ,\ title title Towards single atom computing via high harmonic generation , \ 10.1140/epjp/s13360-023-03649-3 journal journal The European Physical Journal Plus \ volume 138 ,\ pages 123 ( year 2023 ) NoStop

    doi

  13. [21]

    author author W. D. \ Kalfus , author G. J. \ Ribeill , author G. E. \ Rowlands , author H. K. \ Krovi , author T. A. \ Ohki , \ and\ author L. C. G. \ Govia ,\ title title Neuromorphic computing with a single qudit , \ @noop journal journal arXiv e-prints \ ,\ pages arXiv:210...

    2021 arXiv

  14. [22]

    McCaul , author J

    author author G. McCaul , author J. S. \ Totero Gongora , author W. Otieno , author S. Savelev , author A. Zagoskin , \ and\ author A. Balanov ,\ title title Minimal Quantum Reservoirs with Hamiltonian Encoding , \ @noop journal journal submitted \ ( year 2025 ) NoStop

    2025

  15. [23]

    author author J. C. \ Gartside , author K. D. \ Stenning , author A. Vanstone , author H. H. \ Holder , author D. M. \ Arroo , author T. Dion , author F. Caravelli , author H. Kurebayashi , \ and\ author W. R. \ Branford ,\ title title Reconfigurable training and reservoir com...

    doi

  16. [24]

    Lee , author T

    author author O. Lee , author T. Wei , author K. D. \ Stenning , author J. C. \ Gartside , author D. Prestwood , author S. Seki , author A. Aqeel , author K. Karube , author N. Kanazawa , author Y. Taguchi , author C. Back , author Y. Tokura , author W. R. \ Branford , \ and\ ...

    doi

  17. [25]

    Nerenberg , author O

    author author S. Nerenberg , author O. D. \ Neill , author G. Marcucci , \ and\ author D. Faccio ,\ title title Photon number-resolving quantum reservoir computing , \ 10.1364/OPTICAQ.553294 journal journal Optica Quantum \ volume 3 ,\ pages 201--210 ( year 2025 ) NoStop

    doi

  18. [26]

    author author T. W. \ Hughes , author I. A. D. \ Williamson , author M. Minkov , \ and\ author S. Fan ,\ title title Wave physics as an analog recurrent neural network , \ 10.1126/sciadv.aay6946 journal journal Science Advances \ volume 5 ,\ pages eaay6946 ( year 2019 ) NoStop

    doi

  19. [27]

    Wang , author Y

    author author D. Wang , author Y. Nie , author G. Hu , author H. K. \ Tsang , \ and\ author C. Huang ,\ title title Ultrafast silicon photonic reservoir computing engine delivering over 200 TOPS , \ 10.1038/s41467-024-55172-3 journal journal Nature Communications \ volume 15 ,...

    doi

  20. [28]

    Brunner , author B

    author author D. Brunner , author B. J. \ Shastri , author M. A. A. \ Qadasi , author H. Ballani , author S. Barbay , author S. Biasi , author P. Bienstman , author S. Bilodeau , author W. Bogaerts , author F. B \"o hm , author G. Brennan , author S. Buckley , author X. Cai , ...

    doi arXiv

  21. [29]

    author author P. L. \ McMahon ,\ title title The physics of optical computing , \ 10.1038/s42254-023-00645-5 journal journal Nature Reviews Physics \ volume 5 ,\ pages 717--734 ( year 2023 ) NoStop

    2023 doi

  22. [30]

    Pierangeli , author G

    author author D. Pierangeli , author G. Marcucci , \ and\ author C. Conti ,\ title title Photonic extreme learning machine by free-space optical propagation , \ 10.1364/PRJ.423531 journal journal Photonics Research \ volume 9 ,\ pages 1446--1454 ( year 2021 ) NoStop

    doi

  23. [31]

    Jaurigue , author J

    author author J. Jaurigue , author J. Robertson , author A. Hurtado , author L. Jaurigue , \ and\ author K. Lüdge ,\ title title Post-processing methods for delay embedding and feature scaling of reservoir computers , \ 10.1038/s44172-024-00330-0 journal journal Communications...

    doi

  24. [32]

    author author B. J. \ Shastri , author A. N. \ Tait , author T. Ferreira de Lima , author W. H. P. \ Pernice , author H. Bhaskaran , author C. D. \ Wright , \ and\ author P. R. \ Prucnal ,\ title title Photonics for artificial intelligence and neuromorphic computing , \ 10.103...

    doi

  25. [33]

    author author C. M. \ Valensise , author I. Grecco , author D. Pierangeli , \ and\ author C. Conti ,\ title title Large-scale photonic natural language processing , \ 10.1364/PRJ.472932 journal journal Photonics Research \ volume 10 ,\ pages 2846--2853 ( year 2022 ) NoStop

    doi

  26. [34]

    Wang , author K

    author author Z. Wang , author K. M \"u ller , author M. Filipovich , author J. Launay , author R. Ohana , author G. Pariente , author S. Mokaadi , author C. Brossollet , author F. Moreau , author A. Cappelli , author I. Poli , author I. Carron , author L. Daudet , author F. K...

    2024 arXiv

  27. [35]

    Wang , author J

    author author H. Wang , author J. Hu , author Y. Baek , author K. Tsuchiyama , author M. Joly , author Q. Liu , \ and\ author S. Gigan ,\ title title Optical next generation reservoir computing , \ 10.48550/arXiv.2404.07857 journal journal arXiv \ ( year 2024 c ),\ 10.48550/ar...

    doi arXiv

  28. [36]

    Olivieri , author A

    author author L. Olivieri , author A. R. \ Cooper , author L. Peters , author V. Cecconi , author A. Pasquazi , author M. Peccianti , \ and\ author J. S. \ Totero Gongora ,\ title title Adiabatic Energetic Annealing via Dual Single-Pixel Detection in an Optical Nonlinear Ising...

    doi

  29. [37]

    Talukder , author A

    author author R. Talukder , author A. Skalli , author X. Porte , author S. Thorpe , \ and\ author D. Brunner ,\ title title A spiking photonic neural network of 40.000 neurons, trained with rank-order coding for leveraging sparsity , \ 10.48550/arXiv.2411.19209 journal journal...

    doi

  30. [38]

    Gigan ,\ title title Imaging and computing with disorder , \ 10.1038/s41567-022-01681-1 journal journal Nature Physics \ volume 18 ,\ pages 980--985 ( year 2022 ) NoStop

    author author S. Gigan ,\ title title Imaging and computing with disorder , \ 10.1038/s41567-022-01681-1 journal journal Nature Physics \ volume 18 ,\ pages 980--985 ( year 2022 ) NoStop

    2022 doi

  31. [39]

    Rafayelyan , author J

    author author M. Rafayelyan , author J. Dong , author Y. Tan , author F. Krzakala , \ and\ author S. Gigan ,\ title title Large- Scale Optical Reservoir Computing for Spatiotemporal Chaotic Systems Prediction , \ @noop journal journal arXiv:2001.09131 [physics] \ ( year 2020 )...

    2001 arXiv

  32. [40]

    Dong , author M

    author author J. Dong , author M. Rafayelyan , author F. Krzakala , \ and\ author S. Gigan ,\ title title Optical Reservoir Computing Using Multiple Light Scattering for Chaotic Systems Prediction , \ 10.1109/JSTQE.2019.2936281 journal journal IEEE Journal of Selected Topics i...

    2019

  33. [41]

    Dong , author R

    author author J. Dong , author R. Ohana , author M. Rafayelyan , \ and\ author F. Krzakala ,\ title title Reservoir Computing meets Recurrent Kernels and Structured Transforms , \ @noop journal journal arXiv:2006.07310 [cs, eess, stat] \ ( year 2020 b ) ,\ http://arxiv.org/abs...

    2006 arXiv

  34. [42]

    Pierangeli , author V

    author author D. Pierangeli , author V. Palmieri , author G. Marcucci , author C. Moriconi , author G. Perini , author M. De Spirito , author M. Papi , \ and\ author C. Conti ,\ title title Living optical random neural network with three dimensional tumor spheroids for cancer ...

    doi

  35. [43]

    Bauwens , author K

    author author I. Bauwens , author K. Harkhoe , author P. Bienstman , author G. Verschaffelt , \ and\ author G. V. \ der Sande ,\ title title Influence of the input signal’s phase modulation on the performance of optical delay-based reservoir computing using semiconductor laser...

    doi

  36. [44]

    author author R. M. \ Nguimdo , author G. Verschaffelt , author J. Danckaert , \ and\ author G. V. \ der Sande ,\ title title Reducing the phase sensitivity of laser-based optical reservoir computing systems , \ 10.1364/OE.24.001238 journal journal Optics Express \ volume 24 ,...

    doi

  37. [45]

    McCaul , author A

    author author G. McCaul , author A. F. \ King , \ and\ author D. I. \ Bondar ,\ title title Non- Uniqueness of Driving Fields Generating Non - Linear Optical Response , \ 10.1002/andp.202100523 journal journal Annalen der Physik \ volume 534 ,\ pages 2100523 ( year 2022 ) ,\ n...

    doi

  38. [46]

    Masur , author D

    author author J. Masur , author D. I. \ Bondar , \ and\ author G. McCaul ,\ title title Optical distinguishability of Mott insulators in the time versus frequency domain , \ 10.1103/PhysRevA.106.013110 journal journal Physical Review A \ volume 106 ,\ pages 013110 ( year 2022 ) NoStop

    doi

  39. [47]

    McCaul , author A

    author author G. McCaul , author A. F. \ King , \ and\ author D. I. \ Bondar ,\ title title Optical Indistinguishability via Twinning Fields , \ 10.1103/PhysRevLett.127.113201 journal journal Physical Review Letters \ volume 127 ,\ pages 113201 ( year 2021 ) NoStop

    doi

  40. [48]

    McCaul , author C

    author author G. McCaul , author C. Orthodoxou , author K. Jacobs , author G. H. \ Booth , \ and\ author D. I. \ Bondar ,\ title title Driven Imposters : Controlling Expectations in Many - Body Systems , \ 10.1103/PhysRevLett.124.183201 journal journal Phys. Rev. Lett. \ volum...

    doi

  41. [49]

    McCaul , author C

    author author G. McCaul , author C. Orthodoxou , author K. Jacobs , author G. H. \ Booth , \ and\ author D. I. \ Bondar ,\ title title Controlling arbitrary observables in correlated many-body systems , \ 10.1103/PhysRevA.101.053408 journal journal Phys. Rev. A \ volume 101 ,\...

    doi

  42. [50]

    Rahimi \ and\ author B

    author author A. Rahimi \ and\ author B. Recht ,\ title title Random Features for Large-Scale Kernel Machines , \ in\ @noop booktitle Advances in Neural Information Processing Systems ,\ Vol. volume 20 \ ( publisher Curran Associates, Inc. ,\ year 2007 ) NoStop

    2007

  43. [51]

    author author J. W. \ Goodman ,\ @noop title Statistical Optics ,\ edition second edition \ ed.,\ Wiley Series in Pure and Applied Optics\ ( publisher John Wiley & Sons Inc ,\ address Hoboken, New Jersey ,\ year 2015 ) NoStop

    2015

  44. [52]

    Kumar , author V

    author author V. Kumar , author V. Cecconi , author L. Peters , author J. Bertolotti , author A. Pasquazi , author J. S. \ Totero Gongora , \ and\ author M. Peccianti ,\ title title Deterministic Terahertz Wave Control in Scattering Media , \ 10.1021/acsphotonics.2c00061 journ...

    doi

  45. [53]

    He , author X

    author author Z. He , author X. Sui , author G. Jin , author D. Chu , \ and\ author L. Cao ,\ title title Optimal quantization for amplitude and phase in computer-generated holography , \ 10.1364/OE.414160 journal journal Optics Express \ volume 29 ,\ pages 119--133 ( year 202...

    doi

  46. [54]

    author author J. R. \ Janesick ,\ 10.1117/3.374903 title Scientific Charge-Coupled Devices \ ( publisher SPIE ,\ address 1000 20th Street, Bellingham, WA 98227-0010 USA ,\ year 2001 ) NoStop

    2001 doi

  47. [55]

    Hastie , author R

    author author T. Hastie , author R. Tibshirani , \ and\ author J. Friedman ,\ @noop title The Elements of Statistical Learning: Data Mining, Inference, and Prediction \ ( publisher Springer ,\ year 2009 ) NoStop

    2009

  48. [56]

    author author T. Hastie ,\ title title Ridge Regularization : An Essential Concept in Data Science , \ 10.1080/00401706.2020.1791959 journal journal Technometrics \ volume 62 ,\ pages 426--433 ( year 2020 ) NoStop

    2020

  49. [57]

    Te g in , author M

    author author U. Te g in , author M. Y ld r m , author \.I . O g uz , author C. Moser , \ and\ author D. Psaltis ,\ title title Scalable optical learning operator , \ 10.1038/s43588-021-00112-0 journal journal Nature Computational Science \ volume 1 ,\ pages 542--549 ( year 20...

    doi

  50. [58]

    Marcucci , author L

    author author G. Marcucci , author L. Olivieri , \ and\ author J. S. \ Totero Gongora ,\ title title Optimising complexity and learning for photonic reservoir computing with gain-controlled multimode fibres , \ @noop journal journal submitted \ ( year 2025 ) NoStop

    2025

  51. [59]

    \ Huang , author L

    author author G.-B. \ Huang , author L. Chen , \ and\ author C.-K. \ Siew ,\ title title Universal Approximation using Incremental Constructive Feedforward Networks with Random Hidden Nodes , \ 10.1109/TNN.2006.875977 journal journal IEEE Transactions on Neural Networks \ volu...

    2006

  52. [60]

    Xiong , author G

    author author W. Xiong , author G. Facelli , author M. Sahebi , author O. Agnel , author T. Chotibut , author S. Thanasilp , \ and\ author Z. Holmes ,\ title title On fundamental aspects of quantum extreme learning machines , \ 10.1007/s42484-025-00239-7 journal journal Quantu...

    doi

  53. [61]

    Schuld , author R

    author author M. Schuld , author R. Sweke , \ and\ author J. J. \ Meyer ,\ title title Effect of data encoding on the expressive power of variational quantum-machine-learning models , \ 10.1103/PhysRevA.103.032430 journal journal Phys. Rev. A \ volume 103 ,\ pages 032430 ( yea...

    doi

  54. [62]

    Saeed , author M

    author author S. Saeed , author M. Müftüoglu , author G. R. \ Cheeran , author T. Bocklitz , author B. Fischer , \ and\ author M. Chemnitz ,\ title title Nonlinear Inference Capacity of Fiber - Optical Extreme Learning Machines , \ 10.48550/arXiv.2501.18894 journal journal arX...

    doi arXiv

  55. [63]

    Bertschinger \ and\ author T

    author author N. Bertschinger \ and\ author T. Natschl \"a ger ,\ title title Real- Time Computation at the Edge of Chaos in Recurrent Neural Networks , \ 10.1162/089976604323057443 journal journal Neural Computation \ volume 16 ,\ pages 1413--1436 ( year 2004 ) NoStop

    doi

  56. [64]

    Rahimi \ and\ author B

    author author A. Rahimi \ and\ author B. Recht ,\ title title Weighted sums of random kitchen sinks: Replacing minimization with randomization in learning , \ in\ @noop booktitle Advances in Neural Information Processing Systems ,\ Vol. volume 21 ,\ editor edited by\ editor D....

    2008

  57. [65]

    McCaul , author D

    author author G. McCaul , author D. V. \ Zhdanov , \ and\ author D. I. \ Bondar ,\ title title Wave operator representation of quantum and classical dynamics , \ 10.1103/PhysRevA.108.052208 journal journal Physical Review A \ volume 108 ,\ pages 052208 ( year 2023 ) ,\ note pu...

    doi

  58. [66]

    author author M. M. \ Bronstein , author J. Bruna , author T. Cohen , \ and\ author P. Veli c kovi \'c ,\ 10.48550/arXiv.2104.13478 title Geometric Deep Learning : Grids , Groups , Graphs , Geodesics , and Gauges , \ ( year 2021 ),\ http://arxiv.org/abs/2104.13478 arXiv:2104.1...

    doi arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.