arXiv subjects

Uğur Teğin

Publications and source records attributed to Uğur Teğin.

At least 19 recordsLinked to original sources

arXiv2026

Deep Inverse-Designed Nanophotonic Processors with Structural Nonlinearity from Repeated Phase Encoding

Passive nanophotonic scattering regions implement linear optical transformations, and cascading input-independent regions alone does not create functional depth because the resulting transformations collapse into a single linear operator. Here, we introduce repeated phase encoding between inverse-designed passive transformations to generate an input-conditioned multilayer optical map without interlayer photodetection. Each re-encoding introduces additional phase-dependent optical pathways, producing structural nonlinearity with respect to the encoded variables while every scattering region remains passive and linear in the optical field. Under a controlled MNIST depth sweep, classification accuracy increases from 83.53\% with one layer to 93.61\% with seven layers, whereas the input-independent passive control saturates at 86.34\%. The depth trend also persists in a time-multiplexed CIFAR-10 patch model. We further realize three jointly trained $16\times16$ transformations, each independently implemented as an inverse-designed nanophotonic region, with relative complex transmission errors of 7.63\%, 7.59\%, and 8.80\%. The reconstructed electromagnetic stack reaches 91.79\% accuracy after phase calibration, compared with 91.95\% for its surrogate model. These results establish repeated input encoding as a route to functional depth in compact inverse-designed nanophotonic processors.

physics.optics↗

arXiv2026

Low-cost passive single-shot ultrafast imaging at 685 Gfps

Capturing ultrafast transient phenomena conventionally requires streak cameras or computational imaging based on compressed sensing, which lead to complex and costly systems. In this Letter, we demonstrate, to the best of our knowledge, the first fully passive single-shot ultrafast imaging architecture assembled entirely from off-the-shelf, low-cost components. A commercial microlens array combined with a stack of standard microscope cover glasses maps temporal information into multiple spatial channels, and a consumer-grade CMOS image sensor records all delayed replicas within a single camera exposure. The proposed system has a total hardware cost below US\$500 and captures the evolution of a picosecond laser pulse with a temporal sampling interval of 1.46~ps, an effective frame rate of 685~Gfps, and a sequence depth of ten frames. The temporal fidelity of the system is verified by recovering the expected Gaussian pulse profile, and the spatial resolution is characterized through a point-source measurement with a point spread function of 1.86 and 1.62 pixels full width at half maximum along the horizontal and vertical directions, respectively. The proposed architecture presents an alternative approach to single-shot ultrafast imaging with a simple, low-cost, computation-free, and fully passive design.

physics.optics↗

arXiv2026

Scalable Photonic Neural Networks via Surrogate Scattering-Matrix Inverse Design

Inverse-designed nanophotonic media are a promising platform for compact optical neural networks, but training them end to end is expensive because each adjoint iteration couples the full-wave solver to the dataset minibatch, so the number of electromagnetic simulations scales with both the network depth and the batch size. We introduce a two-stage surrogate workflow that decouples task learning from electromagnetic realization. In the first stage, the trainable optical block is represented as a passive complex matrix with bounded singular values and the classification task is solved directly in matrix space at negligible cost. In the second stage, the selected target operator is transferred to a fabrication-aware freeform device through an adjoint problem driven by a Frobenius-norm transmission residual and a reflection penalty, which removes the minibatch dependence from the full-wave loop and yields a smoother loss landscape than intensity-domain cross-entropy. We further introduce a banded-router architecture composed with a fixed evanescent-coupling region, which exploits the bandwidth-additive property of matrix products to realize dense effective operators within a design region roughly half as long as a fully local router would require. The framework is validated on three tasks. On MedMNIST, the realized all-optical classifier reproduces the surrogate accuracy within $0.6$ percentage points after only 20 adjoint epochs. On RSSCN7, the banded router plus evanescent stage improves test accuracy by more than 15 percentage points over a linear readout baseline. A Yin-Yang task confirms that the same framework supports nonlinear decision boundaries. These results indicate that surrogate-guided inverse design is a practical route to training compact photonic processors with simulation budgets orders of magnitude smaller than direct geometry-to-task pipelines.

physics.optics↗

arXiv2026

Resolution-Agnostic Lensless Imaging via Fourier Neural Operators

Lensless cameras based on thin diffusers offer a compact alternative to conventional refractive imaging but rely on computational reconstruction, since the diffuser's point spread function (PSF) globally multiplexes every scene point across the sensor. Here, we report a Fourier Neural Operator (FNO) framework for this reconstruction task. Because a linear shift-invariant forward model reduces to a pointwise multiplication in Fourier space, the spectral-domain kernel of an FNO layer is structurally aligned with the DiffuserCam inverse problem. Using a compact DiffuserCam prototype and a 25,000-image natural-scene dataset, our FNO improves upon a U-Net baseline of comparable parameter count by $2.14$~dB in PSNR and $0.11$ in SSIM. The same FNO, trained exclusively at $128 \times 128$, reconstructs $256 \times 256$ and $512 \times 512$ measurements with less than $1$~dB loss in PSNR and no retraining, demonstrating resolution-agnostic inference. The framework is directly applicable to other lensless modalities with global PSFs, such as multimode-fiber endoscopy.

physics.optics↗

arXiv2026

Recurrent neural networks implemented through spatiotemporal light propagation in optical fibers

Recurrent neural networks excel at temporal tasks and video processing but require energy-intensive sequential memory operations. We demonstrate that multimode optical fibers naturally implement spatiotemporal recurrent computation through passive light propagation. Video frames are encoded onto separate optical beams with controlled time delays; these beams combine and recirculate through a fiber loop where interference and nonlinear propagation generate high-dimensional states encoding both current inputs and fading memory. Remarkably, the entire optical system remains fixed with no trainable parameters or electronic feedback, yet this single physical configuration achieves competitive performance across diverse temporal and spatiotemporal learning tasks: chaotic time-series forecasting, human action recognition, steering angle prediction, and surgical skill assessment. Our results show that recurrent temporal processing can emerge directly from spatiotemporal wave dynamics. This paradigm shift from algorithmic to physical recurrence offers an energy-efficient pathway to temporal artificial intelligence by leveraging intrinsic spatiotemporal optical nonlinearities within multimode fibers.

physics.optics↗

arXiv2026

Multimode fiber laser cavities as nonlinear optical processors

Optical computing provides a promising path toward energy-efficient machine learning, yet implementing nonlinear transformations without complex electronics or high-power sources remains challenging. Here, we demonstrate that continuous-wave multimode fiber laser cavities can function as nonlinear optical processors. Input images encoded as phase patterns on a spatial light modulator undergo high-dimensional transformation through the interplay of multimode interference and gain saturation dynamics. The cavity maps input data into spatially stable, class-separable intensity distributions, enabling a simple linear classifier to achieve accuracies of 85--99\% across diverse benchmarks -- including medical imaging and remote sensing -- with orders of magnitude fewer trainable parameters than deep neural networks. Our results establish multimode fiber lasers as compact, low-power physical processors for scalable optical machine learning.

physics.optics↗

arXiv2026

Self-optimizing multichannel optical computing

Optical computing offers ultrafast, energy-efficient alternatives to conventional digital processors, yet most implementations remain confined to single-channel processing, severely underutilizing light's information capacity. Here we demonstrate a self-optimizing multichannel optical computing architecture based on multi-plane light conversion that natively processes RGB images and structured numerical data throughout the optical domain. We introduce two complementary optimization strategies that enable autonomous performance adaptation without differentiable forward models. First, Bayesian optimization tunes channel mixing coefficients to minimize crosstalk and enhance feature separability at the input level. Second, a hardware-in-the-loop protocol based on self-organized criticality leverages avalanche dynamics to autonomously navigate the high-dimensional phase landscape, enabling the system to self-optimize through stochastic multi-scale perturbations. Across medical imaging, natural image classification, and regression tasks, multichannel processing with random phase masks improves accuracy by 26--58 percentage points over raw pixel baselines, with RGB systematically outperforming grayscale by 5--6 percentage points. Self-optimization strategies provide additional gains of 6--7 percentage points through autonomous adaptation at complementary system levels. Our work establishes self-optimizing multichannel optical computing as a practical platform for real-world machine learning applications.

physics.optics↗

arXiv2025

Optical Spiking Neural Networks via Rogue-Wave Statistics

Optical computing could reduce the energy cost of artificial intelligence by leveraging the parallelism and propagation speed of light. However, implementing nonlinear activation, essential for machine learning, remains challenging in low-power optical systems dominated by linear wave physics. Here, we introduce an optical spiking neural network that uses optical rogue-wave statistics as a programmable firing mechanism. By establishing a homomorphism between free-space diffraction and neuronal integration, we demonstrate that phase-engineered caustics enable robust, passive thresholding: sparse spatial spikes emerge when the local intensity exceeds a significant-intensity rogue-wave criterion. Using a physics-informed digital twin, we optimize granular phase masks to deterministically concentrate energy into targeted detector regions, enabling end-to-end co-design of the optical transformation and a lightweight electronic readout. We experimentally validate the approach on BreastMNIST and Olivetti Faces, achieving accuracies of 82.45\% and 95.00\%, respectively, competitive with standard digital baselines. These results demonstrate that extreme-wave phenomena, often treated as deleterious fluctuations, can be harnessed as structural nonlinearity for scalable, energy-efficient neuromorphic photonic inference.

physics.optics↗

arXiv2025

The role of modes in nonlinear fiber optical computing

We investigate the nonlinear propagation of light in graded-index multimode fiber, utilizing it as an optical computing unit, and quantify how it employs waveguide modes to process information. Using a time-dependent spatiotemporal propagation model with modal decomposition, we evaluate several benchmark regression and classification tasks and study the modal content of the generated speckles, which couples with a simple digital layer to perform optical computing. Analysis of modal entropy and energy-based mode counts reveals that effective computation is confined to a low-dimensional modal subspace, whose identity depends on the task and propagation regime. This also sets a trade-off between modal richness and nonlinear beam self-cleaning. These results establish modal statistics as practical design metrics for fiber-based optical computers.

physics.optics↗

arXiv2025

Engineering Rogue Waves via Multimode Interactions in Integrated Waveguides

We explore rogue wave formation in multimode silicon nitride (Si$_3$N$_4$) waveguides with multimode nonlinear Schrödinger equation-based simulations. Pure fundamental-mode excitation produces smooth propagation without extreme events, whereas higher-order modes or multimode superpositions yield heavy-tailed statistics with bursts exceeding the $8σ$ threshold. These results reveal that rogue wave generation in integrated waveguides is controlled not only by material properties such as nonlinearity and dispersion but also by modal excitation and intermodal nonlinear interactions. Our results identify modal control as a new degree of freedom for engineering extreme spatiotemporal events on photonic chips, with implications for on-chip supercontinuum generation, frequency combs, and nonlinear wave management.

physics.optics↗

arXiv2025

Spatiotemporal Nonlinear Pulse Dynamics in Multimode Silicon Nitride Waveguides

We present an open-source multimode nonlinear Schrödinger equation-based simulation to investigate spatiotemporal nonlinear pulse propagation in thin-film silicon nitride (SiN) waveguides. Using this framework, we analyze femtosecond pulse evolution under diverse excitation conditions in a 6 μm wide SiN waveguide supporting six TE modes. Our results reveal that mode selection and power distribution critically govern nonlinear coupling, soliton fission, and dispersive wave generation, leading to broadband spectra exceeding 3 μm. Our findings reveal that input mode engineering is a powerful strategy for tailoring ultrafast nonlinear dynamics in integrated photonic platforms, with applications in supercontinuum generation, frequency combs, and programmable nonlinear optics.

physics.optics↗

arXiv2025

Real-time surrogate modeling of nonlinear pulse evolution in multimode fibers

Modeling nonlinear pulse propagation in multimode fibers is challenging due to the large number of interacting modes and the resulting spatiotemporal complexity. Traditional optimization methods often become intractable, while learning-based approaches, such as recurrent neural networks, suffer from high computational cost and long inference times. We present a U-Net architecture as a fast, accurate surrogate for modeling nonlinear pulse propagation in multimode fibers. This approach overcomes the intractability of traditional methods while offering low computational cost. Trained on data generated by beam propagation method, our approach achieves an $\sim$88\% average structural similarity index with simulations. The model can generalize to untrained propagation distances, demonstrating convolutional architectures as efficient tools for simulating complex spatiotemporal dynamics in multimode fibers and offering potential for applications like mode decomposition.

physics.optics↗

arXiv2025

Fiber-based diffractive deep neural network

Optical computing has reemerged as a promising alternative computing paradigm for providing energy-efficient information processing in the age of artificial intelligence. Among various photonic neural network platforms, diffractive optical processing systems in free space proved high-performance computing with high parallelism. Here, we report fiber-based diffractive deep neural networks by optimizing the linear coupling of the waveguide modes. Our approach demonstrated high performance in various machine learning tasks such as biomedical disease, fashion, and geospatial classification with a simple readout layer and all-optically. Operating on linear optics, our architecture performs on par with neural networks even in complex datasets where the data cannot be separated using linear operations. These results will enable efficient and scalable diffractive information processing with waveguides for real-life computing, telecommunications, and imaging applications.

physics.optics↗

arXiv2025

Genetically programmable optical random neural networks

Today, machine learning tools, particularly artificial neural networks, have become crucial for diverse applications. However, current digital computing tools to train and deploy artificial neural networks often struggle with massive data sizes and high power consumptions. Optical computing provides inherent parallelism accommodating high-resolution input data and performs fundamental operations with passive optical components. However, most of the optical computing platforms suffer from relatively low accuracies for machine learning tasks due to fixed connections while avoiding complex and sensitive techniques. Here, we demonstrate a genetically programmable yet simple optical neural network to achieve high performances with optical random projection. By genetically programming the orientation of the scattering medium which acts as a random projection kernel and only using 1% of the search space, our novel technique finds an optimum kernel and improves initial test accuracies by 8-41% for various machine learning tasks. Through numerical simulations and experiments on a number of datasets, we validate the programmability and high-resolution sample processing capabilities of our design. Our optical computing method presents a promising approach to achieve high performance in optical neural networks with a simple and scalable design.

cs.ET↗

arXiv2024

Photonic neural networks with spatiotemporal chaos in multimode fibers

Optical computing has gained significant attention as a potential solution to the growing computational demands of machine learning, particularly for tasks requiring large-scale data processing and high energy efficiency. Optical systems offer promising alternatives to digital neural networks by exploiting light's parallelism. This study explores a photonic neural network design using spatiotemporal chaos within grad-ed-index multimode fibers to improve machine learning performance. Through numerical simulations and experiments, we show that chaotic light propagation in multimode fibers enhances data classification accu-racy across domains, including biomedical imaging, fashion, and satellite geospatial analysis. This chaotic optical approach enables high-dimensional transformations, amplifying data separability and differentiation for greater accuracy. Fine-tuning parameters such as pulse peak power optimizes the reservoir's chaotic properties, highlighting the need for careful calibration. These findings underscore the potential of chaos-based nonlinear photonic neural networks to advance optical computing in machine learning, paving the way for efficient, scalable architectures.

physics.optics↗

arXiv2024

Optical computing with supercontinuum generation in photonic crystal fibers

We introduce a novel photonic neural network using photonic crystal fibers, leveraging femtosecond pulse supercontinuum generation for optical computing. Investigating its efficacy across machine learning tasks, we uncover the crucial impact of nonlinear pulse propagation dynamics on network performance. Our findings show that octave-spanning supercontinuum generation results in loss of dataset variety due to many-to-one mapping, and optimal performance requires balancing optical nonlinearity. This study offers guidance for designing energy-efficient and high-performance photonic neural network architectures by explaining the interplay between nonlinear dynamics and optical computing.

physics.optics↗

arXiv2023

Machine Learning with Chaotic Strange Attractors

Machine learning studies need colossal power to process massive datasets and train neural networks to reach high accuracies, which have become gradually unsustainable. Limited by the von Neumann bottleneck, current computing architectures and methods fuel this high power consumption. Here, we present an analog computing method that harnesses chaotic nonlinear attractors to perform machine learning tasks with low power consumption. Inspired by neuromorphic computing, our model is a programmable, versatile, and generalized platform for machine learning tasks. Our mode provides exceptional performance in clustering by utilizing chaotic attractors' nonlinear mapping and sensitivity to initial conditions. When deployed as a simple analog device, it only requires milliwatt-scale power levels while being on par with current machine learning techniques. We demonstrate low errors and high accuracies with our model for regression and classification-based learning tasks.

cs.LG↗

arXiv2022

Programming Nonlinear Propagation for Efficient Optical Learning Machines

The ever-increasing demand for processing data with larger machine learning models requires more efficient hardware solutions due to limitations such as power dissipation and scalability. Optics is a promising contender for providing lower power computation since light propagation through a non-absorbing medium is a lossless operation. However, to carry out useful and efficient computations with light, generating and controlling nonlinearity optically is a necessity that is still elusive. Multimode fibers (MMF) have been shown that they can provide nonlinear effects with microwatts of average power while maintaining parallelism and low loss. In this work, we propose an optical neural network architecture, which performs nonlinear optical computation by controlling the propagation of ultrashort pulses in MMF by wavefront shaping. With a surrogate model, optimal sets of parameters are found to program this optical computer for different tasks with minimal utilization of an electronic computer. We show a remarkable decrease of 97% in the number of model parameters, which leads to an overall 99% digital operation reduction compared to an equivalently performing digital neural network. We further demonstrate that a fully optical implementation can also be performed with competitive accuracies.

cs.ET↗