In this paper we present a simple scheme of dual frequency comb generation, based on a large-signal electrical modulation of a single high-speed semiconductor laser, applicable for symmetric spectroscopic system architectures. Our simulation results, based on a detailed numerical model including parasitic and noise effects, are confirmed through an experiment, reporting almost 40 GHz wide dual optical frequency comb, with 160 beat notes of 10 MHz spacing.

The threads of photonics are eagerly awaited to redefine the future of neuromorphic data processing, especially as computing-intensive artificial intelligence models become an unavoidable part of our everyday lives. However, there is much to be improved within the domain of photonic nonlinear activation functions, as programmable, all-optical, energy-efficient nonlinearities remain beyond the grasp of the current state-of-the-art. In this paper, we address the issue at hand and propose a novel approach in the realization of high-performance all-optical photonic activations. Through simulations and experiments, we show that Fabry-P & eacute;rot laser diodes (FP-LDs) exhibit richness and high programmability of their nonlinear response to input optical pulses with widths as low as 25 ps. We demonstrate a variety of sigmoid-like and inverted PReLU-like trends to be used as all-optical activation functions in photonic neural networks, testing their performance in stringent, real-life training scenarios with randomized data patterns at repetition rates of up to 10 GHz. The programmability of activations is shown using a multitude of experimental operating parameters, among which we highlight the power variation of an additional continuous-wave laser injected into the FP-LD, enriching our approach with all-optical control of all-optical activations. With very low static power consumption of our active element, we achieve a record-breaking energy draw on the order of pJ to hundreds of fJ per nonlinear operation. (c) 2025 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC) license (https://creativecommons.org/licenses/by-nc/4.0/).

In this paper, we present a simple optical frequency comb generation technique based on the optimized structure consisting of a gain-switched DFB laser and reflective semiconductor optical amplifier cascade. Bias currents of the two devices are tailored as multi-harmonic waveforms, using a meta-heuristic algorithm optimization for maximizing the number of comb lines with flatness within 3 dB margin. We report flat-top comb bandwidths in range from 125 to 171 GHz, with up to 29 comb lines and free spectral range tunability from 5 to 12.5 GHz.

In this paper we theoretically investigate application of a bistable Fabry-Pérot semiconductor laser under optical-injection as all-optical activation unit for multilayer perceptron optical neural networks. The proposed device is programmed to provide reconfigurable sigmoid-like activation functions with adjustable thresholds and saturation points and benchmarked on machine learning image recognition problems. Due to the reconfigurability of the activation unit, the accuracy can be increased by up to 2% simply by adjusting the control parameter of the activation unit to suit the specific problem. For a simple two-layer perceptron neural network, we achieve inference accuracies of up to 95% and 85%, for the MNIST and Fashion-MNIST datasets, respectively.

The paper explores the performance enhancement of a photonic reservoir computer through the reconfigurability of all-optical nonlinear node based on Fabry-Pérot laser diode. By benchmarking various tasks with differing computational demands, we demonstrate how reconfigurability offers flexible computational capabilities, enabling the reservoir computer to adapt to specific tasks while striking an optimal balance between nonlinear transformation and linear memory. We achieve the state-of-the-art results using a basic delay-line reservoir computer concept with a small number of virtual nodes (∼30 nodes).

Background/Objectives: Inexperienced dentists and dental students are especially prone to misdiagnosis, and this represents a huge problem regarding antimicrobial stewardship. We aimed to develop a mobile app for rational antibiotic prescribing in dentistry based on local–systemic symptoms and patient factors, rather than solely on diagnosis, to tackle misdiagnosis. Methods: The study involved 64 participants, 50 of which were third-year dental students attending a pharmacology course focusing on antimicrobials, comprising lectures and practical sessions without (noAPP group, n = 22) or with (APP group n = 28) the assistance of a mobile application. The other 14 participants were practicing dentists who decided to register and use the application. All registered users of the application were asked to take a feedback survey, while learning outcomes were evaluated via a pharmacology quiz. Results: A decision tree was used for application development. In total, 76 impressions were collected on the application. The majority of the impressions were related to odontogenic–endodontic infections. Multiple linear regression analysis did not reveal differences in survey responses between practicing dentists and undergraduate students in the feedback survey responses. There was a significant difference in the mean pharmacology test scores between the noAPP and APP groups (5.50 ± 1.80 vs. 7.21 ± 1.03, p = 0.0001). Conclusions: The dentalantibiotic.com application was developed to support rational antibiotic prescribing, in view of tackling misdiagnosis, among inexperienced dentists, as well as to assist in undergraduates’ pharmacology learning, and the current study shows its large impact as an educational tool. The majority of participants considered it easy to use, efficient in facilitating the right antibiotic choice, and useful for everyday decision-making.

We experimentally validate the all-optical activation functions in Fabry-Perot lasers under optical injection for random and non-random inputs. Sigmoid-like activations for 100 ps pulses are reconfigured using injection parameters, consuming 1.4 pJ per nonlinear operation.

The article presents a short statistical overview of photonics research in the Western Balkans, with more spotlight aimed at the field of optical frequency combs with focus on the overview of the research conducted at the University of Belgrade-School of Electrical Engineering. We give a review of proposed schemes for single and dual optical frequency combs generation, based on optimized gain-switching and electro-optic modulator operation and expand our previous results with analytical approximations for comb lines intensities for the gain-switched laser.

In this Letter, we present a detailed theoretical and experimental investigation of optical bistability and tristability in dual injection-locked Fabry-Perot laser diodes. The proposed device can be reconfigured between the bistable and tristable regimes, simply by adjusting the power level of the injected control optical signal. The tristability presented in the experiment is achieved for relatively low optical input powers between 1.03 and 1.25 mW, with the output signal ratio of up to 7 dB between stable states. Such a device is a potential candidate for designing trits, a bit analogy in ternary computational logic.

The expressiveness and learning capabilities of neural networks depend on the activation functions nonlinearity. We present a concept of photonic device with nonlinear transfer function which can be flexibly reconfigured to optimize the performance of different machine learning tasks, such as classification or time-series prediction.

In this paper, a novel reservoir computing architecture based on the single nonlinear node with two delayed inputs is presented. Its performance is investigated on time-series forecasting tasks, for two different profiles of activation functions implemented in the optical domain. Simulation results show reduced NRMSE value to around 3% for Mackey-Glass time series prediction 11 steps in advance, when sine squared optical activation function based on Mach-Zehnder modulator is used.

In this paper we propose a small-signal model of a reflective semiconductor optical amplifier (RSOA) based fiber cavity laser (FCL), which can be used for determining the small-signal modulation response and the corresponding −3dB bandwidth. The model is encapsulated in a boundary value problem (BVP), which, through the boundary condition, accounts for the fiber cavity transfer function. Starting from the numerical solution of the BVP, we investigate the modulation response and the −3dB bandwidth of an RSOA-FCL. We show that, due to the round-trip group delay, the RSOA-FCL's modulation response exhibits oscillatory behavior. The bandwidth of an RSOA-FCL is analyzed with respect to the RSOA's bias current density and active region length, the distribution fiber length, and the spectral width of an optical band-pass filter. We find that, in case of significant fiber dispersion, the bandwidth of FCL is limited by the lower of the two bandwidths — RSOA's or fiber's, usually the latter one. In case of small or negligible fiber dispersion, the bandwidth of FCL can be improved by increasing the RSOA's bias current or fiber length and/or selecting the optimal RSOA active region length, and is limited by the RSOA's performance.

We present an all-optical perceptron that utilizes an injection-locked Fabry-Perot laser diode as an adaptive activation unit. The evaluation of the proposed neuron was performed theoretically on the problem of binary classification of pairs of hand-written digits. Results for the MNIST dataset reveal an average accuracy of over 97% for activation function profile tailored by adjusting the laser diode bias current.

We present an approach for the generation of an adaptive sigmoid-like and PReLU nonlinear activation function of an all-optical perceptron, exploiting the bistability of an injection-locked Fabry-Perot semiconductor laser. The profile of the activation function can be tailored by adjusting the injection-locked side-mode order, frequency detuning of the input optical signal, Henry factor, or bias current. The universal fitting function for both families of the activation functions is presented.

In this paper we present and theoretically investigate a simple scheme for dual optical frequency comb generation residing on a pulse modulation of a semiconductor laser. The modulation current is composed of two superimposed waveforms, generated by two independent step recovery diodes, driven by two sine generators. Our model of the modulated laser estimates dual-combs comprising 300 teeth pairs within 10 dB margin spanning up to 60 GHz with total optical power of 0.4 mW.

We demonstrate the use of meta-heuristics algorithms for flatness optimization of optical frequency combs (OFCs). Without any additional component for flatness compensation, the laser alone is explored when driven by optimized bias current and radio frequency (RF) driving signals composed by multiple harmonics. The bias current amplitude and RF harmonic amplitudes and relative phases are optimized using particle swarm optimization (PSO) and differential evolution (DE) algorithms. The numerical results lead to a 9 lines-GS-laser-based OFC spectrum with 2.9 dB flatness. An online experimental optimization using the DE algorithm results in a 7-line-GS-laser-based OFC with 2 dB flatness.

This manuscript tackles the issue of circuit parasitics of reflective semiconductor optical amplifiers (RSOAs), which has been often identified as the main culprit for their relatively poor modulation performance. Building upon the findings related to the beneficial effect of inductive peaking (IP) in the case of small-signals and following the principle of photonic-electronic co-design, in this paper we extend our study and reveal that the same effect can be beneficial for large-signal modulation as well. By using multi-Gaussian probability distribution for analyzing the histograms of the sampled signal at the RSOA's output, we explore the evolution of the back-to-back ${\bf Q}$-factor under various operating conditions and benchmark the large-signal modulation performance of the IP enhanced RSOA against the one with parasitics excluded. The study confirms that the IP effect indeed provides enhancement of the ${\bf Q}$-factor especially in the case of low input optical powers and low to moderate current densities of bits 0 and 1. We reveal that, provided that the modulation depth is kept fixed, ${\bf Q}$-factor can be maximized by finding the optimum current corresponding to bit 0. The major and overall achievement of the IP implementation is the boost of the ${\bf Q}$-factor through simple modulation current engineering, breaking the limit for post-processing free high quality transmission (${\bf Q}\ {\geq 7}$) for a broad range of bit-rates and seeding optical powers.

In this paper, we present and theoretically investigate a simple and power efficient scheme for dual optical frequency comb generation residing on a single directly modulated semiconductor laser driven by two superimposed current waveforms. Our detailed model estimates dual combs comprising teeth pairs within 20 dB margin spanning up to 116 GHz with power per teeth pair up to 8.8 µW. In addition, we report dual combs with 40 GHz span comprising ultraflat teeth pairs, with flatness of the order of 1 dB.

We present a small-signal model of the multiple quantum well (MQW) reflective semiconductor optical amplifier (RSOA) with the accompanying parasitic circuit, followed by a numerical simulation of its modulation response. We analyze the corresponding -3 dB bandwidth's dependence on the bias current, optical power of the input signal, and the RSOA's active region length. Material, electrical and optical parameters, and the overall design of MQW RSOA are derived bottom-up from the fundamental principles. It is shown that the modulation response, resulting from either intrinsic or parasitic-like model (including transport effects), usually provides high -3 dB bandwidth, which is clamped by the chip's parasitics, leading to a relatively poor external modulation bandwidth. This issue can be overcome by an advanced design of the RSOA structure, as the one proposed in this paper and with the optimized bonding, which may improve the external modulation bandwidth significantly through the inductive peaking effect (IPE). Moreover, IPE's efficiency increases if the intrinsic or parasitic-like modulation bandwidth is in the proximity of the parasitic circuit resonant frequency.

We propose and experimentally demonstrate a power efficient dual-stage optical frequency comb using laser gain switching followed by a dual-drive Mach-Zehnder modulator (DD-MZM). The laser is initially gain switched at ∼ 9.5 GHz and the resultant comb is then expanded using a dual-drive Mach-Zehnder modulator driven at ∼ 19 GHz with signal amplitudes below 1.5 V. The setup generates an optical frequency comb, with 12 lines within 3 dB flatness, in a power efficient manner. Theoretical analysis is presented and verified through simulation and experimental results.