Uncovering vulnerable connections in the aging brain using reservoir computing. - In: Emerging Topics in Artificial Intelligence (ETAI) 2023, (2023), PC1265508
We used reservoir computing to explore the changes in the connectivity patterns of whole-brain anatomical networks derived by diffusion-weighted imaging, and their impact on cognition during aging. The networks showed optimal performance at small densities. This performance decreased with increasing density, with the rate of decrease being strongly associated with age and performance on behavioural tasks measuring cognitive function. This suggests that a network core of anatomical hubs is crucial for optimal functioning, while weaker connections are more susceptible to aging effects. This study highlights the potential utility of reservoir computing in understanding age-related changes in cognitive function.
Self-induced ultrafast electron-hole-plasma temperature oscillations in nanowire lasers. - In: Physical review applied, ISSN 2331-7019, Bd. 20 (2023), 3, S. 034045-1-034045-12
Nanowire lasers can be monolithically and site-selectively integrated onto silicon photonic circuits. To assess their full potential for ultrafast optoelectronic devices, a detailed understanding of their lasing dynamics is crucial. However, the roles played by their resonator geometry and the microscopic processes that mediate energy exchange between the photonic, electronic, and phononic subsystems are largely unexplored. Here, we study the dynamics of GaAs-AlGaAs core-shell nanowire lasers at cryogenic temperatures using a combined experimental and theoretical approach. Our results indicate that these NW lasers exhibit sustained intensity oscillations with frequencies ranging from 160GHz to 260GHz. As the underlying physical mechanism, we have identified self-induced electron-hole plasma temperature oscillations resulting from a dynamic competition between photoinduced carrier heating and cooling via phonon scattering. These dynamics are intimately linked to the strong interaction between the lasing mode and the gain material, which arises from the wavelength-scale dimensions of these lasers. We anticipate that our results could lead to optimised approaches for ultrafast intensity and phase modulation of chip-integrated semiconductor lasers at the nanoscale.
Data-informed reservoir computing for efficient time-series prediction. - In: Chaos, ISSN 1089-7682, Bd. 33 (2023), 7, 073109, S. 073109-1-073109-11
We propose a new approach to dynamical system forecasting called data-informed-reservoir computing (DI-RC) that, while solely being based on data, yields increased accuracy, reduced computational cost, and mitigates tedious hyper-parameter optimization of the reservoir computer (RC). Our DI-RC approach is based on the recently proposed hybrid setup where a knowledge-based model is combined with a machine learning prediction system, but it replaces the knowledge-based component by a data-driven model discovery technique. As a result, our approach can be chosen when a suitable knowledge-based model is not available. We demonstrate our approach using a delay-based RC as the machine learning component in conjunction with sparse identification of nonlinear dynamical systems for the data-driven model component. We test the performance on two example systems: the Lorenz system and the Kuramoto-Sivashinsky system. Our results indicate that our proposed technique can yield an improvement in the time-series forecasting capabilities compared with both approaches applied individually, while remaining computationally cheap. The benefit of our proposed approach, compared with pure RC, is most pronounced when the reservoir parameters are not optimized, thereby reducing the need for hyperparameter optimization.
Data-driven forecasting of nonequilibrium solid-state dynamics. - In: Physical review, ISSN 2469-9969, Bd. 107 (2023), 18, 184306, S. 184306-1-184306-18
We present a data-driven approach to efficiently approximate nonlinear transient dynamics in solid-state systems. Our proposed machine-learning model combines a dimensionality reduction stage with a nonlinear vector autoregression scheme. We report an outstanding time-series forecasting performance combined with an easy-to-deploy model and an inexpensive training routine. Our results are of great relevance as they have the potential to massively accelerate multiphysics simulation software and thereby guide the future development of solid-state-based technologies.
Spontaneous emission noise resilience of coupled nanolasers. - In: Frontiers in photonics, ISSN 2673-6853, Bd. 4 (2023), 1169988, S. 01-06
We investigate the spontaneous emission noise resilience of the phase-locked operation of two delay-coupled nanolasers. The system is modeled by semi-classical Maxwell-Bloch rate equations with stochastic Langevin-type noise sources. Our results reveal that a polarization dephasing time of two to three times the cavity photon lifetime maximizes the system’s ability to remain phase-locked in the presence of noise-induced perturbations. The Langevin noise term is caused by spontaneous emission processes which change both the intensity auto-correlation properties of the solitary lasers and the coupled system. In an experimental setup, these quantities are measurable and can be directly compared to our numerical data. The strong parameter dependence of the noise tolerance that we find may show possible routes for the design of robust on-chip integrated networks of nanolasers.
Multiplexed random-access optical memory in warm cesium vapor. - In: Optics express, ISSN 1094-4087, Bd. 31 (2023), 6, S. 10150-10158
The ability to store large amounts of photonic quantum states is regarded as substantial for future optical quantum computation and communication technologies. However, research for multiplexed quantum memories has been focused on systems that show good performance only after an elaborate preparation of the storage media. This makes it generally more difficult to apply outside a laboratory environment. In this work, we demonstrate a multiplexed random-access memory to store up to four optical pulses using electromagnetically induced transparency in warm cesium vapor. Using a Λ-System on the hyperfine transitions of the Cs D1 line, we achieve a mean internal storage efficiency of 36% and a 1/e lifetime of 3.2 µs. In combination with future improvements, this work facilitates the implementation of multiplexed memories in future quantum communication and computation infrastructures.
Deriving task specific performance from the information processing capacity of a reservoir computer. - In: Nanophotonics, ISSN 2192-8614, Bd. 12 (2023), 5, S. 937-947
In the reservoir computing literature, the information processing capacity is frequently used to characterize the computing capabilities of a reservoir. However, it remains unclear how the information processing capacity connects to the performance on specific tasks. We demonstrate on a set of standard benchmark tasks that the total information processing capacity correlates poorly with task specific performance. Further, we derive an expression for the normalized mean square error of a task as a weighted function of the individual information processing capacities. Mathematically, the derivation requires the task to have the same input distribution as used to calculate the information processing capacities. We test our method on a range of tasks that violate this requirement and find good qualitative agreement between the predicted and the actual errors as long as the task input sequences do not have long autocorrelation times. Our method offers deeper insight into the principles governing reservoir computing performance. It also increases the utility of the evaluation of information processing capacities, which are typically defined on i.i.d. input, even if specific tasks deliver inputs stemming from different distributions. Moreover, it offers the possibility of reducing the experimental cost of optimizing physical reservoirs, such as those implemented in photonic systems.
Photonic reservoir computing with non-linear memory cells: interplay between topology, delay and delayed input. - In: Emerging Topics in Artificial Intelligence (ETAI) 2022, (2022), 1220408, S. 1220408-1-1220408-7
Photonic reservoir computing is an emerging topic due to the possibility to realize very fast devices with minimal training effort. We will discuss the reservoir computing performance of memory cells with a focus on the impact of delay lines and the interplay between coupling topology and performance for various benchmark tasks. We will further show that additional delayed input can be beneficial for reservoir computing setups in general, as it provides an easy tuning parameter, which can improve the performance of a reservoir on a range of tasks.
Optimizing the cavity-arm ratio of V-shaped semiconductor disk lasers. - In: Physical review applied, ISSN 2331-7019, Bd. 18 (2022), 6, S. 064070
Passively mode-locked semiconductor disk lasers have received tremendous attention from both science and industry. Their relatively inexpensive production combined with excellent pulse performance and great emission-wavelength flexibility make them suitable laser candidates for applications ranging from frequency-comb tomography to spectroscopy. However, due to the interaction of the active medium dynamics and the device geometry, emission instabilities occur at high pump powers and thereby limit their performance potential. Hence, understanding those instabilities becomes critical for an optimal laser design. Using a delay-differential equation model, we are able to detect, understand, and classify three distinct instabilities that limit the maximum achievable pump power for the fundamental mode-locking state and link them to characteristic positive-net-gain windows. We furthermore derive a simple analytic approximation in order to quantitatively describe the stability boundary. Our results enable us to predict the optimal laser-cavity configuration with respect to positive-net-gain instabilities and therefore may be of great relevance for the future development of passively mode-locking semiconductor disk lasers.
Master memory function for delay-based reservoir computers with single-variable dynamics. - In: IEEE transactions on neural networks and learning systems, ISSN 2162-237X, Bd. 0 (2022), 0, S. 1-14
We show that many delay-based reservoir computers considered in the literature can be characterized by a universal master memory function (MMF). Once computed for two independent parameters, this function provides linear memory capacity for any delay-based single-variable reservoir with small inputs. Moreover, we propose an analytical description of the MMF that enables its efficient and fast computation. Our approach can be applied not only to single-variable delay-based reservoirs governed by known dynamical rules, such as the Mackey-Glass or Stuart-Landau-like systems, but also to reservoirs whose dynamical model is not available.