Reseach area C: Adaptive solid-state nanosystems
C02
C02 Opto-electronic neuromorphic architectures
Prof. Dr. Rudolf Bratschitsch - Physics Institute
Prof. Dr. Wolfram H. P. Pernice - Physics Institute
Prof. Dr. Wilfred G. van der Wiel - Physics Institute and MESA+ (University of Twente)Project description
We will develop adaptive nanoscale opto-electronic networks for machine learning in materio. Memory functionality is embedded via phase-change materials (PCMs). Learning capability is obtained by combining local field enhancement through plasmonic nanoparticles (NPs) with optical and electrical feedback. NP single-electron transistors will employ PCMs as tunnel barriers that can be programmed by ultra-short optical pulses combined with feedback from electrical high-frequency signals. We will study both regular and disordered NP networks created via bottom-up self-assembly and top-down nanofabrication. Our long-term goal is to realize matter-like processors that communicate with each other, and to analyse electrical sensory input, providing intelligent response for machine-learning tasks.
Ivonne Bente Opto-electronic neuromorphic architectures | Pernice
Lorenzo Cassola Silicon-Based Neuromorphic Computation in Materio | van der Wiel
Reinier Cool Incorporating Memory into Unconventional Computing Devices| van der Wiel
C03
C03 Self-assembly of hybrid nanostructures for neuromorphic electronics
Prof. Dr. Andreas Heuer - Institute of Physical Chemistry
Prof. Dr. Bart Jan Ravoo - Organic Chemistry Institute
Prof. Dr. Wilfred G. van der Wiel - Physics Institute and MESA+ (University of Twente)Project description
Disordered networks of functionalized metal nanoparticles can be configured into devices such as classifiers. Here we significantly enhance the functionality of these devices by introducing organic ligands that can be switched between different configurations and/or charge states. Additionally, we will also use magnetic nanoparticles. These new features will not only enhance the addressability, but will also introduce memory, allowing to tackle new (time-dependent) problems. By directly connecting chemical synthesis, physical experiment, and simulation of data-driven and physical models, the project will explore the development of intelligent matter based on Material Learning.
Jonas Mensing Kinetic Monte Carlo Model for Computing Functionalities in Nanoparticle Networks | Heuer
Lisa Schlichter Self-Assembly of Hybrid Nanostructures for Brain-Inspired Electronics | Ravoo
Marc Beuel Self-Assembly of Hybrid Nanostructures for Neuromorphic Electronics | van der Wiel
C04
C04 Adaptive magnonic networks for nanoscale reservoir computing
Prof. Dr. Rudolf Bratschitsch - Physics Institute
Prof. Dr. Sergej O. Demokritov - Institute of Applied Physics
Prof. Dr. Wolfram H. P. Pernice - Physics InstituteProject description
We plan to develop nanoscale reservoir computing devices based on adaptive magnonic networks with embedded memory functionality. Adaptive networks based on both lithographically patterned and self-assembled nanostructures will be realized, which will be controlled using ultrafast optical programming. In the mid-term, we will combine the developed controllable building blocks into ordered and disordered networks with multiple input and output terminals, allowing implementation of reservoir-computing devices at the nanoscale. In the long term, we strive to employ interconnected magnonic reservoirs for realizing adaptive surfaces responding to magnetic, electric, and optical stimuli.
Jannis Bensmann Ultrafast Dynamics in Magnetic Nanosystems | Bratschitsch
Kirill Nikolaev Dynamics of magnon Bose-Einstein condensate | Demokritov
Dmitrii Raskhodchikov Spin Wave Systems for Reservoir Computing | Pernice
C05
C05 Coherent nanophotonic neural networks with adaptive molecular systems
Jun.-Prof. Dr. Benjamin Risse - Faculty of Mathematics and Computer Science
Jun.-Prof. Dr. Carsten Schuck - Physics InstituteProject description
This project targets the implementation and performance evaluation of elementary linear and nonlinear building blocks for coherent optical artificial neural networks. The focus will lie on integrating a large variety of nonlinear photo-responsive chemical and molecular systems developed within this CRC with nanophotonic devices that allow for straightforward replication. We will assess the characteristics of the resultant building blocks for dedicated training, regularization and explainable AI strategies to derive tailored analysis and optimization algorithms. This interdisciplinary combination will yield nanophotonic neural network components and accompanying digital twins that pave the way for large-scale artificial intelligence.
Marlon Becker Explainable Deep Learning and Nanophotonic Neural Networks | Risse
C06
C06 Mixed-mode in-memory computing using adaptive phase-change materials
Prof. Dr. Wolfram H. P. Pernice - Physics Institute
Prof. Dr. Martin Salinga - Institute of Materials PhysicsProject description
We develop neuromorphic architectures that exploit phase-change materials to implement in-memory computing. Nanophotonic waveguides will allow for realizing high-bandwidth neuromorphic processors with both optical and electrical feedback. Nanoscale artificial synapses will be based on phase-change materials, which are electrically programmed into multiple memory states for weighted optical readout. Using material engineering and spatially resolved phase-state assignment, we will create interconnected logic arrays for mixed-mode information processing.
Zhongyu Tang Development of Lossless Photonic Memories with Novel Materials | Pernice
Akhil Varri Neuromorphic photonic computing | Pernice
Daniel Wendland Neuromorphic Photonic Computing | Pernice
Nishant Saxena Phase Change Materials for Programmable Electronic-Photonic Systems | Salinga
Niklas Vollmar Electro-Optical in-Memory Computing With Phase Change Materials | Salinga