| Abstract: |
| In ensembles of coupled oscillators, the synergy between topological
features and the underlying dynamics may lead to interesting
self-organized phenomena. In the first part of my talk, I will present a
system that is capable of exhibiting such complex dynamics: a SQUID
(superconducting quantum-interference device) metamaterial, i. e. an
artificially structured medium of periodically arranged, weakly coupled
SQUIDs, which shows extraordinary electromagnetic properties and
tunability. From a dynamical point of view, the single SQUID is a highly
nonlinear system exhibiting extreme multistablity and chaos. I will talk
about the emergent collective behavior in SQUID metamaterials, in
particular spatiotemporal pattern formation and chimera states.
The fundamental building block of superconducting devices is the
Josephson junction (JJ), which inherently exhibits neuromorphic
properties and can mimic basic neuron-like behavior.
Superconductor-based neuromorphic systems are particularly attractive
because they combine ultrafast operation, approaching terahertz
frequencies, with extremely low, sometimes negligible, power
dissipation. In the second part of my talk, I explore how Josephson
junctions can serve as physical substrates for machine learning,
focusing on reservoir computing applications. In particular, we
investigate time-multiplexed reservoir computing, which exploits the
strong nonlinearity and rich dynamical response of Josephson junctions
to perform information processing tasks efficiently. |
|