A new paradigm of reservoir computing exploiting hydrodynamics

Abstract

Nonlinear waves have played a significant role in the development of the science of complexity throughout history. More recently, they have also contributed to the emergence of a novel paradigm in reservoir computing called neuromorphic computing by waves. In this paradigm, information is transmitted through wave dynamics, which process the data to perform complex tasks with minimal energy consumption. Leveraging nonlinear hydrodynamic waves for computational purposes, we have created the Aqua-Photonic-Advantaged Computing Machine by Artificial Neural Networks (Aqua-PACMANNs). This system utilizes wave propagation in shallow water as the primary physical phenomenon, with minimal electronic components required, limited to a camera for detection purposes. As a proof of concept, we have successfully implemented an exclusive-not-or logic gate using the Aqua-PACMANN architecture. This accomplishment demonstrates the potential of fluid dynamic neuromorphic computing and opens up possibilities for a new class of computing systems.