Skyrmion based energy efficient straintronic physical reservoir computing

TL;DR

This paper demonstrates that patterned thin films hosting multiple skyrmions can implement energy-efficient physical reservoir computing by leveraging their nonlinear magnetization dynamics for temporal pattern recognition.

Contribution

It introduces a novel skyrmion-based reservoir computing model utilizing nonlinear breathing dynamics and physical interactions for recognition tasks.

Findings

Achieved 100% accuracy in classifying sine and square waves.

Showed that skyrmion interactions influence reservoir computing capacity.

Validated the energy efficiency of skyrmion-based reservoir computing.

Abstract

Physical Reservoir Computing (PRC) is an unconventional computing paradigm, which exploits nonlinear dynamics of reservoir blocks to perform recognition and classification tasks. Here we show with simulations that patterned thin films hosting several skyrmions, particularly one, two, four and nine skyrmions, can implement energy efficient reservoir computing. This reservoir computing (RC) block is based on nonlinear breathing dynamics of skyrmions, which are coupled to each other through dipole interaction and spin waves, in response to a voltage generated strain. This nonlinear and coupled magnetization dynamics is exploited to perform temporal pattern recognition. Two performance metrics, namely Short-Term Memory (STM) and Parity Check (PC) capacity are studied to demonstrate the potential of such skyrmion based PRC in addition to showing it can classify sine and square waves with 100% accuracy. Furthermore, our study demonstrates that nonlinear magnetization dynamics and interaction through spin waves and dipole coupling have a strong influence on STM and PC capacity, thus explaining the role of physical interactions in a dynamical system on its ability to perform Reservoir Computing (RC).