Litcius/Paper detail

Easy-plane spin Hall nano-oscillators as spiking neurons for neuromorphic computing

Danijela Marković, Matthew W. Daniels, Pankaj Sethi, Andrew D. Kent, M. D. Stiles, Julie Grollier

2022Physical review. B./Physical review. B27 citationsDOIOpen Access PDF

Abstract

Ferromagnetic nanoconstrictions can serve as integrate-and-fire neurons, communicating via spin-wave pulses analogous to neural spikes. The nanoconstriction geometry leads to easy-plane oscillators which manifest phase dynamics analogous to damped driven oscillators. The authors propose a sample geometry where nanoconstrictions undergo phase slips and produce pulses of spin-wave signals when triggered by a small spin torque. They show with the help of macrospin modeling and micromagnetic simulations that these spin-wave pulses can excite other nanoconstriction neurons and that they possess the basic fan-in and fan-out capability required for neuromorphic systems.

Topics & Concepts

Neuromorphic engineeringSpiking neural networkPhysicsAnisotropyJosephson effectArtificial neural networkSpin (aerodynamics)Plane (geometry)Condensed matter physicsTopology (electrical circuits)Computer scienceMathematicsQuantum mechanicsElectrical engineeringArtificial intelligenceEngineeringGeometrySuperconductivityThermodynamicsAdvanced Memory and Neural ComputingNeural Networks and Reservoir ComputingMagnetic properties of thin films
Easy-plane spin Hall nano-oscillators as spiking neurons for neuromorphic computing | Litcius