Neuromorphic computation with a single magnetic domain wall

Abstract Machine learning techniques are commonly used to model complex relationships but implementations on digital hardware relatively inefficient due poor matching between conventional computer architectures and the structures of algorithms they required simulate. Neuromorphic devices, in particular reservoir computing architectures, utilize inherent properties physical systems implement machine so have potential be much more efficient. In this work, we demonstrate that dynamics individual domain walls magnetic nanowires suitable for implementing paradigm hardware. We modelled a wall placed two anti-notches nickel nanowire using both 1D collective coordinates micromagnetic simulations. When driven by an oscillating field, exhibits non-linear within well created analogous those Duffing oscillator. exploit modulating amplitude applied field inject time-multiplexed input signals into reservoir, show how allows us perform tasks including: classification (1) sine square waves; (2) spoken digits; (3) non-temporal 2D toy data hand written digits. Our work lays foundation creation nanoscale neuromorphic devices which analysis tasks.