Skyrmion Based Physical Reservoir Computing*

Physical reservoir computing (PRC) is one of the leading unconventional computing paradigms which is capable of performing recognition and classification tasks by exploiting nonlinear dynamics of reservoir blocks [1, 2]. We demonstrate the use of thin films hosting several skyrmions, particularly one, two, three, four and nine skyrmions, as a block for implementing reservoir computing. Our concept utilizes the breathing of skyrmions in response to an applied electric field, which are coupled through dipole interaction and spin waves. This nonlinear [3] and coupled magnetization dynamics is exploited to perform temporal pattern recognition. Two performance metrics, namely short-term memory and parity check capacity were studied to demonstrate the potential of high performance of skyrmion based PRC. Furthermore, our study demonstrates that nonlinear magnetization dynamics and interaction through spin waves and dipole interaction affect memory and parity check capacity and explain the role of physical interactions in this dynamical system in its ability to perform reservoir computing.

[1] C. Du et al., Nat Commun 8, 2204 (2017)

[2] D. Pinna et al., Phys. Rev. Applied, 14, 054020 (2020)

[3] M. Azam et al., J. Appl. Phy. 124, 152122 (2018)