2021年第68回応用物理学会春季学術講演会

講演情報

一般セッション(口頭講演)

10 スピントロニクス・マグネティクス » 10.3 スピンデバイス・磁気メモリ・ストレージ技術

[17p-Z19-8~24] 10.3 スピンデバイス・磁気メモリ・ストレージ技術

2021年3月17日(水) 15:15 〜 19:45 Z19 (Z19)

金井 駿(東北大)、日比野 有岐(産総研)

19:30 〜 19:45

[17p-Z19-24] Simulation of reservoir computing utilizing artificial spin ice

〇(DC)Kwan Hon1、Yuki Kuwabiraki2、Minori Goto1,3、Ryoichi Nakatani2、Yoshishige Suzuki1,3、Hikaru Nomura1,2,3 (1.Grad. Sch. Eng. Sci., Osaka Uni.、2.Dept. Mater. Sci., Grad. Sch. Eng., Osaka Uni.、3.CSRN-Osaka)

キーワード:artificial spin ice, reservoir computing, nanomagnet logic

We propose the use of artificial spin ice for reservoir computing and evaluate its short-term memory and nonlinear computational abilities. We perform numerical simulation based on the macrospin model to investigate the beahviour of artifical spin ice at temperature of 0 K. The artifical spin ice is formed by 72 nanomagnets arranged in honeycomb lattice.The central nanomagnets of the aritifcal spin ice acts as the input layer in reservior computing framework. After an input is sent to the central nanomagnets, external magnetic fields are successively applied to update the magnetization (reservior) state. The optimzied short-term memory capacity of 3.5 and nonlinear compuation capcity of 2.9 are found when the external magnetic field is near the switching field of the nanomagnets. In addition, the capcacites can be adjusted by engineering the shape anistropy of the nanomagnets.