A Nanotechnology-Ready Computing Scheme based on a Weakly Coupled Oscillator Network

Abstract

With conventional transistor technologies reaching their limits, alternative computing schemes based on novel technologies are currently gaining considerable interest. Notably, promising computing approaches have proposed to leverage the complex dynamics emerging in networks of coupled oscillators based on nanotechnologies. The physical implementation of such architectures remains a true challenge, however, as most proposed ideas are not robust to nanotechnology devices’ non-idealities. In this work, we propose and investigate the implementation of an oscillator-based architecture, which can be used to carry out pattern recognition tasks, and which is tailored to the specificities of nanotechnologies. This scheme relies on a weak coupling between oscillators, and does not require a fine tuning of the coupling values. After evaluating its reliability under the severe constraints associated to nanotechnologies, we explore the scalability of such an architecture, suggesting its potential to realize pattern recognition tasks using limited resources. We show that it is robust to issues like noise, variability and oscillator non-linearity. Defining network optimization design rules, we show that nano-oscillator networks could be used for efficient cognitive processing.

Two-dimensional mutually synchronized spin Hall nano-oscillator arrays for neuromorphic computing

“…Spin transfer torque nano-oscillators (STNOs) are one of the most promising candidates addressing these requirements (23,24). Free running STNOs can interact with other STNOs to synchronize via coupling mechanisms that can be electrical (25,26), exchange based (27), dipolar (28), or due to spinwave propagation (29)(30)(31)(32)(33)(34), and this way deliver higher power and more coherent microwave signals. A recent study has highlighted a comparable performance of one STNO (35) in vowel recognition rate compared to state-of-the-art CMOS.…”

Coherent Spin Pumping in a Strongly Coupled Magnon-Magnon Hybrid System

“…In spintronics, exchange-induced magnetic excitations, called spin waves, or magnons [3,4], are good candidates because information can be encoded by both the amplitude and the phase of spin waves. For example, the interference of coherent spin waves can be engineered for spin wave logic operations [5][6][7]; the coherent interaction of spin-torque oscillators leads to mutual synchronization [8][9][10][11][12][13], which can be applied in artificial neural networks [14,15]; and the coherent coupling between magnons and microwave cavities [16][17][18][19][20][21][22] opens up new opportunities for magnon-based quantum information science [23,24].Recently, strong coupling between two magnonic systems has been observed [25][26][27], which allows excitations of forbidden spin wave modes and high group velocity of propagating spin waves [28,29]. The coupling is dominated by the exchange interaction at the interface of the magnetic bilayers, providing a new pathway to coherently transfer magnon excitations between two magnetic systems possessing distinctive properties: from conductor to insulator, from uniform to nonuniform mode and from high-damping to low-damping systems.…”

Thermal coupling and effect of subharmonic synchronization in a system of two VO2 based oscillators

“…The effect of subharmonic synchronization studied in this paper might increase the number of states of coupled oscillators system. Usage of this effect for pattern recognition and classification problem (similar to [3]) will allow for increase of stored pattern number in ONN while working according to FSK-scheme without increasing the number of oscillators. The authors have proposed an effective algorithm to determine the value of synchronization efficiency of two oscillators and the value of subharmonic ratio at the synchronization frequency.…”

“…Therefore, the research dynamics of oscillatory neural networks (ONN) and their application to solve the problems of classification, clusterization and pattern recognition attracts a lot of attention. [3][4][5]. Hardware implementations of oscillatory neural networks are based both on current CMOS devices (e.g., phase-locked loop circuits [6] or Van der Pol oscillators [7]) and on emerging new devices, such as, spin-torque nano-oscillators [8], switches based on materials with metal-insulator [1,9] or charge density wave [10,11] transitions, and oxide RRAM [12,13].…”