Chiral Non-stationary Magnonic Auto-oscillator for neuromorphic computing

Description du projet

De nouveaux nano-oscillateurs pour mieux imiter les fonctions cérébrales

Inspirée par la manière dont le cerveau humain traite les informations, l’informatique neuromorphique est un domaine en plein essor qui vise à créer des systèmes plus rapides et plus efficaces sur le plan énergétique. Avec le soutien du programme Actions Marie Skłodowska-Curie, le projet CNMA fera un pas en avant en développant des nano-oscillateurs à couple de spin innovants appelés auto-oscillateurs magnoniques non-stationnaires chiraux. Ces dispositifs utiliseront des courants électriques modulés dans le temps et des signaux spéciaux pour imiter les capacités cérébrales, telles que la mémoire à court terme et les réponses non linéaires à des fréquences élevées de l’ordre du gigahertz. La recherche proposée se concentrera sur l’amélioration de l’efficacité et de la rapidité de ces dispositifs en vue de leur utilisation dans des réseaux inspirés du cerveau. En outre, le projet explorera la manière de connecter plusieurs appareils afin d’augmenter leur puissance de calcul.

Objectif

This project aims to advance the field of neuromorphic computing by exploiting novel non-stationary dynamics in spin torque nano-oscillator, called here chiral non-stationary magnonic auto-oscillators. The core innovation lies in the numerical and experimental demonstration of time-modulated electric current (TMEC) stimulation combined with secondary stimuli to induce robust nonlinear excitations capable of sustaining neuromorphic computations. Key features such as short-term memory, echo-state properties and strong non-linear responses are achieved in the GHz frequency range, specifically targeting skyrmion and vortex textures.

The project will focus on optimising CNMA for analogue neural networks by exploring time-dependent stimulation, improving energy efficiency and computational speed. A crucial part of the research will be a close collaboration with the renowned group at the University of Kaiserslautern, where I will receive specialised training in advanced experimental techniques for magnetoresistive and spectral analysis. This will be cross validated with micromagnetic simulations, combining the expertise of the fellow and the host group. The success of the project is also guaranteed by international cooperation with renowned research centres providing samples.

Towards the end of the project, we will numerically demonstrate the scalability of the CNMA system by coupling multiple CNMAs via spin waves in a ferromagnetic strip, mimicking additional neural connections. This approach promises to significantly increase the computational power of CNMA-based reservoir computing systems. Through this partnership, fellow aim to establish a strong international research network, which will open opportunities for long-term collaborations and enhance my position within the scientific community. This extensive cooperation will also play a crucial role in securing future employment within academic institutions or R&D sectors in industry.