Project Manager

Philippe Talatchian, CEA Researcher at SPINTEC

The HOBBIT project aims to study the emergence of thermally activated collective magnetization, which could give rise to an unprecedented self-adaptive network. This network would replicate out-of-equilibrium critical dynamics observed in nature, such as in the brain or forest fires. The goal is to lay the foundation for a new type of computation, capable of operating with variable devices, noisy signals, and minimal energy consumption, similar to the human brain

Challenges

> Analyze the symmetry properties of various distortions and their implications on altermagnetic spin separation, which allows for the classification of materials based on symmetry consideration

> Contribute to spintronic neuromorphic computing: Advance research in the field of spintronic neuromorphic computing, thermally activated dynamics, and computational neuroscience, while exploring new pathways for brain-inspired computing with low energy consumption

> Build spintronic neural networks: Develop fully thermally activated spintronic neural networks, with stochastic spintronic neurons (SMTJs/STNOs) and non-volatile synapses, for cognitive tasks such as image and speech recognition

> Improve SMTJs and STNOs: Work on enhancing these devices to achieve fast (GHz to THz), reliable stochastic behavior and tunable functionalities at the nanoscale (probability and dwell time)..

Positioning the Project with the PEPR

The project will exploit unexplored spintronic stochastic properties and leverage resources from the SPINCOM and SPINCARAC projects (RF platform). It will establish connections between SPINTHEORY (out-of-equilibrium thermal dynamics) and TOAST (THz stochastic dynamics). HOBBIT is directly linked to AI and opens up possibilities for neuroscience and complex systems. It will support the algorithms from the EMERGENCES project (PEPR-IA) and benefit from the efforts of PEPR-Electronique, promoting the development of innovative spintronic technologies for the digital and green transitions.