Physics for Artificial Intelligence and Data

Artificial intelligence, cloud data, ... have in common a double challenge: to always push the limits of their performance further while limiting their energy impact. Fundamental concepts and cutting-edge technologies, particularly in quantum physics, semiconductors and photonics, are at the heart of this digital revolution and its challenges.

This course aims to study, design, and integrate elementary components and functions in complex systems: datacenters, neural networks... It is based on a comprehensive approach from materials to systems combining modeling, simulation, manufacturing and characterization.

Introduction to physical concepts and issues. Ferroelectric materials for non-volatile memories, application to computation in memory and to hardware neural networks. TP: deposition of ferroelectric oxide layers, characterization of ferroelectric properties. BE: behavioral modeling of ferroelectric capacitors and simulation of neuromorphic memory circuits Integrated Photonics circuits and architectures, integration with datacenters: TP: realization of waveguides by UV lithography, characterization of optical properties. BE: FDTD simulation of functions such as micro-ring resonators, analysis of optical interconnections in datacenters, reconfigurable networks and optical servers Restitution