Description du projet
Une approche photonique pour la résolution de problèmes mathématiques complexes
Les verres de spin sont des matériaux magnétiques dont l’orientation magnétique semble totalement aléatoire, comme s’ils étaient gelés. Ces matériaux servent de modèles pour des problèmes mathématiques de grande envergure, par exemple en intelligence artificielle, en logistique et en séquençage de l’ADN, qui sont difficiles à résoudre avec la technologie informatique actuelle. Le projet BEC-NETWORKS, financé par l’UE, envisage de créer des condensats de Bose-Einstein à partir de photons en envoyant plusieurs faisceaux laser dans des microcavités optiques. Cette plateforme expérimentale permettra aux chercheurs de réaliser des simulations ultrarapides de systèmes de spin classiques et de résoudre le problème de l’état fondamental dans les verres de spin.
Objectif
Despite large advances in both algorithms and computer technology, even typical instances of computationally hard problems are too difficult to be solved on today’s computers. Unconventional computational devices that break with the usual paradigms of digital electronic computers can help to overcome these limitations. In this project, a network of coupled photon Bose-Einstein condensates will be developed and used as experimental platform to perform ultrafast simulations of classical spin systems. Specifically, the network will be capable of solving the ground-state problem in spin glasses (disordered magnets). The latter constitutes a well-known combinatorial problem that can be mapped mathematically to many other computationally hard problems in machine learning, logistics, computer chip design and DNA sequencing. In a proof-of-principle experiment, I aim to demonstrate that the proposed spin glass simulator performs this computationally hard optimisation problem significantly faster than any other computer today. I have pioneered the Bose-Einstein condensation of photons in optical microcavities, which has enabled us to investigate this genuine quantum-mechanical effect with all-optical methods. In a recent work of my group, we experimentally demonstrate controllable phase relations between photon Bose-Einstein condensates in an optical microcavity. The investigated device realises an optical analogue of a Josephson junction. Similar to a transistor for electronics, a controllable photonic Josephson junction represents the key component for ultrafast optical spin glass simulation and, thus, is the crucial basis for the proposed project. The BEC-NETWORKS project will be the main research project of my research group at the University of Twente.
Champ scientifique
- natural sciencesbiological sciencesgeneticsDNA
- engineering and technologymaterials engineering
- natural sciencescomputer and information sciencesartificial intelligencemachine learning
- natural sciencesphysical sciencescondensed matter physicsbose-einstein condensates
- natural sciencesphysical sciencestheoretical physicsparticle physicsphotons
Programme(s)
Régime de financement
ERC-COG - Consolidator GrantInstitution d’accueil
7522 NB Enschede
Pays-Bas