News

Statistical physics provides novel strategies for low-power computers

  • Faculté des Sciences, des Technologies et de Médecine (FSTM)
    23 septembre 2021
  • Catégorie
    Recherche
  • Thème
    Physique & sciences des matériaux

Physicists from the University of Luxembourg in collaboration with the University of Louvain have reported a comprehensive theory on how to construct rigorous stochastic models of complex electronic circuits at the single electron level and in a thermodynamically consistent way. The scientists have published their findings in Physics Review X, the prestigious journal of the American Physical Society, committed to publishing key advances in fundamental research in all branches of pure, applied and interdisciplinary physics.

Understanding and describing thermal fluctuations in complex electronic circuits is a fundamental problem with very relevant practical implications. The main reason is that thermal fluctuations are one of the limiting factors preventing the design of faster computers with lower energy demands. Also, new computing schemes are being considered where thermal fluctuations are exploited as a resource to solve problems that require randomness, and this enables large energy savings. However, thermal fluctuations are usually described with ad-hoc treatments involving approximations that are difficult to control.

In order to provide this new theory, Dr. Nahuel Freitas from the Department of Physics and Materials Science (DPhyMS) and co-workers have leveraged recent developments in the field of stochastic thermodynamics, that provide new rigorous tools and concepts of universal validity to understand the role of thermal fluctuations in many different systems. When these developments are combined with modeling methods used in engineering, the resulting formalism allows to build stochastic models of a large family of electronic circuits with practical relevance. In particular, they have applied it to the kind of circuits used nowadays to design computers. “Information processing alone is responsible for a significant share of the global carbon emissions, and this share is only expected to increase. We are interested in the physics of computation at a fundamental level but we also want to contribute to the search for new energy-efficient computing schemes in realistic architectures.”, says Nahuel Freitas.

As a first non-trivial application, they proposed and studied a design of a stochastic neuron (a basic component for the physical implementation of artificial neural networks) that uses thermal fluctuations as a resource, and that can be built with the same technology that powers regular computers.

“Our work provides bridges between statistical physics, thermodynamics, and computer design” says Professor Massimiliano Esposito, head of Complex Systems and Statistical Mechanics group within the Physics and Materials Science Department. “Building on the latest fundamental findings on the nature of thermal noise, this paper opens the way to design novel computing paradigms exploiting noise rather than avoiding it.“ adds Prof. Esposito.

Article “Stochastic Thermodynamics of Nonlinear Electronic Circuits: A Realistic Framework for Computing Around “, Physical Review X, September 2021kT