On Wednesday, 21st of December at 2:30 am, there will be our next GoLP VIP Seminar by Prof. E. Paulo Alves (University of California, Los Angeles). The event will take place at Anfiteatro Abreu Faro – Complexo I at IST.

Title: Accelerating our understanding of complex nonlinear plasma dynamics using machine learning

Abstract: The increasing quantity and quality of plasma data being produced by laboratory experiments, spacecraft observations and high-fidelity numerical simulations is creating new opportunities for innovation in the way we do plasma physics. In particular, deep learning techniques are offering powerful new ways of building highly predictive data-driven models for various important applications (including disruption prediction of fusion plasmas). The inherent complexity of these data-driven models, however, limits their interpretability, challenging the development of a theoretical understanding of the plasma physics underlying the data. In this talk, I will focus on how sparse regression techniques can be used to infer interpretable plasma physics models (in the form of nonlinear partial differential equations) directly from the data of fully kinetic parKcle-in-cell (PIC) simulations. The potential of this approach will be illustrated through the recovery of the fundamental hierarchy of plasma physics models, from the Vlasov equation to magnetohydrodynamics, based solely on data of complex plasma dynamics captured by first-principles PIC simulations. I will discuss how this new data-driven methodology offers a promising new route to accelerate the development of new reduced models of complex nonlinear plasma phenomena and to design computationally efficient algorithms for mulK-scale plasma simulations.

Short bio of the speaker: Paulo Alves is an Assistant Professor in the UCLA Physics and Astronomy Department. He obtained his PhD in Plasma Physics at the Instituto Superior Tecnico in Lisbon, Portugal, in 2015. Following his PhD, he joined the SLAC National Accelerator Laboratory as a Research Associate and joined the UCLA faculty in 2020. His research aims to understand the plasma physics underpinning extreme astrophysical environments and high-energy-density laboratory experiments. His interests span a broad range of topics, from understanding how the most energetic particles in the Universe are accelerated, to harnessing plasmas to control and amplify ultra-intense lasers. His research combines analytic theory and state-of-the-art numerical simulations to understand the nonlinear and multi-scale dynamics of plasmas in these systems from first- principles. His research also explores how modern techniques from the fields of Artificial Intelligence and Machine Learning can be combined with traditional theoretical and computational techniques from Plasma Physics to develop advanced computational algorithms for multi-scale modeling of laboratory and astrophysical plasmas.