BEGIN:VCALENDAR VERSION:2.0 PRODID:-//linuxsoftware.nz//NONSGML Joyous v1.4//EN BEGIN:VEVENT SUMMARY:Data-driven modeling of collisional kinetic plasma dynamics DTSTART:20250609T160000Z DTEND:20250609T180000Z DTSTAMP:20260629T175701Z UID:fdc55c54-64d0-4d09-a54c-eed83ceaf7e6 SEQUENCE:2 CREATED:20250606T092704Z DESCRIPTION:Computational plasma physics has seen significant advances in the ability to model nonlinear plasma dynamics from first-principles. Howe ver\, capturing the complex multi-scale dynamics of plasmas\, especially i n non-equilibrium and strongly coupled regimes\, remains challenging. This talk will highlight recent work that explores the potential of combining machine learning algorithms with traditional plasma kinetic simulation fra meworks to address these challenges.The first part of the talk will focus on integrating  Graph Neural Networks (GNNs) as surrogate models for kine tic plasma simulators.  We demonstrate that GNNs can accurately replicate the dynamics of a one-dimensional plasma for arbitrary degrees of collisi onality\, and generalize effectively to conditions well beyond their train ing data. We also show that GNNs can implicitly learn a computationally ef ficient collisional operator in 1D. In the second part\, we discuss the us e of self-consistent electromagnetic Particle-in-Cell (PIC) simulations th at resolve the inter particle fields that mediate collisional interactions to guide the development of data-driven\, collisional operators. These da ta-driven operators can shed light on the nature of collisions in strongly coupled regimes and\, in the future\, elucidate how collisional operators are modified in far-from-thermodynamic-equilibrium conditions. LAST-MODIFIED:20250606T092718Z LOCATION:Online URL:http://df.vps.tecnico.ulisboa.pt/pt/eventos/data-driven-modeling-of-co llisional-kinetic-plasma-dynamics/ X-ALT-DESC;FMTTYPE=text/html:

Computational plasm a physics has seen significant advances in the ability to model nonlinear plasma dynamics from first-principles. However\, capturing the complex mul ti-scale dynamics of plasmas\, especially in non-equilibrium and strongly coupled regimes\, remains challenging. This talk will highlight recent wor k that explores the potential of combining machine learning algorithms wit h traditional plasma kinetic simulation frameworks to address these challe nges.


The first part of the talk will focus on integrating  Graph Neural Networks (GNNs) as surrogate models for kinetic plasma simu lators.  We demonstrate that GNNs can accurately replicate the dynamics o f a one-dimensional plasma for arbitrary degrees of collisionality\, and g eneralize effectively to conditions well beyond their training data. We al so show that GNNs can implicitly learn a computationally efficient collisi onal operator in 1D.

In the second part\, we discuss the use of self-consistent electromagnetic Particle-in-Cell (PIC) simulations that re solve the inter particle fields that mediate collisional interactions to g uide the development of data-driven\, collisional operators. These data-dr iven operators can shed light on the nature of collisions in strongly coup led regimes and\, in the future\, elucidate how collisional operators are modified in far-from-thermodynamic-equilibrium conditions.

END:VEVENT END:VCALENDAR