• Working as a Postdoc in the field of theoretical plasma astrophysics
  
• Currently working on machine learnig applications to nonlinear dynamical systems in the context of turbulence
• Looking for collaborations involving data analysis, machine learning and statistical methods
  
• You can find my academic CV on my personal website

My name is Felipe Nathan de Oliveira Lopes, and I am currently a postdoctoral researcher at the Centre for Mathematical Plasma Astrophysics at the University of Leuven (KU Leuven) in Belgium. I hold a M.Sc. in Theoretical Physics from the University of Brasília in Brazil, and an M.Eng. in Nuclear Sciences from the Polytechnic University of Catalonia in Spain, while an intern at the Barcelona Supercomputing Center, where I worked with turbulence stabilisation via fast ions in controlled thermonuclear fusion. I also have a Ph.D. in Physics from the Ruhr-Universität Bochum in Germany, and at the same time I worked at the Max Planck Institute for Plasma Physics in Munich, where I developed a geometrical formulation of hamiltonian field theory applied to modelling reduced (gyro)kinetic turbulence in astrophysical and space plasmas.

My primary research interests lie at the intersection of theoretical plasma astrophysics, applied differential geometry, and data analysis / machine learning. My experience at the Barcelona Supercomputing Center also expanded my expertise in numerical simulations and high-performance computing.

At KU Leuven, I am involved in an FWO-funded project focused on using physically informed machine learning techniques to better understand energy dissipation in collisionless astrophysical systems—one of the enduring open questions in space and astrophysical plasmas. More details about the project can be found on my personal webpage.

In my spare time, I like to experiment with mathematical approaches for extracting structure and insights from high-dimensional data. In particular, I have a strong interest in topological data analysis. Currently, I am exploring how persistent homology can be applied to time series analysis with the goal of developing topologically informed hidden Markov models for robust motif detection in time series and complex dynamical systems.