Pedro Francisco Gil — Stellarator Optimization, University of Washington

Home

I am a Postdoctoral Researcher at the University of Washington, working with Prof. Dr. Alan A. Kaptanoglu on stellarator optimization. Our work focuses on single-stage methods that optimize the plasma and coils together, GPU-accelerated optimization, and machine learning for stellarator design.

I completed my PhD at Ludwig Maximilian University Munich and Max-Planck-Institute for Plasma Physics under the supervision of Prof. Dr. Hartmut Zohm, where I studied how to design a stellarator to confine electrons and positrons and advance fusion reactors. My research interests include nonlinear optimization, and the application of artificial intelligence and machine learning to complex optimization problems.

Education

PhD in Stellarator Optimization, Ludwig Maximilian University of Munich and Max-Planck-Institute for Plasma Physics, Munich, Germany, 2026
MSc in Physics and Nanotechnology, DTU, Danmarks Tekniske Universitet, Lyngby, Denmark, 2022
MSc, Grande Ecole Program, ENSTA Paris, Institut Polytechnique de Paris, Paris, France, 2021
Preparatory Classes, Lycée du Parc, Lyon, France, 2018

Research

My work translates physical problems into tractable optimization frameworks. I focus on reformulating ill-posed problems and navigating highly non-convex optimization landscapes.

Current main topic

Coil optimization for stellarators is non-trivial as there can be an infinite amount of coil shapes that reproduce the target magnetic field. Recent work where we applied an augmented Lagrangian formulation to the electromagnetic coil problem resulted in Pareto-optimized fusion configurations.

Interests

Nonlinear and constrained optimization
Convex optimization and duality
Numerical linear algebra and large-scale solvers
Stochastic and robust optimization under uncertainty
Machine learning and AI for scientific design problems
Stellarator and fusion magnet design

Publications

2026 Stellarator coils for future fusion reactors via an augmented Lagrangian approach P. F. Gil, W. Li, J. Stratton, A. A. Kaptanoglu, E. V. Stenson — Physical Review Letters 137, 065101
2026 Efficient computation of stellarator coils with an augmented Lagrangian optimization method P. F. Gil, W. Li, J. Stratton, A. A. Kaptanoglu, E. V. Stenson — Physical Review E
2026 Designing a buildable optimized stellarator to confine electron-positron plasmas P. F. Gil, J. Smoniewski, P. Huslage, R. Jorge, T. Thun, E. Buglione-Ceresa, T. Schuler, S. Fingl, G.-H. Ducas, E. V. Stenson — Journal of Plasma Physics
2026 A proof-of-concept for automated AI-driven stellarator coil optimization with in-the-loop finite-element calculations A. A. Kaptanoglu, P. F. Gil — arXiv:2603.15240
2026 Stochastic single-stage stellarator optimization using fixed-boundary equilibria P. F. Gil, J. Smoniewski, R. Jorge, P. Huslage, E. V. Stenson — Plasma Phys. Control. Fusion
2026 A non-planar ReBCO test coil with 3D-printed aluminum support structure for the EPOS stellarator P. Huslage, T. Schuler, P. F. Gil, V. Brack, D. Schmeling, D. A. R. Orona, E. von Schoenberg, T. Thun, J. Smoniewski, E. V. Stenson — IEEE Transactions on Applied Superconductivity 36(3), 1–5
2026 Manufacturing tolerances of non-planar coils for an optimized tabletop stellarator P. F. Gil, V. Brack, T. Schuler, P. Huslage, J. Smoniewski, E. V. Stenson — IEEE Transactions on Applied Superconductivity 36(3), 1–5
2025 Strain optimisation for ReBCO high-temperature superconducting stellarator coils in SIMSOPT P. Huslage, E. J. Paul, M. Haque, P. F. Gil, N. Foppiani, J. Smoniewski, E. V. Stenson — Journal of Plasma Physics 91(2), E71