AI & Physics
AI techniques have the potential to fundamentally transform traditional approaches to scientific research. In physics, these techniques are employed to tackle complex problems in a broad variety of fields, including condensed matter physics, materials science and quantum information technologies. Our research work on AI focuses on creating and tailoring new algorithms, models and methodologies to address the specific challenges of scientific research in physics. We are also committed to explore how physics insights can help to develop fundamentally new AI approaches and frameworks.
Selected publications
Below we showcase our recent scientific papers at the crossroads of Artificial Intelligence and Physics, aimed at advancing the understanding of the fundamentals and applications in a variety of different areas.
- "Generative adversarial networks for data-scarce radiative heat transfer applications," J.J. García-Esteban, J.C. Cuevas, J. Bravo-Abad. Machine Learning: Science and Technology 5, 015060 (2024).
- "Tackling multimodal device distributions in inverse photonic design using invertible neural networks," M. Frising, J. Bravo-Abad, F. Prins. Machine Learning: Science and Technology 4, 02LT02 (2023).
- "Deep learning enabled inverse design in nanophotonics," S. So, T. Badloe, J. Noh, J. Bravo-Abad, J. Rho. Nanophotonics 9, 1041-1057 (2020).
Training the next generation of professionals in the field
Over the years, we have had the privilege of advising a talented cohort of students across various academic levels. Below we showcase the innovative projects we have developed together, underscoring our commitment to nurturing the next generation of professionals in this dynamic field.
- Michel Frising, PhD Thesis title: "Machine learning for probabilistic inverse design in nanophotonics", 2018-2022 (La Caixa Fellowship, co-adviced with Prof. Ferry Prins, placement: postdoc at UAM).
- Miguel Dalmau, Master Thesis title: "Graph convolutional neural networks for the prediction of properties of small organic molecules", 2021-2022 (co-adviced with Prof. Eduardo Hernández).
- David Alonso, Master Thesis title: "A deep learning approach to light transmission through single subwavelength apertures", 2020-2021 (co-adviced with Prof. Ferry Prins).
- Juan José García, Master Thesis title: "Deep learning in radiative heat transfer", 2019-2020 (co-adviced with Prof. Juan Carlos Cuevas).
- Pablo Gallego, Undergraduate Thesis title: "Application of deep learning techniques to the simulation of light scattering by a metallic particle", 2021-2022 (co-adviced with Prof. Esteban Moreno).
- Roberto Rodiño, Undergraduate Thesis title: "Artificial intelligence for the simulation of nanophotonic structures", 2019-2020.
Short courses and master classes at UAM
We have created a short series of lectures that cover the fundamental concepts as well as the most relevant applications of AI to the natural sciences. They are aimed at students, researchers and professionals wanting to start applying AI to their particular fields of specialization but are new to the topic and wonder “where to begin?”
- AI for Scientists Bootcamp 2021 (June 2021).
- AI for Scientists Bootcamp (Oct.-Nov. 2020).
- Introduction to artificial intelligence for scientific problems (July 2020).
AI for Physics in conference sessions
- Artificial Intelligence for Condensed Matter Physics. Co-organized with Alexandre Dauphin at the XXXVIII Biennial of Physics of the Spanish Royal Physics Society, July 2022.
- Artificial Intelligence for Condensed Matter Physics. Co-organized with Eliska Greplova at the Biennial of the Condensed Matter Division of the Spanish Royal Physics Society and of the European Physical Society, August 2020.