I conducted this work as a part of my Masters project in 2024, presented it as a scheduled talk at the SPAICE Conference in 2024, a presented it as a poster at the 37th NeurIPS Machine Learning for the Physical Sciences Workshop in 2023. I have now submitted an updated version of this project to the Journal on Geophysical Research Machine Learning and Computation.
In this work, I train a surrogate neural likelihood model to accelerate Hamiltonian Monte Carlo sampling from the posterior over parameters describing the transport of galactic cosmic rays through our solar system.
Read the updated 2025 paper in review for publication in JGR Machine Learning and Computation
Read the 2024 paper presented at SPAICE
Read the 2023 NeurIPS paper on ML4PhysicalSciences workshop page