SFA II: Neutron star mergers

Neutron stars pack more mass than the Sun into a sphere only about 25 kilometers across. When two of them collide, they shake the fabric of spacetime – producing gravitational waves – and light up across the entire electromagnetic spectrum, including powerful gamma-ray bursts. Simulating such mergers requires solving Einstein's equations on supercomputers, yet even the largest simulations cannot resolve the tiny turbulent eddies that amplify magnetic fields to extreme strengths – the very fields believed to power the gamma-ray jets. This project trains deep-learning models, including physics-informed neural networks, on zoomed-in high-resolution simulations to represent this unresolvable small-scale physics in full merger simulations, and tests the predictions against real observations. It is multi-scale, reaching from tiny turbulent eddies to jets of astronomical size, and interdisciplinary across general relativity, plasma physics, machine-learning theory, and observational multimessenger astronomy.