Again! – but faster, better, and with more physics: ML-accelerated inference of galaxy properties in deep and wide surveys of the universe – Joel Leja (Penn State US)

Abstract: The inference of the physical properties of galaxies at cosmological distance requires modeling a wide range of physics, including e.g. stellar evolution and atmospheres; dust attenuation and re-emission; nebular physics;  AGN emission; and more. Bayesian inference is often used to map the inevitable degeneracies, and the large amount of physics and wide parameter space means these codes are typically not fast. Yet current and near-future surveys of the universe will yield spectra for millions of galaxies and imaging for billions. I will introduce new tactics employed to speed up this inference, ranging from neural net emulators of key physics (photoionization modeling; stellar spectra) to efficient gradient-enhanced GPU-accelerated high-dimensional sampling to rapid simulation-based inference. These yield speed-ups of somewhere between 100x and 1,000,000x, with different trade-offs in flexibility and accuracy. In addition to permitting sophisticated high-dimensional modeling on the industrial scales of modern surveys, I will discuss qualitatively new science directions enabled by these breakthroughs, such as modeling entire galaxy populations rather than one-at-a-time approaches and extremely high dimensional modeling of individual systems, e.g. spatially resolved modeling.

Link YouTube: https://youtu.be/2TPPMJ7qPME