Organized by: Prof. Lu Lu

Events on Wednesday, February 26th, 2025

Beyond the Observable: A Machine Learning perspective on modern Cosmology

Time: 4:00 pm - 5:00 pm

Place: 5280 CH &

Speaker: Dr. Carolina Cuesta Lazaro, Institute for Artificial Intelligence and Fundamental Interactions, MIT

Abstract: Our observations have painted a simple portrait of cosmic evolution, yet fundamental questions remain unanswered: What is the nature of dark matter, the invisible substance that makes up most of the matter in the Universe? What is driving the accelerated expansion of the cosmos? How did it begin? In this talk, I will present machine learning frameworks that bridge the gap between numerical simulations and increasingly precise astronomical observations to decode these invisible components. I will demonstrate how generative models can serve as both efficient emulators and tools for parameter estimation, enabling direct inference from high-dimensional astronomical data without relying on simplified summary statistics. I will then show how we can probabilistically reconstruct the dark matter distribution from observed galaxy clustering, while being robust to uncertainties in galaxy formation physics. Finally, I will discuss how machine learning enables us to reconstruct the primordial Universe by inferring the initial conditions that evolve into our observed cosmos. Through these methods, we can also identify potential anomalies in the cosmic web that might signal departures from our standard cosmological model. These advances come at a pivotal moment, as we enter an era of extremely precise cosmological surveys that may transform current statistical tensions into discoveries of new fundamental physics.

Host: Sridhara Dasu

Add this event to your calendar