Speaker: Ken Osato (Chiba University)

Talk title: The Differentiable Universe: From Summary Statistics to Field-Level Analysis

Abstract
Understanding the nature of dark matter and dark energy is one of the fundamental challenges in modern cosmology. Recent advances in astronomical observations have enabled us to map the distribution of galaxies and matter across vast regions of the Universe with unprecedented precision. These observations provide critical clues to understanding the composition and evolution of the Universe. Traditionally, cosmological analyses have relied on summary statistics, such as correlation functions and power spectra, to extract information from these enormous datasets. While these methods have been highly successful, they do not capture all the information contained in the observations, particularly in the complex structures formed by nonlinear gravitational evolution.
Recent developments in differentiable cosmological simulations offer a new approach to overcoming these limitations. These simulations allow us to calculate how changes in the underlying physical parameters affect the predicted distribution of matter in the Universe. This capability opens up new possibilities for analyzing observed cosmic structures more directly, rather than relying solely on summary statistics. In this talk, I will review the challenges of current cosmological analyses and introduce recent advances in differentiable simulations and their applications to field-level analysis. I will discuss how these new approaches could help us make better use of astronomical observations and ultimately deepen our understanding of dark matter, dark energy, and the evolution of the Universe.