Matteo Biagetti - Persistent Homology of Cosmic Structures

Persistent Homology of Cosmic Structures

by Matteo Biagetti (IFPU)

Abstract: Cosmologists search for answers about essential properties of the Universe on large scales, to understand its origin, evolution, and fate. In the current cosmological paradigm, the Λ Cold Dark Matter model, these properties are described by a small set of numbers determining the initial conditions for the formation of structures, the expansion rate, and the amount of matter, and radiation in the Universe. Galaxy surveys are a unique probe of the ΛCDM model as they observe angular positions and redshifts across large cosmological volumes, allowing us to trace the evolution of structures as a function of time.  Standard methods exploit the measurement of the correlation of galaxy pairs as a function of distance, as theoretical and numerical predictions for this observable exist and are reliable on large scales. As experiments will observe galaxy maps over wider and wider ranges of scales and redshifts and with better and better signal-to-noise ratios, these predictions will not improve as fast. I propose to explore novel summary statistics that are better suited for the datasets we have. Persistent homology is a method for computing topological features of a map at different spatial resolutions. It is built on a mathematically well-defined framework that is optimal for extracting information from high-dimensional, incomplete, and noisy datasets, such as galaxy maps. The program will establish an analysis pipeline for current and future surveys and a publicly accessible framework for likelihood analysis and the production of fast simulations for results validation. The final goal is to address fundamental questions in cosmology, such as the origin of structures, the amplitude of matter fluctuations, the expansion history of the universe, and the detection of the sum of neutrino masses.

The seminar will be held in hybrid form in the Library seminar room (Castello Basevi) and on

Caratteristiche dell'evento

Inizio evento 08-02-2023 11:30
Iscritti 0
Posti disponibili 35