Events Daily

Wednesday, May 1, 2024

HEP/Pheno Journal Club
Event Type: ArXiv Discussion
Time: 12:30 PM - 1:30 PM
Location: 726 Broadway, 940, CCPP Seminar

Cosmic Tensions and Early Dark Energy
Michael Toomey, MIT
Event Type: HEP Seminar
Time: 2:00 PM - 3:00 PM
Location: 726 Broadway, 940, CCPP Seminar
Abstract: Cosmological datasets consistently exhibit growing tension with the Λ Cold Dark Matter (ΛCDM) model of cosmology. The most striking discrepancy is between direct measurements of today's expansion rate (H0) using Cepheid-calibrated supernovae and the expansion rate inferred from ΛCDM fits to early-Universe data sets, namely the cosmic microwave background (CMB). This discrepancy now sits at a 5σ tension. More recently, measurements of galaxy clustering, as parameterized by the amplitude of dark matter density fluctuations (σ8), suggest further tension within the concordance model. One leading proposal to resolve the H0 tension is Early Dark Energy (EDE), which posits a slow-rolling scalar field active shortly before CMB formation. This field acts as a new, brief source of energy injection, modifying the size of the sound horizon—a critical standard ruler for constraining H0. In the first half of this talk, I will discuss the most recent constraints from combined analyses of CMB and large-scale structure datasets, which show that the original axion-like EDE model is strongly disfavored. I will also demonstrate that these constraints are not influenced by prior volume effects, which have been argued to artificially disfavor the model in Bayesian MCMC analyses of EDE. In the final part of the talk, I will present the first constraints on early dark energy using theory-informed priors in an analysis with CMB and large-scale structure datasets. Leveraging normalizing flows, a type of machine learning architecture, this represents the most stringent constraints on the model to date. Along the way, I will also comment on some general insights gained from the detailed analysis of the EDE scenario and how this should inspire model builders to construct future, well-motivated extensions of ΛCDM.

Link to the Event Video