Statistically Exploring Cracks in the Lambda Cold Dark Matter Model
Diaz Rivero, Ana
MetadataShow full item record
CitationDiaz Rivero, Ana. 2020. Statistically Exploring Cracks in the Lambda Cold Dark Matter Model. Doctoral dissertation, Harvard University Graduate School of Arts and Sciences.
AbstractThe Lambda Cold Dark Matter (LCDM) model has long been lauded for its exquisite fit to large-scale cosmological observations. Recent years, however, have revealed cracks in this model and the theories that underpin it. Tensions have arisen between the values of some cosmological parameters inferred from high-redshift and low-redshift independent observables. Little progress has been made in confirming, or at least improving, our understanding about some of the most important theoretical foundations the model rests on, including inflation, dark energy, and dark matter.
This dissertation aims to shed light on some of the most important outstanding issues in the LCDM picture of the universe and develop statistical tools to (1) extract more and better information from cosmological and astrophysical observables, and to (2) improve the accuracy of statistical inference pipelines, questioning established modeling choices and assumptions that can give us unfair confidence in our model of the universe, skew our understanding of the universe, or both. In the first part of the dissertation we focus on probes of cold dark matter on the kiloparsec scale, a regime that remains untested and is very sensitive to dark matter microphysics. We present novel ways of mapping the distribution of dark matter on these scales that offer a variety of advantages compared to traditional methods: sensitivity to lower masses (smaller scales), much more model-independence, and a significant speed-up. In the second part of the dissertation we address the impact of canonical modeling choices in cosmological data analyses and how this relates to the important tensions we observe between high-precision datasets. We use and develop data-driven methods that can incorporate our ignorance into statistical inference pipelines, removing important assumptions that are usually baked into analyses.
We conclude by assessing the reach of our work and placing it in the context of the broader field, envisioning how it will shape research in the years to come.
Citable link to this pagehttps://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37369445
- FAS Theses and Dissertations