Exploring the Nature of Dark Matter Through Inflation, Primordial Black Holes, and Large Scale Structure

In this thesis I explore dark matter via three avenues: through cosmic inflation, the evolution of primordial black holes, and large scale structure. Each project studies a theoretical framework and works towards observational consequences of dark matter. This is important as a means to distinguish between the various proposed candidates of dark matter. First, I explore the paradigm of warm natural inflation (WNI), an inflationary paradigm in which the primordial field generates a thermal bath. Through this theoretically well motivated model, I am able to show that not only does it satisfy constraints from the cosmic microwave background, it is able to generate enough primordial black holes to explain dark matter. Additionally, I show that the generation of scalar induced gravitational waves (SIGWs) through this process will be detectable by future observations. Second, I explore the hierarchical merging or coagulation of primordial black holes through multiple generations. Specifically, I study the conditions necessary to trigger runaway merging. This is done via a GPU accelerated coagulation code. I find that for asteroid mass black holes, no significant merging occurs and it behaves similar to cold dark matter. Finally, I explore large scale structure finding methods via machine learning (ML) tools, with a focus on identifying cosmic voids. The identification of voids is important as a possible means to constrain the time dependence of dark energy. I use a Kmeans clustering algorithm to segment cosmic structures and find it segments voids similar to other structure finders. I then train a UNET neural network to predict void finding at higher redshifts and achieve moderate accuracy, with accuracy generally decaying at higher redshifts. I then finally build a pipeline to extract the surfaces of the void-like regions using Connected Component Labeling (CCL) and the Marching Cubes algorithm.