Self-gravitating scalar field configurations, ultra light dark matter and galactic scale observations

Abstract

We live in an era of data-driven cosmology. Increasingly accurate and precise data from various sources has enabled us to establish the concordance model of cosmology knownasthe!CDMmodel. However,manyfundamentalquestionsaboutourUniverse remain unanswered, such as the mystery of the physical nature of dark matter. In this thesis, we will examine the exciting possibility that dark matter consists of spin-zero particles with the smallest possible mass consistent with the existence of dwarf galaxies in the Universe, i.e. m → 10→22 eV. There exist many experimental constraints on the couplings of such a new scalar with Standard Model particles. We however, focus our attention on its coupling to itself, i.e. its self-interactions. In other words, we do not make any assumptions about any non-gravitational interactions of dark matter with known elementary particles, except that they are small enough to be consistent with existing upper limits. We adopt a signature-driven approach and ask, what can galactic scale observations say about this class of dark matter scenarios. This kind of Ultra Light Dark Matter (ULDM) is expected to form stable self gravitating configurations, whose macroscopic properties depend on the mass m and self-coupling ω. We explore the observational consequences of describing the inner re gions of galactic halos using such self-gravitating scalar field configurations (also called solitons) and ask what values of ω can be probed if m → 10→22 eV. First, we demonstrate a method to obtain constraints in the ω ↑ m plane, using observational upper limits on the amount of mass contained within the central region of a galactic halo containing a supermassive black hole. By obtaining solutions of the Gross-Pitaevskii-Poisson equations, that characterise stationary configurations of the scalar field, we showed that both attractive and repulsive self-interactions with strength ω →±10→96 ↑10→95 can be probed by such upper limits. Next, we tackled recent constraints on the ULDM scenario with ω =0,which stated that ULDM with m ↓ [10→24 eV,10→20 eV] cannot simultaneously satisfy observed rotation curves of galaxies as well as an empirically expected power law between the mass of the soliton and mass of the halo. We found that, for dark matter dominated low surface brightness galaxies from the SPARC catalogue, ULDM with m =10→22 eV and ω ↭ 10→90 can satisfy both observed rotation curves as well as a modified power law relation that includes the effects of self-interactions. Another interesting consequence of ULDM with m → 10→22 eV is the tunnelling of dark matter from a satellite dwarf galaxy orbiting around the centre of a large halo over cosmological timescales, due to the tidal effects of the host halo. Using the for malism of quasi-stationary solutions, we studied the effects of self-interactions on such a system and found that attractive self-interactions aid the self-gravity of the satellite galaxy against the tidal effects of the halo, extending the lifetime of the satellite. On the other hand, repulsive self-interactions do the opposite and shorten the lifetime of satel lite galaxies. We applied this to the Fornax dwarf spheroidal with a known core mass and orbital period, and found that ULDM with m =10→22 eV and ω ↫ ↑2.12 ↔ 10→91 will enable the dwarf galaxy to survive on cosmological timescales, evading a recent constraint for the same mass but with ω =0, thereby remaining consistent with the ULDMparadigm. Additionally, in the context of ULDM, we also explore machine learning models like Artificial Neural Networks (ANNs) to learn from observed data. Using simulated rotation curves of several dwarf galaxies, we train neural networks to learn the relation ship between the rotation curve and parameters of the dark matter density profile (such as ULDM particle mass m, scaling parameter s, core-to-envelope transition radius rt and NFWscale radius rs), along with Baryonic parameters (such as the stellar mass-to light ratio ”↑). We then feed the observed rotation curves of the galaxies and attempt to infer appropriate parameter values along with their uncertainties. Here, we explore the importance of noise in the training data, and compare the inferred parameter values and uncertainties obtained using two different methods to those obtained using standard Bayesian methods. We find that the trained neural networks can extract parameters that describe observations well for the galaxies we studied. This thesis is an effort to glean some insight into the nature of dark matter using galactic scale observations. Our work highlights the importance of taking the effects of self-interactions into consideration, as future experiments attempt to detect ULDM. Our work also highlights the importance of developing and embracing new techniques to learn from data.