The fundamental nature of dark matter (DM) so far eludes direct detection experiments, but it has left its imprint in the large-scale structure (LSS) of the Universe. I will present a search using cosmic microwave background (CMB) and galaxy surveys for ultra-light DM particle candidates called axions that are well motivated from high energy theory. In combining these datasets, I will discuss how the presence of axions can improve consistency between these probes and, in particular, help alleviate the S_8 cosmological parameter tension (the discrepancy in the amplitude of density fluctuations as inferred from CMB and galaxy data). I will then present complementary searches for ultra-light and light (sub-GeV) DM using a LSS probe called the Lyman-alpha forest. By combining complementary large- and small-scale structure probes, I will demonstrate how current and forthcoming cosmological data will systematically test the nature of DM. In order to model novel DM physics accurately and efficiently in CMB and LSS probes, I will present new machine learning approaches using neural network "emulators" to accelerate DM parameter inference from days to seconds and active learning to reduce massively the computational expense.

Large self interactions between dark matter particles alter the predicted properties of dark matter halos and may help address small-scale structure issues, while maintaining the successes of standard cold dark matter at large scales. Self interactions allow for efficient heat transfer within a halo, which can be modeled using a gravothermal fluid approximation. In this talk, I will discuss how the choice of the particle physics model for self interactions, as well as the environment of the halo, impacts the phases of halo evolution.

Upcoming surveys of cosmic structures will probe scales ranging from the nonlinear regime to scales close to the cosmological horizon. This opens up the possibility of probing the LCDM model, as well as early universe scenarios with non-Gaussianity. Modeling the galaxy bispectrum is challenging, as it involves general relativity, radiation, and large nonlinearities. In this talk, I will present a numerical modeling of the matter bispectrum on the light cone including relativistic and radiation effects. This is a crucial step towards modeling the observable bispectra, i.e. the weak lensing bispectrum and the galaxy bispectrum.

While the standard $\Lambda$CDM model provides an astonishing fit to cosmological data, there is a 4-6 $\sigma$ tension in the Hubble constant, $H_0$, between the value inferred from early-Universe observables, Planck CMB anisotropy spectra, and the value determined by local measurements from SH0ES collaboration using Type 1a SN calibrated with Cepheid variables. As an effort to solve this tension, we develop a method of searching for data-driven solutions to the Hubble tension given cosmological dataset, based on the standard Fisher bias formalism. As an example, we apply our method to time-varying electron mass modifying cosmological recombination history. With Planck CMB data, we find a solution entirely resolving the Hubble tension. However, once BOSS BAO and PantheonPlus uncalibrated SN data are included, it is not possible to fully solve the tension. The method we develop in this work can be a useful tool to search for any other possible extensions to the $\Lambda$CDM model, and is expected to inspire a model-building effort from the cosmology and particle physics community.

TBA

In stellar evolution, one of the fundamental ingredients is the chemical composition. At the stellar surface, the chemical abundances change during the life of the stars due to the presence of the chemical transport mechanisms. One of these processes is atomic diffusion. This process presents great success in simulating solar-type star evolution.
However, for F-type stars, the inclusion of atomic diffusion shows unrealistic chemical variations at the stellar surface, indicating the need for competing transport processes. In this talk, we present the possibility of parameterizing the effects of the competing processes using the turbulent mixing. We successfully found that it is possible to avoid the chemical over-variations at the stellar surface. We also found that the effects of radiative acceleration in F-type stars can be reproduced using turbulent mixing.
Then, we explored the performance of using the parameterised turbulent mixing in the inference of the global properties from a sample of observed Kepler stars. We also compared its results with the current models that neglect atomic diffusion in F-type stars. We found that considering turbulent mixing and atomic diffusion induce a difference in the inferred age up to 20%.

The modeling and interpretation of molecular spectra from interstellar environments require the knowledge
of both radiative and collisional processes of energy transfer between chemical species. Accurate
state-to-state rate coefficients of interstellar molecules in collision with the main collisional partners H2,
H and He are then necessary.

Such collisional data can be obtained through quantum time-independant scattering calculations by
resolving the close-coupled equations. This method is nowadays one of the most accurate for describing
collisions at very low temperature. However, species such as radicals possess very complex energetic
structures because of the presence of non zero electronic and even nuclear spins. Also, accurate description
of fine and hyperfine transitions implying two interacting molecules is a true theoretical challenge.
Then, this method is not fully applicable in terms of memory and CPU time.

This thesis work consists in quantifying the collisional excitation of these open-shell molecules in astrophysical
media. The development of methodological and numerical tools are then required to overcome
these problems and to characterize physical and chemical processes as precisely as possible. The provided
collisional data are expected to interpret observations and corroborate experiments performed in
laboratory.

The redshifted 21 cm signal from neutral hydrogen is a very useful probe of the cosmic evolution. A particularly interesting phase of this evolution is the Epoch of Reionization (EoR), when the first ionizing sources were born in the early universe. The current and the upcoming low frequency radio telescopes, like the upgraded GMRT (Giant Metrewave Radio Telescope) or the Square Kilometre Array (SKA), are expected to play a crucial role in studying cosmic evolution during the EoR and the post-EoR era, and thus addressing many important but unresolved issues. One of the challenging aspects in this regard is the fast and efficient data processing necessary for detecting the weak cosmological signal in presence of the strong, confusing foreground. However, the foreground is interesting on its own right. In this talk, I will share some of our recent efforts to study the low radio frequency foreground using data from the uGMRT, an SKA pathfinder, as stepping stones towards constraining the 21 cm signal. This work may help us to understand the near-future pathways of studying the atomic hydrogen power spectrum from the EoR/post-EoR era; it also allows us to study, in details, the properties of a large population of extragalactic sources as well as of the interstellar medium of the Milky Way.

Reconciling the standard model's theoretical predictions with precision measurements from multiple probes stands as one of the most pressing challenges of modern cosmology. During the last decade, increasing evidence for persistent discrepancies in the value of cosmological parameters when inferred from model-dependent and -independent probes opened the door for a new paradigm. In this talk, I will address whether (part of) the cosmological tensions can be attributed to missing physics in the standard model. In particular I will focus on tests of different models with interactions between dark energy and dark matter.