Investigating the star formation rate density (SFRD) of galaxies throughout cosmic time is a fundamental inquiry in the field of modern astrophysics and cosmology. Galaxy surveys have been employed to trace the star formation rate (SFR) of galaxies up to a redshift of approximately z ~ 11. However, these surveys fail to capture the complete information due to the presence of multiple faint and undetectable galaxies, particularly at high redshifts. To complement traditional surveys, the line intensity mapping (LIM) technique has emerged as a valuable approach, which involves measuring the spatial fluctuations of the total spectral-line emission from galaxies and the intergalactic medium. There are many LIM experiments running and even more planned for the near future. All these experiments target different spectral lines depending on their objective. Some of the most popular lines are: CO, [CII], HI 21 cm, Hα, and Lyα.
Investigating the star formation rate density (SFRD) of galaxies throughout cosmic time is a fundamental inquiry in the field of modern astrophysics and cosmology. Galaxy surveys have been employed to trace the star formation rate (SFR) of galaxies up to a redshift of approximately z ~ 11. However, these surveys fail to capture the complete information due to the presence of multiple faint and undetectable galaxies, particularly at high redshifts. To complement traditional surveys, the line intensity mapping (LIM) technique has emerged as a valuable approach, which involves measuring the spatial fluctuations of the total spectral-line emission from galaxies and the intergalactic medium. There are many LIM experiments running and even more planned for the near future. All these experiments target different spectral lines depending on their objective. Some of the most popular lines are: CO, [CII], HI 21 cm, Hα, and Lyα.
Team Members:
Objectives:
Given the imminent surge of data expected in the upcoming years from all the planned experiments, it is of high importance to develop tools and models that will be able to manage and comprehend the forthcoming data. Therefore the objectives of this project can be summarized as follows:
- Undertake the challenge of component separation in scenarios where multiple spectral lines are blended within a map, with the objective to first denoise the map and then disentangle the individual spectral lines/components
- Given an intensity map extract astrophysical or cosmological information (e.g., luminosity functions, star formation rate, halo bias, matter power spectrum)
Methodology: The initial focus of my methodology will be shifted to build a large data set to train and test this network. I plan to do this using the SIDES-Uchuu simulation. This is a powerful and realistic simulation that replicates the far infrared (FIR) and sub-millimeter sky and eventually simulating the anticipated LIM data from a wide range of LIM experiments. Obtaining the training dataset will demand running the simulation for different and multiple realizations of the dark matter halo catalogues, which serve as the starting point of the simulation. In this way I will be able to generate both the overall signal map and the individual component maps. For a better visualization of this concept see the following figure. The next step will involve deciding the most suitable neural network architecture (e.g., CNN, GAN) best tailored for the purposes of this project.