ΛCDM (Λ - Cold Dark Matter) is the current most widely accepted paradigm to describe the cosmology of the Universe. Here, Λ is the cosmological constant initially proposed by Einstein which is associated with dark energy (which makes up most of the Universe) and 'cold' dark matter, which suggests dark matter particles with velocities far less than the speed of light c. The ΛCDM model is able to describe the cosmology of the Universe, mostly consistent with observations and theoretical predictions based on the standard model of particle physics and assuming general relativity on cosmological scales. This model is able to predict an expanding Universe and large-scale structure formation due to density and quantum fluctuations evolving over cosmic time, which is supported by the observations of the cosmic microwave background emission and studies of the evolution of large-scale structures. However, there are some observed discrepancies that the ΛCDM model cannot explain, and this study is built on one such premise.
Team Members:
Objective:
The objective of this study is to conduct a morpho-kinematic analysis of high redshift quasar sources and understand their role in the overall galaxy evolution timeline. The goal is to study the observed data of the multiple merger Hot DOG system - W2246-0526, also known as W2246.
● Separating out the different components from spectral datacubes:
○ Allowing for a more quantitative identification of different sources which were only previously identified qualitatively.
○ Allowing for a better dynamics of interacting systems at high redshifts
● Developing novel deep learning algorithms to analyze the role of different high-redshift sources in the galaxy evolution timeline
Methodology:
For the initial steps of this project:
● Spectral cube simulations will be used, with controllable parameters, and component separation techniques will be applied to analyze the feasibility of such techniques on real data.
○ The first type of spectral cube simulations would be very simple ones (generated from Gaussians and the like) where each and every parameter (scale length, width, angle of inclination, axis of rotation, velocity magnitudes, etc.) can be controlled.
○ The second type would be spectral cubes generated from state-of-the-art cosmological simulations incorporating radiative transfer from the FIRE (Feedback in Realistic Environments) simulation project, where mock IFU cubes are being constructed by collaborators at the University of Connecticut.
● Component analysis and separation methods such as principal component analysis (PCA), Independent component analysis (ICA), Generalized Morphological Component Analysis (GMCA), etc. would be implemented. The algorithms consistent with the simulated cubes would be implemented on the ALMA observations of W2246, and eventually JWST observations, and the different kinematics of the sources would be analyzed.
● Deep learning techniques will be developed to predict the role of high redshift Hot DOGs in the galaxy evolution timeline
○ The training dataset would be constructed using datacubes from cosmological simulations (example image from FIRE on the right), as the time-evolution of sources is possible. In addition, due to being theoretical, the simulations provide a wealth of data to make robust datasets for machine learning.
○ The goal is to train the model and evaluate them on simulated as well as observed datacubes and have the model predict the stage of galaxy evolution by learning about the various stages of simulated galaxy evolution via the time-evolved sources.