Redshift Survey

Redshift surveys map the three-dimensional distribution of galaxies to constrain cosmological parameters, aiming to understand the universe's large-scale structure and evolution. Current research focuses on improving inference methods using machine learning, particularly neural networks like graph neural networks and PointNets, to analyze the complex, high-dimensional data from these surveys more efficiently and robustly, often bypassing traditional summary statistics. These advancements, including the use of Bayesian neural networks and accelerated inference techniques like those offered by CosmoPower-JAX, are crucial for handling the massive datasets expected from next-generation surveys and extracting more precise cosmological information.