Abstract: Modern maps of the Universe reveal that on large scales, galaxies are not uniformly distributed, but trace out intricate patterns—filaments, clusters, and vast empty regions called voids. Together, these structures form the 'cosmic web', which arises from the gravitational clumping of dark matter. Reconstructing the 3D distribution of matter  is important for understanding how the Universe evolves. In this talk, I will introduce a new method that uses neural networks to recover the underlying matter density and velocity fields from observed galaxy data. I will also explain how neural networks connect to more familiar statistical tools, helping to demystify them. I will then compare our approach with a traditional technique called the Wiener filter and show how the neural network does a better job of capturing the more complex non-linear patterns in the data. Finally, I will present our recent results using this method on real observations from the 2MRS galaxy survey, which allows us to create a new map of the nearby Universe and show some new results on physics inspired machine-learning.

About the Speaker: Punyakoti Ganeshaiah Veena is currently a postdoc at the University of Genova in Italy. Prior to this, she was at the Technion in Israel. She completed her PhD in cosmology from the University of Groningen, the Netherlands. She specialises in computational cosmology and uses neural networks to study large-scale structures of the Universe.

We look forward to your active participation.