12:00–1:00 pm ERC 401
Nick Kokron (Stanford University) "Improving models of structure formation statistics by combining perturbation theory and simulations"
A deluge of high precision data from Stage IV cosmological galaxy surveys is imminent. One of the key challenges that faces cosmology today is ensuring models for the statistics measured by these surveys, such as galaxy clustering and lensing, are sufficiently accurate so that the most information is extracted from these data. In this talk I'll discuss a promising new class of models, known as hybrid effective field theory (hybrid EFT), that combines perturbation theory-based approaches with the fully nonlinear dark matter distributions produced by cosmological N-body simulations. I will survey a few recent results pointing to hybrid EFT being a capable model to understand the tracer-matter connection. Namely, I'll show how an emulator based on hybrid EFT can describe clustering and lensing statistics to scales of k~0.6 h/Mpc to 1% accuracy, how these models can be used to describe beyond two-point statistics such as k-Nearest Neighbour Cumulative Distribution Functions (kNN-CDFs) and how they may be used as field-level descriptors of galaxies to learn about their stochasticities. I will also point to future directions of work enabled by hybrid EFTs and the challenges to deploying them at full industrial scale as fiducial models for next generation surveys.
