A point cloud approach to generative modeling for galaxy surveys at the field level

TL;DR

This paper presents a diffusion-based generative model for galaxy distributions in 3D space, enabling both data emulation and inference without binning, demonstrated on dark matter haloes in simulations.

Contribution

It introduces a novel point cloud diffusion model for galaxy surveys, allowing direct modeling of 3D galaxy distributions and conditional likelihood computation.

Findings

Successfully reproduces key galaxy distribution statistics

Enables inference of galaxy fields from simulation data

Offers a flexible framework for cosmological data analysis

Abstract

We introduce a diffusion-based generative model to describe the distribution of galaxies in our Universe directly as a collection of points in 3-D space (coordinates) optionally with associated attributes (e.g., velocities and masses), without resorting to binning or voxelization. The custom diffusion model can be used both for emulation, reproducing essential summary statistics of the galaxy distribution, as well as inference, by computing the conditional likelihood of a galaxy field. We demonstrate a first application to massive dark matter haloes in the Quijote simulation suite. This approach can be extended to enable a comprehensive analysis of cosmological data, circumventing limitations inherent to summary statistic -- as well as neural simulation-based inference methods.