Applying PQMass to Learning the Universe cosmological simulations

Sammy Sharief ( Université de Montréal )

Cosmological simulations are vital in studying the universe’s initial conditions and evolution to present days. However, due to high computational cost, these simulations do not span the same range as surveys, limiting their use. In the face of these challenges, many simulations, such as those by the Learning the Universe collaboration, are utilizing generative models to help lower the computational cost of these simulations. Crucially, to ensure the accuracy of these generative models, we employ the PQMass, which provides a probabilistic assessment of model quality, allowing us to detect biases and ensure that the generated simulations faithfully represent the statistical properties of observational data. By applying PQMass, we enhance our ability to infer cosmological parameters and initial conditions.