Sommaire:

---- Note the unusual day! ----
The SPT-3G receiver on the South Pole Telescope measures the cosmic microwave background with unprecedented precision. The on-going analysis of SPT-3G data is expected to yield cosmological constraints of higher precision than Planck that are independent of the satellite mission's results. With this constraining power comes the responsibility to employ a flexible, robust, and efficient data analysis pipeline.
In this talk, I present an in-progress view of a new likelihood software developed for SPT-3G data analysis. The likelihood code is written in python and can be used in conjunction with widely available analysis software (e.g. Cobaya). The software is designed to leverage the power of the JAX-library, such that when paired with a differentiable theory code (e.g. CosmoPower) it provides a fully differentiable analysis framework from cosmological parameters to the log likelihood. This pairing gives easy access to robust Fisher matrices with minimal computational cost. I highlight a range of applications that are greatly simplified in this framework and demonstrate how it can be leveraged to carry out a volume of consistency tests on the data otherwise computationally unfeasible. Finally, I provide an overview of other types of applications that are only made possible through the gradient-information, such as more efficient exploration of the likelihood surface.

Pour plus d'informations, merci de contacter Poulin V.