Séminaire Doctoral / Seminar PhD |
| « A hierarchical approach for field level inference » |
|
Anirban Bairagi |
| Advancing cosmological parameter inference with reduced uncertainties is a vibrant area of research, especially with the wealth of data from next-generation surveys like Euclid, DESI, and the Vera Rubin Observatory. This talk focuses on Simulation-Based Inference (SBI), utilizing summary statistics such as the Power Spectrum (P(k)) and Bispectrum (B(k)). However, these summaries fail to fully harness the non-Gaussian and non-linear features of the cosmological density field. To extract this crucial information, a field-based analysis is preferable, although its success hinges on model architecture, hyperparameter tuning, and the ability to fit high-resolution density fields into GPUs. We introduce a hierarchical approach that combines small-scale information from sub-volumes (patches) with large-scale information from the Power Spectrum. Our method demonstrates enhanced Fisher information about the parameters compared to using P(k) or B(k) alone. |
|
vendredi 23 février 2024 - 16:00 Salle des séminaires Évry Schatzman, Institut d'Astrophysique |
| Page web du séminaire / Seminar's webpage |