8–12 Sept 2026
CBPF
America/Sao_Paulo timezone

Machine Learning-Based Emulators for the non-linear Matter Power Spectrum in Current Cosmological Surveys

Not scheduled
20m
Auditório Ministro João Alberto Lins e Barros (CBPF)

Auditório Ministro João Alberto Lins e Barros

CBPF

Rua Dr. Xavier Sigaud, 150 - Urca Rio de Janeiro - RJ - Brasil CEP: 22290-180
Poster + Fireslide

Speaker

Vicente Pedreros (Universidad de Chile)

Description

Unraveling the nature of the cosmic expansion of the Universe, driven by Dark Energy (DE), remains one of the key challenges in modern cosmology. Modified Gravity (MG) models provide an alternative to the standard paradigm of the cosmological constant. These models predict distinct signatures in the growth of structures, yet analytical predictions are challenging due to the highly non-linear equations involved. To overcome this problem, we study the approach to MG as an effective DE fluid. To this end, we have incorporated linear DE perturbations into the N-body code \concept, which relies on the Einstein–Boltzmann solver \texttt{CLASS}. We replaced the \texttt{CLASS} version used by \concept\ with \texttt{mochi_CLASS}, a patched version that includes models within the Horndeski family. By writing linear DE perturbations ($\delta\rho_{\text{DE}}$, $\delta P_{\text{DE}}$, and $\sigma_{\text{DE}}$) in terms of metric perturbations in the synchronous gauge, the N-body code is able to evolve the effective fluid using the Boltzmann equations. Our current simulations encompass a range of cosmological scenarios –$w_0w_a$CDM, $f(R)$ Hu-Sawicki, Cubic Galileon, and Brans-Dicke–, each extended to include massive neutrinos. Furthermore, we are exploring artificial neural network architectures, including Long Short-Term Memory (LSTM) models, to develop an emulator of the non-linear matter power spectrum. This emulator would provide a fast and flexible tool to confront MG predictions with data from current and upcoming cosmological surveys, such as Euclid, DESI, 4MOST, LSST, thereby tightening the constraints on DE parameters in these models.

Author

Vicente Pedreros (Universidad de Chile)

Co-author

Prof. Domenico Sapone (Universidad de Chile)

Presentation materials

There are no materials yet.