Seminar: Potential Gradient Descent Acceleration of L-PICOLA Simulations with Non-Gaussian Initial Conditions

Title: Potential Gradient Descent Acceleration of L-PICOLA Simulations with Non-Gaussian Initial Conditions
Speaker: Greco Alejandro Peña Pinto

Abstract:

Cosmic structure formation helps us understand how the universe evolved, from small primordial perturbations to large superclusters. Fast N-body codes generate dark matter fields quickly but often miss small-scale details.

Machine learning can help improve this resolution. We tested the Potential Gradient Descent (PGD) method on L-PICOLA simulations with non-Gaussian initial conditions.

We present preliminary results for the power spectrum and halo mass function. This seminar is ideal for anyone interested in machine learning, cosmological simulations, or structure formation.

Date / Time: 28 November 2025, 11:15–12:00
Location / MS Teams: OCW 0017 / Join on MS Teams