Research
Meet Robert Pearce
LEADS Lab
August 2025 - Present
Undergraduate Researcher
Under the supervision of Dr. Paul La Plante in UNLV’s LEADS Lab, I use the Bridges-2 supercomputer at the Pittsburgh Supercomputing Center to run large-scale numerical simulations with the Zerion reionization model. These simulations produce kSZ temperature maps, optical depth estimates, and ionization histories, which I analyze by computing their angular power spectra (Cl) to quantify statistical features of the reionization process. The central objective of this work is to develop a machine learning–based emulator capable of predicting the kSZ angular power spectrum directly from reionization model parameters. To construct the training set, I perform Latin Hypercube Sampling over the parameter space, process simulation outputs stored in HDF5 format, and derive the corresponding Cl spectra through Fourier-based power spectrum estimation. I then train a neural network to learn the mapping from physical model parameters to their predicted spectra, enabling rapid exploration of cosmological parameter space without the computational cost of running new simulations. This project integrates high-performance computing, numerical cosmology, and machine learning to create a scalable tool for studying the timing and structure of reionization. It is part of my ongoing undergraduate research with Dr. Paul La Plante, utilizing the Bridges-2 supercomputer at the Pittsburgh Supercomputing Center.