Nicolas Bartolomeo

Nicolas Bartolomeo

Biography

Nicolas Bartolomeo is a rising junior majoring in Mathematics, minoring in ComputerScience, and working towards a certificate in Geographical Information Science at the University of Maryland, Baltimore County where he is a member of the Meyerhoff Scholars Program. This program aligns with his values of supporting underrepresented groups in STEM and supports his mission of pursuing a PhD in Applied Mathematics. His research experiences include an internship at the Ohio State University and a sustained project in the Biswas and Hoffman lab at his home institution. At OSU, he investigated the relationship between extreme weather event factors and median rent prices in the aftermath. At UMBC, he is researching applications of Markov-ChainMonte-Carlo Simulations for earthquake modeling. From these experiences, he gained an interest in computational and mathematical modeling applications in settings that contribute toward change-provoking conclusions. While participating in the MIT Summer Research Program, he is working under the guidance of Dr. Benoit Forget and Dr. W. Reed Kendrick in the Computational ReactorPhysics Group. Outside of research, he is a Teaching Assistant in the Mathematics department and a member of Pi Mu Epsilon, STEM PhD Society, and Student Government Association.

Abstract

Developing a Novel Self-Scatter Sampling Scheme for the Neutron Transport Equation

The neutron transport equation allows us to understand the interactions of neutrons and is used to model their behavior and distribution within nuclear reactor settings. This equation is typically solved numerically and can be quite costly; however, approximation techniques are often employed to mitigate the high computational requirements. One common simplification involves approximating the angular variable within the equation that describes a neutron’s direction of travel. While this is typically beneficial for most nuclides, it can lead to errors in edge cases where there is an ‘over-correction’, resulting in negative probabilities that cause simulations in MonteCarlo settings to break down. This project develops an algorithm that treats corrections differently by adding a cap to avoid these negative probabilities and adjusting the sampling kernel of the scattering angle to replicate the desired behavior. Through incremental testing of increasingly complex settings to refine its implementation and generalization, we expect to be able to run these simulations in Monte Carlo settings without fail. This will enable Monte Carlo simulations using simplified data models to validate deterministic simulations, thus providing high-fidelity verification of the models. On a larger scale, this will strengthen results used towards predicting the behavior of nuclear reactors.