Ivan Casas-Rocha

Hi! I am a Ph.D. student in mathematics at the University of Minnesota, advised by Yulong Lu. My research provides theoretical guarantees and understanding of score-based generative models, and sampling, using tools from stochastic analysis, functional inequalities, and high-dimensional probability. I am particularly interested in understanding the structure of generative models in high-dimensions.

I completed my B.S. in mathematics at the University of Central Florida, where I did my undergraduate thesis under Dr. Robert Jenkins on the Inverse Scattering Transform for the Nonlinear Schrödinger equation. Previously, I worked with Dr. Zi-Xia Song on extremal graph theory, particularly co-critical graphs.

email: casas061 umn edu

GitHub · LinkedIn · CV

News

  • Nov 2026: Poster, Delocalization of Bias for Score-based Generative Models, at the NSF Mathematical Foundations of Artificial Intelligence Conference, University of Pennsylvania.
  • Oct 2026: Attended the Flatiron Institute x IBM workshop Emerging Directions in Probabilistic Modeling (sampling, diffusion models, and optimal transport) and presented a poster.
  • Summer 2026: Attended the Summer School on Stochastic Dynamics and Theoretical Machine Learning at Duke University.