×
Methods and Scientific Applications
This workshop brings together recent advances in probabilistic modeling, with a focus on emerging methods for representing, learning, and sampling complex probability distributions. Topics of interest include kernel and transport-based methods, discrete and structured generative models, conditional and constrained modeling, and new perspectives on density modeling and sampling.
The workshop also highlights scientific applications of these ideas, including molecular dynamics, inverse problems, climate and weather modeling, and related problems in the physical and life sciences. More broadly, it aims to bring together researchers from machine learning, statistics, applied mathematics, and computational science to discuss common ideas, emerging challenges, and new opportunities in modern probabilistic modeling.
Optimal transport, kernel-based sampling, and flow-based approaches for distribution modeling
Discrete sampling and generative modeling via diffusion models and flow matching
Advanced MCMC methods, diffusion-based sampling, and density estimation
Conditional generation, constrained optimization, and model steering
Probabilistic methods for PDEs, molecular simulation, free energy estimation, biology, and climate and weather modeling
We are excited to feature talks from leading researchers in probabilistic modeling, machine learning, and scientific applications.
The Flatiron Institute is the internal research division of the Simons Foundation, home to the Center for Computational Mathematics.
IBM's state-of-the-art research and innovation hub in the heart of Manhattan, featuring collaborative spaces and cutting-edge facilities.
In-person applications are now closed, and selected participants have been notified. The workshop will be streamed as a Zoom webinar, open to everyone.
Register for the Zoom webinar to watch the talks live.
Register for Zoom WebinarIn-person applications closed on August 22, 2026
