Events

Back to Upcoming EventsThis Event has Passed

FDS Postdoctoral Applicants

Gibbs sampling from log-concave distributions under smoothness assumptions

Speaker: Neha Wadia Flatiron Institute Thursday, December 19, 2024 10:00AM - 11:00AM and via Webcast: https://yale.zoom.us/j/99788429879

“Gibbs sampling from log-concave distributions under smoothness assumptions”

Abstract: The Gibbs sampler, also known as the coordinate hit-and-run algorithm, is a Markov chain that is widely used to draw samples from probability distributions in arbitrary dimensions. At each iteration of the algorithm, a randomly selected coordinate is resampled from the distribution that results from conditioning on all the other coordinates. Although the Gibbs sampler is several decades old, non-asymptotic guarantees that identify the dimension dependence of its convergence behavior have only just begun to emerge. Building on the recent work of Aditi Laddha and Santosh Vempala, in which they establish a mixing time bound for Gibbs sampling from a uniform distribution supported on a convex body in $\mathbb{R}^n$ that is polynomial in $n$, I will discuss a new mixing time bound for Gibbs sampling from strongly log-concave distributions supported on $\mathbb{R}^n$ under some smoothness assumptions.

Bio: Neha Wadia is a postdoctoral fellow at the Center for Computational Mathematics of the Flatiron Institute. She holds undergraduate and masters degrees in physics from Amherst College and the Perimeter Institute, respectively, and a PhD in biophysics from the University of California, Berkeley. Neha is interested in foundational problems in machine learning and the theory of computing, and in algorithmic challenges at the intersection of machine learning with the natural sciences. Outside of research, Neha spends a lot of time thinking about, reading about, and in pursuit of good food.

Add To: Google Calendar | Outlook | iCal File