Title: Sampling from high-dimensional constrained distributions
Speaker: Dr. Vissarion Fisikopoulos (Senior software engineer at Oracle and Coordinator of the GeomScale project)
Abstract: Sampling from a given probability distribution constrained by a convex
space is a fundamental problem that arises across various fields, from
optimization to quantitative finance and systems biology. These problems
are critical for modeling uncertainty, making decisions under
constraints, and understanding complex systems. However, designing
efficient sampling algorithms in high-dimensional spaces remains a
significant challenge.
In this talk, we will introduce the relevant theoretical background,
define the main problems, and delve into the core challenges associated
with sampling in constrained spaces. We will explore how these
challenges manifest in practical applications, briefly touching on real
world examples from biology and finance. Furthermore, we will outline
the current state-of-the-art methods available for some important
constraint models, focusing on their computational efficiency,
scalability, and accuracy.
From a practical viewpoint, we will demonstrate the implementation of
these methods in open-source software, providing hands-on insights into
their implementation and applicability. The discussed software is
publicly available at \url{https://github.com/GeomScale}, enabling
researchers and practitioners to integrate these advanced sampling
techniques into their workflows.
Short Bio: Previously, he was a postdoc researcher at Algorithms Research Group in Université libre de Bruxelles (ULB) working with Prof. Samuel Fiorini. Before that he was a visiting researcher at National Institute for Mathematical Sciences in South
Korea and researcher at the University of Athens.
He holds a Phd in Computer Science, from University of Athens supervised by Prof. Ioannis Emiris. Also, a M.Sc. in Logic, Algorithms and Computation, from University of Athens. His diploma (5-year degree) is in Computer Engineering & Computer Science from CEID,
University of Patras.
You can attend online via MS-Teams by clicking the following link
