Fellow members of the Mathematics Department,
There will be an *in person* CAM Seminar
TODAY 10/20/21
3:35 - 4:35pm
Ayres 112
The speaker, title, and abstract are below.
Best,
Abner
------------------
Speaker: Paul Laiu
Affiliation:ORNL
Title: Data-driven approximation to entropy-based moment closures
Abstract:
Moment models approximate the kinetic equations by tracking the
evolution of a small number moments of the kinetic distribution. The
behavior of these models depends heavily on the moment closure, which
prescribes the kinetic information that is lost in the moment approach.
Entropy-based moment closures inherit many structural features of
kinetic equations, while their use is limited by several implementation
challenges. In this talk, I will present a data-driven approach to
construct entropy-based closures. The proposed closure learns the
entropy function by fitting the map between the moments and the entropy
of the moment system, and thus does not depend on the space-time
discretization of the moment system and specific problem configurations
such as initial and boundary conditions. With convex and
approximations, this data-driven closure inherits several structural
properties from entropy-based closures, such as entropy dissipation,
hyperbolicity, and H-Theorem. We illustrate this approach for a simple
linear transport equation in slab geometry. For two-moment models, a
convex fit can be constructed with splines. For larger systems, convex
splines are not available, so we resort to a fit that uses a neural
network. We test the approximation on two- and three- moment systems
and find that the resulting systems provide a cheaper alternative to
standard entropy-based closures.
This is joint work with Graham Alldredge (Berlin), Martin Frank
(Karlsruhe), Cory Hauck (Oak Ridge), and Will Porteous (Austin).
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
To leave the list go here:
https://listserv.utk.edu/cgi-bin/wa?SUBED1=MATHTALK&A=1
|