[Amath-seminars] Thursday seminar, Mary Gates 389, 4-5pm, Huan Lei
Loyce M. Adams
lma3 at uw.edu
Mon Mar 27 14:05:57 PDT 2017
Please note the room change for this quarter's Thursday seminar.
It is Mary Gates Hall 389. This Thursday, we will have a visitor for
PNNL, Huan Lei. His title and abstract are below. I hope to see all
of you there.
Speaker: Huan Lei, PNNL
Title: A data-driven approach to quantify quasi-equilibrium and
non-equilibrium properties for complex multiphysics systems
One of the grand challenges of modeling multi-scale multiphysics systems
roots in the multifaceted amorphous nature of these systems. Traditional
computational approaches developed on isolated scales, or coupled between
phenomenological descriptions with ad hoc scale separation show limitations
due to lack of universal field variables of such systems. We propose a
data-driven method to quantify quasi-equilibrium and non-equilibrium
properties for complex physical systems with high dimensional stochastic
space based on generalize polynominal chaos (gPC) expansion and Mori-Zwanzig
projection method. To alleviate the high-dimensionality, we propose a
numerical method to enhance the sparsity by defining a set of collective
variables within active subspace, which enable us to accurately quantify
the quasi-equilibrium properties with respect to the amorphous states
where traditional approaches such as sparse grid method shows limitation
therein. Moreover, non-equilibrium properties further depends on the
non-local memory term arising from the high-dimensional unresolved states.
We propose a data-driven method based on appropriate parametrization to
compute the memory kernel of the generalized Langevin Equation (GLE). The
approximated kernel formulation satisfies the second fluctuation-dissipation
conditions with consistent invariant measure. The proposed method enables
us to characterize equilibrium properties such as reaction rate where mean
field approximation shows limitation.
More information about the Amath-seminars