|
|
|
|
|
DMS colloquium talk by Wei Cai (Southern Methodist University)
DMS colloquium talk by Beatrice Riviere (Rice University)
T. T. Phuong Hoang
Abstract: We present and analyze a
general framework for constructing high order explicit
local time stepping (LTS) methods for
hyperbolic conservation laws. In
particular, we consider the model problem discretized
by Runge-Kutta discontinuous Galerkin (RKDG) methods
and design LTS algorithms based on the strong
stability preserving Runge-Kutta (SSP-RK) schemes that
allow spatially variable time step sizes to be used
for time integration in different regions of the
computational domain. The proposed algorithms are of
predictor-corrector type, in which the interface
information along the time direction is first
predicted based on the SSP-RK approximations and
Taylor expansions, and then the fluxes over the region
of the interface are corrected to conserve mass
exactly at each time step. Following the proposed
framework, we derive the corresponding LTS
schemes with accuracy up to the fourth order, and
prove their conservation property and nonlinear
stability for the scalar conservation laws. Numerical
results on various test cases are also presented to
demonstrate the performance of the proposed LTS
algorithms.
Sanghyun Lee
Location
and time: Parker Hall 328, 2:00 pm - 3:00 pm
Abstract: We present and analyze
enriched Galerkin finite element methods (EG) to
solve coupled flow and transport system in porous
media such as viscosity and density-driven flows.
The EG is formulated by enriching the conforming
continuous Galerkin finite element method (CG)
with piecewise constant functions. This approach
is shown to be locally and globally conservative
while keeping fewer degrees of freedom in
comparison with discontinuous Galerkin finite
element methods (DG). Linear solvers and dynamic
mesh adaptivity techniques using entropy residual
and hanging nodes will be discussed. Some
numerical tests in two and three dimensions are
presented to confirm our theoretical results as
well as to demonstrate the advantages of the EG.
No
seminar
Yingda Cheng
Braxton Osting
DMS colloquium talk by Robert Lipton (Louisiana State University)
Xiaojing Ye
Department of Mathematics and Statistics,
Georgia State University
Abstract: Decentralized consensus optimization
has extensive applications in many emerging big data,
machine learning, and sensor network problems. In
decentralized computing, nodes in a network privately
hold parts of the objective function and need to
collaboratively solve for the consensual optimal
solution of the total objective, while they can only
communicate with their immediate neighbors during
updates. In real-world networks, it is often difficult
and sometimes impossible to synchronize these nodes, and
as a result they have to use stale and stochastic
gradient information which may steer their iterates away
from the optimal solution. In this talk, we focus on a
decentralized consensus algorithm by taking the delays
of gradients into consideration. We show that, as long
as the random delays are bounded in expectation and a
proper diminishing step size policy is employed, the
iterates generated by this algorithm still converge to a
consensual optimal solution. Convergence rates of both
objective and consensus are derived. Numerical results
on some synthetic optimization problems and on real
seismic image reconstruction will also be presented.
Minglei Yang
Abstract: The semi-linear
nonlocal diffusion equations have been
applied in a wide variety of applications,
including, e.g., contaminant flow in
groundwater, the dynamics of financial
markets and plasma physics. In this work, we
focus on developing and analyzing novel
probabilistic numerical approaches for solving
several types of semi-linear nonlocal
diffusion equations in both unbounded and
bounded high dimensional spaces. And we also
introduce one application of initial-boundary
value partial integro-differential equation
problem, the approximation of the runaway
probability of electrons in fusion tokamak
simulation. Analysis of the approximation
errors of the proposed schemes and several
numerical examples will be presented to verify
the accuracy and effectiveness of our
approaches.
Somak Das
Abstract: Most of our
contemporary mathematical models are based on partial
differential equations. However, the varied levels of
randomness pose difficulties for such systems to be
accurately modeled using deterministic partial
differential equations. In such settings we use
stochastic partial differential equations to incorporate
the randomness. To determine the optimal control for the
stochastic system in this project, we adopt the
stochastic gradient descent algorithm. With vast
data-sets being customary for training of most machine
learning algorithms, the stochastic gradient descent
method is one of the efficient ways to obtain the
optimal control. Another class of algorithms, adaptive
gradient, has also widespread applications in large
scale stochastic optimizations. The algorithm adjusts
its step-size at every iteration depending on the
current gradient value unlike stochastic gradient where
we need to re-tune the step-size manually. In this talk
we show the results obtained from these algorithms.
|
Thi-Thao-Phuong Hoang, Auburn University |