(January 22) Dr. Andreas Veeser:
Stability of Flat Interfaces During Semidiscrete Solidification
-
The stability of flat interfaces with respect to a spatial
semidiscretization of a solidification model is analyzed. The
considered model is the quasi-static approximation of the
modified Stefan problem. The stability analysis relies on an
argument developed by Mullins and Sekerka for the undiscretized
case. The obtained stability properties differ from those with
respect to the quasi-static model for certain parameter values
and relatively coarse meshes. Moreover, consequences on
discretization issues are discussed.
(January 29) Dr. Daniel Kessler :
Numerical Analysis and Simulation of a Solutal Phase-Field Model
-
We characterize a binary alloy during isothermal solidification
by its relative concentration and an order parameter for
solidification: the phase-field. In an isotropic model, the
evolution of these two quantities is described by a coupled
parabolic system with Lipschitz continuous nonlinearities.
Anisotropy depending on the direction of the gradient of the
solution can also be added to the model. We discretize the system
with P1 finite elements in space and a semi-implicit Euler scheme
in time. The convergence of the scheme is proved in the isotropic
case, using a generalized elliptic projector. Finally, for the
anisotropic model we find a clear distinction between low and
high anisotropy regimes via numerical simulations. (February 5) Professor James Stone:
Numerical Models of Accretion Flows Around Black Holes -
Observations of black hole candidates at the center of the Milky
Way and other galaxies indicate they radiate far less energy than
expected based on the inferred rate of accretion of surrounding
plasma. The resolution of this discrepancy may lie in the
dynamics of the accreting plasma. It is now possible to compute
the structure and evolution of accretion flows around compact
objects such as black holes from first principles, by solving the
equations of radiation magnetohydrodynamics using numerical
methods. I will summarize the results of a series of numerical
experiments which investigate the properties of time-dependent
flows in 2D and 3D, and which are providing new insights into the
physics of accretion flows. (February 12) Professor Christoph Schwab:
Two-scale Regularity and Sparse Grids for Homogenization Problems -
We consider elliptic homogenization problems in periodic
structures. We present a two-scale regularity theory for the solutions of
such homogenization problems in d-dimensional domains D (d=1,2,3)
with coefficients and/or domains oscillating at period
e < < diam(D). We show that the solutions u(x,e) allow for high regularity
independent of e if they are viewed as traces of certain
functions in 2d-dimensions. We show that tensorized Finite Elements in dimension 2d converge
independently of e at the price of high complexity. We show that a FEM based on sparse grids in 2d-dimensions
converges independently of e in the physical domain D. The
complexity of the sparse two-scale FEM is shown to equal, up to
log(N), that of standard FEM in the physical d-dimensional domain
on meshes of width e < h < 1. The results are joint work with A.M. Matache and were obtained in
the Project `Homogenization and Multiple Scales 2000'. (February 19) Pingwen Zhang:
Moving Mesh Methods Based on Harmonic Maps
-
In this talk, we will propose a moving mesh method which also
contains two parts, a solution algorithm and a
mesh-redistribution algorithm. A framework for adaptive meshes
based on the Hamilton-Schoen-Yau theory was proposed by Dvinsky.
We will extend Dvinsky's method to provide an efficient solver
for the mesh-redistribution algorithm. The key idea is to
construct the harmonic map between the physical space and a
parameter space by an iteration procedure, and to update the
interior and boundary grids simultaneously, rather than
considering them separately. Application of the proposed moving
mesh scheme is illustrated with some two- and three-dimensional
problems with large gradients. The numerical experiments show
that our methods can accurately resolve detail features of
singular problems in 3D.
(February 26) Victoria E. Howle:
Solving 3D Incompressible Navier-Stokes via Parallel
Block Preconditioning -
We consider parallel steady state and time dependent solutions of
the 3D Lid Driven Cavity problem. This talk focuses mainly on
two issues in resolving the set of nonlinear equations that
result from discretization: (1) nonlinear solver performance and
how different nonlinear solvers yield vastly different linear
subproblem characteristics; (2) solution of the linear
subproblems via block preconditioning along the lines of Kay &
Loghin, Elman, and Silvester & Wathen. Specifically, both
Newton's method and a fixed point technique based on an Oseen
iteration are considered for the nonlinear problem. Block
preconditioning methods using approximate Schur complement
operators in conjuction with algebraic multigrid for subproblems
will be discussed to solve the Oseen submatrix problems. For
Newton's method, two preconditioners are considered: one applied
to an explicitly calculated Jacobian matrix and the other applied
to the Oseen matrix. We illustrate good convergence rates,
similar to those observed by other authors, for two dimensional
steady state and time dependent problems, time dependent three
dimensional problems, and low Reynolds number steady state three
dimensional problems. However, the steady state large Reynolds
number 3D lid driven cavity problem is significantly more
problematic. We will illustrate the utility of these ideas with a
variety of benchmark problems and discuss some of the
difficulties associated solution of steady 3D problems with large
Reynolds numbers, as well as possible remedies. (March 12) Prof. Zhimin Zhang:
Recovery Techniques of the Finite Element Method -
Recovery and post-processing are common practice
in scientific and engineering computation with or without
adaptivity. Some popular recovery methods, such as weighted
averaging, local $L^2$-projection, the Zienkiewicz-Zhu patch
recovery will be discussed and compared. Some superconvergence
results will be presented.
(March 19) Richard S. Falk:
Approximation by Quadrilateral Finite Elements
-
Approximation properties of quadrilateral finite element spaces
of scalar functions and of vector-valued functions defined by the
Piola transform are discussed. In particular, we consider
neccessary conditions on the spaces of functions defined on the
reference element for optimal order approximation on the physical
element after bilinear mappings. As applications, we demonstrate
degradation of the convergence order on quadrilateral meshes as
compared to rectangular meshes for some standard finite element
spaces. (March 19) Prof. Alfred Schmidt:
Multi-mesh finite element methods for coupled systems of PDEs
with applications to thermo-elasticity and problems with phase
transitions -
Many models for physical problems lead to coupled systems of
PDEs, where different components of the solution exhibit a
strongly different local behavior in the underlying domain. An
optimal discretization for a numerical simulation should be
adapted to this behavior and use different locally refined meshes
for different components. We present a general framework of
adaptive finite element methods for coupled systems and show
applications to problems from thermo-elasticity, semiconductor
crystal growth with thermal convection, and phase field
simulations for phase transitions. (April 2) Prof. Tobias von Petersdorff:
Boundary Element Methods for Maxwell Equations in Lipschitz
Domains -
We consider the scattering of electromagnetic waves at a bounded
domain which can either be a perfect conductor or a dielectric
material. We want to use a boundary integral formulation for the
approximation, so that we only have to discretize the boundary of
the scatterer. Until recently, only results for smooth domains
were available (e.g. the book by Nedelec), but most applications
have domains with edges and vertices. We derive equivalent boundary integral formulations for both the
perfect conductor and the dielectric case, and show that boundary
element methods with Raviart-Thomas or Brezzi-Douglas-Marini
elements converge quasioptimally. The main difficulty for time-harmonic Maxwell equations is that
electric and magnetic terms give opposite signs in the
corresponding ``energy bilinear form''. In order to obtain a
coercivity property one has to split functions into two parts
using a ``Hodge decomposition''. For the discrete case we have to
show that there is a discrete Hodge decomposition which is
``close'' to the continuous Hodge decomposition. This is joint work with A. Buffa, R. Hiptmair, C. Schwab. (April 9) Drs. Nail A. Gumerov and Ramani Duraiswami:
Fast Multipole Methods for the Helmholtz Equation: New Fast
Translation Operators -
The 3-D Helmholtz equation arises in a number of important
applications, including acoustical and electromagnetic wave
scattering problems. The use of fast multipole techniques for the
solution of this equation is an active and promising area of
research. In this talk we will first present an overview of the
FMM, with the goal of identifying the different parts of the
algorithm, and the state-of-the-art in developing efficient
implementations for these. One of the most important issues in the FMM is the complexity of
the translation operation. The practical implementation of an FMM
algorithm and the cross-over point at which it begins to be
useful, are critically dependent on this step. Previous authors
have either accomplished translations with expensive algorithms,
or with more inexpensive methods that use an approximate
quadrature which causes additional errors and stability problems. To solve this problem we derive new "translation theorems" for
the fundamental solutions of the Helmholtz equation. Results of
some multiple scattering problems using these theorems will be
presented. Using the obtained relations we are able to derive a
sparse matrix decomposition of the FMM translation operators for
the Helmholtz equation. This decomposition allows evaluation of
translations using the theoretical minimum of operations
required. Our expressions do not require approximate quadrature
as do those used in the literature, and are much more
computationally efficient. We also can derive tighter
error-bounds and complexity estimates for the Helmholtz FMM. (Support of NSF Award 0086075 is gratefully acknowledged)
(April 16) Professor John Osborn:
On Principles for the Selection of Shape Functions
for the Generalized Finite Element Method -
Effective shape functions for the Generalized Finite Element Method should
reflect the available information on the solution. This information is fuzzy
because the solution is, of course, unknown, and, typically, the only
available information is the solution's inclusion in various function spaces.
It is desirable to select shape functions that perform robustly over a family
of relevant situations. Quantitative concepts of robustness are introduced and
discussed. We show in particular that, in one dimension, polynomials are
robust when the only available information consist in inclusions in the usual
Sobolev spaces. If some additional information is available, if, {\it e.g.,}
the approximated function is constrained by certain boundary conditions, then
polynomials may perform poorly---relative to the optimal shape functions for
approximating functions satisfying the boundary conditions---and
some other family of shape functions should be used. This talk is based on
joint work with I. Babu\v{s}ka and U. Banerjee. (April 23) Dr. Cheng Wang:
A Fourth Order Numerical Solver for the PlaneGeostrophic Equations Using Staggered Grid
-
The numerical method system of the planetary geostrophic
equations (PGE), which has played a central role in large-scale
ocean circulation theory, is taken into consideration. Such
system can be viewed as asymptotic description of the primitive
equations (PEs) of the atmospheric and oceanic flow as the Rossby
number approaches to zero. There is only one prognostic equation
in the system, which is for the temperature field. The horizontal
velocity field is statically determined by the planetary
geostrophic balance, and the vertical velocity field is
determined by incompressibility condition, along with its
vanishing boundary condition. A fourth order finite difference method for the PGEs using MAC
grid is analyzed in the talk. Different physical variables are
evaluated on staggered numerical grid points. At each time step
(stage), the temperature transport equations is explicitly
updated by fourth order long-stencil approximation to the
convection and diffusion terms. Subsequently, the temperature
gradient can be obtained by finite difference operation. The
horizontal velocity field is solved by a fourth order
approximation to a first order ODE in vertical direction at each
fixed horizontal point by using the combined information of the
discrete value of the temperature gradient and the vanishing
vertically averaged horizontal velocity field in the earlier
derivation. The vertical velocity field is solved by a compact
fourth order scheme for a second order ODE in vertical direction
at each fixed horizontal point, with the vanishing Dirichlet
boundary condition at the top and bottom. The proposed method is shown both numerically and theoretically
to preserve stability and fourth order convergence. Some
numerical results, including an accuracy check and a physical
example of a large-scale thermocline circulation, are presented.
(May 2) Prof. Randolph E. Bank:
A New Paradigm for Parallel Adaptive Meshing Algorithms
-
We present a new approach to the use of parallel computers with
adaptive finite element methods.
This approach addresses the load balancing problem in a new
way, requiring far less communication than current approaches.
It also allows existing sequential adaptive PDE codes such as
pltmg and mc to run in a parallel environment without a large
investment in recoding.
In this new approach, the load balancing problem is reduced
to the numerical solution of a small elliptic problem on a single
processor, using a sequential adaptive solver, without requiring
any modifications to the sequential solver.
The small elliptic problem is used to produce {\em a
posteriori} error estimates to predict future element densities
in the mesh, which are then used in a weighted bisection of the
initial mesh.
The bulk of the calculation then takes place independently on
each processor, with no communication, using the same sequential
adaptive solver.
Each processor adapts its region of the mesh independently,
and a nearly load-balanced mesh distribution is usually obtained
as a result of the initial weighted bisection.
Only the initial fan-out of the mesh decomposition to the
processors requires communication. Two additional steps requiring boundary exchange
communication are employed after the individual processors reach
an adapted solution, namely the construction of a global
conforming mesh from the independent subproblems, followed by a
final smoothing phase using the subdomain solutions as an initial
guess. We present a series of convincing numerical experiments which
illustrate the effectiveness of this approach. This work is joint
with Michael Holst. (May 3) Prof. Randolph E. Bank:
Multigrid: From Fourier to Gauss
-
Multigrid iterative methods are one of the most significant
developments in the numerical solution of partial differential
equations in the last twenty years. Multigrid methods have been
analyzed from several diverse perspectives, from Fourier-like
spectral decompositions to approximate Gaussian Elimination, with
each perspective yielding new insight. In this lecture we will
introduce the multigrid method, and survey several of the tools
used in its analysis. (May 7) Prof. Phillip Colella:
Adaptive Mesh Refinement for Multiphysics Problems
-
In this talk, we will discuss a variety of algorithmic ideas that
arise in extending the block-structured adaptive mesh refinement
approach of Berger and Oliger to time-dependent problems in which
a mixture of elliptic, parabolic and hyperbolic behaviors appear,
such as in imcompressible fluid flows, low-Mach-number
combustion, and problems with volume-of-fluid representations of
fixed and free surfaces. Some of the issues to be addressed
include the formulation of appropriate coarse-fine matching
conditions; the influence of refinement in time on matching
conditions for elliptic and parabolic problems, and on
free-stream preservation; and error estimators at refinement
boundaries. We will also discuss some of the software design
issues that arise in supporting a diverse collection of
algorithms, applications, and users. (May 14) Prof. Kunibert Siebert:
Fully Localized A Posteriori Error Estimators and Barrier Sets for
Contact Problems -
We discretize the elliptic obstacle problem, which constitutes a
model contact problem, with finite elements. Enforcing the
unilateral constraint solely at the nodes, we obtain a practical,
but in general non-conforming method. We present a~posteriori
error estimators for the maximum norm error. They are reliable,
exhibit optimal order, and, except for the resolution of the
obstacle, vanish within the discrete contact set. Moreover, we
use these estimators to construct a~posteriori barrier sets for
free boundaries under a natural stability condition on the
continuous problem. We illustrate these results with simulations
in 2d and 3d showing the impact of localization in mesh grading
within the contact set along with quasi-optimal meshes. This is joint work with R.,H. Nochetto (University of Maryland)
and A. Veeser (Universita degli Studi di Milano).
|