Math 621 - Numerical Methods for Partial Differential Equations
Fall 2010 - Matthias K. Gobbert
Presentations of the Class Projects
-
Friday, December 17, 2010, 01:05-01:15
Modeling of Intracellular Calcium Waves and Sparks
in one Spatial Dimension
Zana Coulibaly, Department of Mathematics and Statistics
A model for the flow of calcium on the scale of one heart cell is
given by a system
of time-dependent reaction-diffusion equations coupled by non-linear
stochastic reaction
terms and discrete release site. In this project, we turn the model from
stochastic to deterministic and use
finite difference to simulate the behavior of the system on different
mesh size. The results we obtain
suggests that solving the problem on finer meshes results in solving a
different problem than having a
smoother approximation to the solution.
This work is joint with advisor Dr. Brad Peercy.
-
Friday, December 17, 2010, 01:20-01:30
Numerical Solution of a Distributed Optimal Control Problem with
Semilinear Elliptic PDE Constraints Using Adjoint Methods
Jyoti Saraswat, Department of Mathematics and Statistics
In this project we solve distributed optimal control problem with
semilinear elliptic PDE constraints. Our aim is to find the optimal
control variable and the state variable which minimizes the cost
functional. Each cost functional evaluation requires a nonlinear PDE
solve. In order to solve the optimal control problem we use Newton's
method which requires the gradient and the Hessian. The non trivial
approach here is the use of adjoint methods to obtain the expressions
for the gradient and the Hessian.
This work is joint with advisor Dr. Andrei Draganescu.
-
Friday, December 17, 2010, 01:35-01:45
Performance Characteristics of Linear Solvers Using ODE15S in MATLAB
Eric Katzaman, Department of Physics
It is commonly known that MATLAB contains linear solver
algorithms capable of solving partial differential equations using the
Method of Lines approach; however, best practices associated with
using these linear solvers are not as well known. We present
performance characteristics on a solved system of partial differential
equations with known analytical solution using the Jacobian option in
conjunction with the NDFk and BDFk methods in ode15s. A
convergence study using the lowest acceptable tolerance threshold is
also reported.
This work is joint with instructor Dr. Matthias Gobbert.
-
Friday, December 17, 2010, 01:50-02:00
Numerical Solution of an Elliptic PDE using Finite Differences on a
Staggered Grid
David Stanley, Department of Mathematics and Statistics
In this paper we will solve a 2-D elliptic PDE using the Finite
Differences method. However, this PDE will contain a spatially dependent
diffusivity coefficient inside the divergence term which we will take into
account by using a staggered grid of the domain. In other words, the
values of the solution will be positioned on the nodes of the grid while
the values of the diffusivity coefficient will be positioned in the
intervals. Since we have a Dirichlet boundary condition, only the
interior of the domain will be taken into consideration. A convergence
analysis will also be performed by solving the PDE with several refinments
of the mesh to determine if the rate of convergence agrees with the
theoretical value of quadratic convergence.
This work is joint with advisor Dr. Andrei Draganescu.
-
Friday, December 17, 2010, 02:05-02:15
Using Monte Carlo Method to Simulate Laser Propagation
in Multilayer Tumor
Yonghui Chen, Department of Mechanical Engineering
Monte Carlo method simulations of photon propagation offer a flexible
yet rigorous approach toward photon transport in turbid tissue.
This method simulates the 'random walk' of photons in a medium
that contains absorption and scattering. In my presentation
I developed a Matlab program to simulate laser absorption,
scattering, reflection and refraction in a multilayer tumor
embedded in mice tissue. The result show that the energy of laser
will mostly deposit in the target area as we expected.
This work is joint with advisors Dr. Ronghui Ma and Dr. Liang Zhu.
-
Friday, December 17, 2010, 02:20-02:30
Simulations of a Model for Calcium Waves in a Heart Cell
Using COMSOL Multiphysics
David W. Trott, Department of Mathematics and Statistics
Experiments have shown that calcium waves can occur from
spontaneously generated calcium sparks in a heart cell. A model for
this process is given by a system of reaction-diffusion equations.
We have in the past developed a special-purpose C code with MPI for
this model that successfully handles the high-resolution meshes of
the desired three-dimensional domains. We will explore the basic
framework of the model and investigate whether a general-purpose
package such as COMSOL Multiphysics can be used.
This work is joint with advisor Dr. Matthias Gobbert.
-
Monday, December 13, 2010, 04:05-04:15
ADI As Preconditioner for Krylov Subspace Methods
Kyle Stern, Department of Mathematics and Statistics
The alternating directions implicit (ADI) method is a classical
iterative method for numerically solving linear systems arising from
discretizations of partial differential equations. We propose to use ADI
as a preconditioner for Krylov subspace methods for linear systems arising
from the discretization of partial differential equations. This talk uses
a finite difference approximation of an elliptic test problem as test
problem for the computational experiments. The method is attractive
because it allows highly efficient matrix-free implementations both in
serial MATLAB and parallel C with MPI.
This work is joint with advisor Dr. Matthias Gobbert.
-
Wednesday, December 08, 2010, 04:05-04:15
Modeling Tissue Necrosis Dependent Perfusion Coefficient during Magnetic
Nanoparticle Hyperthermia
Anilchandra Attaluri, Department of Mechanical Engineering
Magnetic nanoparticle hyperthermia (MNH) is treatment procedure used to
treat cancer in which heat is used to physically ablate the tumor.
The Pennes bio-heat equation is solved using the finite element method
on a spherical domain in one spatial dimension (in an interval)
to obtain the temperature field. The goal of this work is to set up
the model which can determine the total tissue damage due to the heat
induced by MFH. The thermal damage is dependent on both the magnitude
of change in temperature and the duration of the elevated temperature.
From the energy point of view, the energy needed to elevate the tissue
to a certain temperature during a time duration is directly proportional
to both the temperature elevation and the treatment duration.
Once the temperature field is obtained from the FEM solution of
the bio-heat equation, an Arrhenius equation will be used to predict the
tissue damage.
This work is joint with advisor Dr. Liang Zhu.
Copyright © 2001-2010 by Matthias K. Gobbert. All Rights Reserved.
This page version 1.0, December 2010.