|
I often require students to complete
projects in my math courses. There are many benefits to doing
this, including:
- Introducing students to the idea of performing research
on a mathematical topic.
- Improving students' technical communication skills,
including technical document preparation, and composing and
delivering presentations.
- Alerting students to some of the many applications of
mathematics in the real world.
- In some courses, getting students to work together as
part of a group.
I try to use student projects as a teaching opportunity on
multiple levels. I usually announce the project and outline
the requirements early in the semester, often on the first day
of class. Usually the student(s) are allowed to choose the
topic of their project, with my approval and assistance if
needed. I then establish milestones for the students to
accomplish throughout the semester, giving them short-term
tasks while keeping their attention focused on the long-term
goal of completing the project. An example of this is to
require the students to give me a short summary of their
progress about midway through the semester. If a written
report is to be turned in as part of the project, I will often
require the students to give me a draft to review and provide
comments and suggestions. I thoroughly enjoy working with
students on projects and am usually impressed with how well the
students do.
Some examples of courses and
projects:
Numerical Methods (Fall
2007): In this junior-level course, students were
to form groups of at most 5 persons and complete a project that
discussed how numerical methods are used for research on a
problem they find interesting (see
full project description). The groups were to produce a
technical document describing the problem and to give a
presentation on the problem to the rest of class. Selected
projects are listed below:
- Bridge
Flutter Analysis by D. Frank, D. Kammer, and F.
Ding.
This project is concerned with approximating flutter in
cable-stayed bridges. It focuses on two numerical methods
to approximate the critical wind velocity needed to create
the flutter phenomenon in a bridge. It discusses the
methods in detail, and then gives an example that shows the
accuracy of the two methods.
- Applications
of Monte Carlo Methods in Portfolio Analysis by M.
Albrecht, J. Combes-Knoke, and M. Seminatore.
Monte Carlo numerical methods are used extensively in areas
such as pricing American derivative securities, they are
also incorporated to calculate the efficient frontier of an
investment portfolio as well as providing insight into
investment portfolios. Other usages include the calculation
of Value at Risk (VaR) as a relevant risk measure and net
present value of projects.
- Satellite
Launch and Alignment into Orbit by R. Chatterjee, V.
Devaraj, F. Pineda, and J. Shoemaker.
This project focuses on the two major aspects of space
exploration: satellite launch and satellite placement into
orbit. The delicate nature of these calculations
necessitates numerical methods and computer calculations.
The margin of error is very small, so slight errors in
cancelation, among other potential pitfalls, could derail
the entire launch process. The same factors apply to
placing a satellite into proper orbit. Ultimately, accuracy
in both launch and alignment are crucial to satellite
implementation.
- Centroidal
Voronoi Tessellations by E. Allen, Z. Girouard, P.
Goswami, and B. Leary
A centroidal Voronoi tessellation (CVT) is a partitioning
of points based on minimizing distances relative to centers
of mass. This paper will study three numerical methods
involved in computing CVT’s. The first goal is to
discuss the different applications of CVT’s, ranging
from animal territories to optimal placement of
resources. The second goal is to analyze our methods of
generating CVT’s. Lloyd’s method is a
deterministic method while MacQueen’s method is
probabilistic. We want to look at when they converge, how
quickly they converge, and pros and cons of each. We will
talk about how these numerical methods are utilized in
finding CVT’s. Finally, we will look at the
effectiveness of using a non-Euclidean method to compute
anisotropic CVT’s.
- Numerical
Modeling of a Parasite-Symbiont-Host Model by N. Komarov and M.
Nemaric
This project presents a new population model to describe
the interactions of a host, parasite, and symbiont. From
the equilibrium criteria, we derive restrictions on the
constants in the model, and discuss certain aspects of the
resulting equilibria. We then model this interaction
numerically, using the forward Euler, or tangent line
method, and use these approximations to discuss the
implications of some of the equilibria. Finally, we discuss
some possible expansions of this model.
- Numerical
Weather Prediction by R. Barrett and E. Tucker
This project has several focuses with the intent of
surveying the development of numerical weather prediction
over the past century. First, it examines the early days of
weather prediction and the work of Lewis Fry Richardson,
one of the pioneers in the field. In this segment of the
report, some of the first techniques are studied and the
reasons for their failure are explained. After this, we
move on to look at the role supercomputers played in the
continuing advancement of this field. Two numerical methods
techniques for solving relevant equations, the Finite
Difference Method and the Spectral Method, are also
explained. In the final portion of this report, a more
in-depth look at one of the current and most advanced
numerical weather prediction systems is provided. The
HIRLAM technique employs these two methods to divide
weather forecasting into physics and dynamics.
- Financial
Simulations and Variance Reduction by A. Dabholkar,
V. Hidayat, F. Ho, D. Levin, and J. Qi
Valuation of American options is one of the more difficult
problems in option-pricing theory. American options are
heavily used in many financial markets, but pricing remains
challenging. Unlike European options, which can only be
exercised at expiration, American options can be exercised
at any point during the lifetime of the option. There exist
explicit formulas to price European options (e.g.
Black-Scholes-Merton), but no such formulas exist for
American options due to their more complex nature. Instead
numerical methods are generally used.
- Spectral
Methods for Burger's Equation by J. Rothenberg and
G. Peim
Burger's Equation can be solved numerically using several
different spectral methods. These methods are efficient to
calculate approximate solutions to
Burger's Equation. Burger's equation is a good
simplification of Navier-Stokes Equation when the velocity
is in one spacial dimension and the net force acts
in one direction. Burger's Equation is used to analyze
traffic congestion and acoustics. We discuss the background
of the equation, analytical solution, and
the Galerkin and Collocation spectral approximations.
- Artificial
Neural Network Machine Learning Through Numerical
Optimization by J. Dunn and X. Li
An antificial neural network (ANN) consists of a group of
interconnected simple processing elements, known as
artificial neurons, which provide a robust approach to
approximating vector-valued functions. For many types of
machine learning applications, such as handwriting
recognition and facial recognition, ANNs are among the most
accurate learning methods in existence.
- B-Spline
Interpolation and Its Scientific Applications by A.
Bakelmun, A. Charnas, G. Domville, and R. Lazrus
B-spline interpolation is a method commonly used to
approximate three-dimensional curves in space. The method
estimates the curve by substituting finite, known, data
points into approximate equations. The accuracy to which
the interpolation can estimate a curve varies greatly
depending on the field of study, the type of curves to be
approximated, and the equations and data points chosen for
the calculations. B-spline interpolation is applied to
various fields of study such as medicine, meteorology, and
chemistry.
Mathematical Studies II (Spring
2009): In this lecture course, designed for
advanced sophmores, the topics covered include vector spaces
and some linear analysis. At the end of the semester, students
(in groups of one or two) gave a brief presentation that
summarized key ideas and results that we covered during the
semester. Topics presented included:
- Infinite vs. Finite Dimensional Vector Spaces
- Approximating Solutions of Linear Systems of
Equations
- Eigenvalues and Eigenvectors
- Self-Adjointness and Positivity
- Bilinear and Alternating Forms
- Orthogonality
- Canonical Forms
- Classical Inequalities and Their Proofs
Finite Element Methods (Fall
2009): In this graduate course, students will
apply the finite element method to a problem that they find
interesting. The application of the finite element method to
their problem will usually require working through existence,
unidueness, and error estimate proof, as well as adapting
existing finite element code to the problem. They will write a
technical report on their project (in the style of a research
journal article) and give a presentation to the rest of class.
This project offers the potential to produce articles that can
be published in student research journals, and the small class
size allows for the student and instructor to work closely
together on the project. Project milestones include (a)
deciding on a topic, (b) writing a brief description of the
objectives of the project, (c) composing a draft technical
document, and (d) improving the draft, conducting simulations
of a problem related to the project, and presenting the
results.
Contact Info:
Jason S Howell;
Department of Mathematics; College of Charleston; 66 George Street; Charleston, SC 29424; 843-953-1016 (office); 843-953-1410 (fax);
email: howelljs at cofc dot edu
Copyright © 2012 Jason
Howell,
please send me any questions or comments.
|
|
|