I will return to NASA/Goddard Space Flight Center this summer to continue working
in the Laboratory for Extraterrestrial Physics with the NEAR Earth Asteroid Rendezvous Mission
science team. I also designed, developed and maintain our website. The NEAR Mission was the
first NASA Discovery Mission, as well as the first to have a satellite orbit an asteroid,
A433Eros in particular. The satellite entered rendezvous on February 14, 2000, and will continue
to orbit the asteroid until February 14, 2001. There are six data gathering instruments onboard
the satellite, and all are functioning properly.
 |
 |
I work directly with the X-Ray/Gamma-Ray
Spectroscopy team, and Tim McClanahan is my supervisor. I began working for him as an intern
during the summer of 1998 and returned again funded by Catholic University of America, as an
independent contractor during the later part
of the summer of 1999. Under his guidance, in addition to an affinity for software engineering,
I was able to develop a data visualization system
that is assisting both the scientific analysis and documentation endeavors of this mission.
I actually engineered four
software applications for this mission. They are, the
NEAR-DVS, NEAR-CVI, NEAR-VCI, and NEAR-CRS, which were designed as follows:
The NEAR and yet so far Data Visualization System (NEAR-DVS)
 |
. . . is a visual data-mapping tool that automatically generates an interactive
three-dimensional rendition of an asteroid, A433Eros in particular,
in a manner such that it can be oriented at will in the virtual xyz-coordinate
system of the computer. It reads in a number of input files and outputs
NEAR-DVS.wrl .
This application was named by Dr. Arunas Dagys, Professor of Mathematics and Computer
Science, Saint Xavier Univeristy. |
The NEAR.miss C to VRML Interface (NEAR-CVI)
|
. . . is also a visual data-mapping tool that automatically generates an interactive
two or three-dimensional color-coded rendition of any object in a manner such that the
it can be oriented at will in the virtual xy-coordinate or xyz-coordinate system of the
computer, depending upon the dimension of your model. It reads in a number of input
files and outputs triangulated_plate_model.wrl .
|
|
|
This application was named by Maryanne Terpstra, PhD. Candidate, Marquette University.
|
The NEAR bullseye VRML Color Index (NEAR-VCI)
 |
. . . is a two-dimensional color-coded pie chart of IDL color definitions, that allows the
scientists to easily choose color schemes for their data models. Each distinct slice
of the pie chart contains one of the available colors for the scheme under consideration.
It reads in color_ratios.in and outputs near_bullseye.wrl . |
|
This application was named by Dr. Susan Beal, Professor of Mathematics and Computer
Science, Saint Xavier Univeristy. |
The NEAR and dear to my heart Color Ratio Synthesizer (NEAR-CRS)
|
This program reads in, from a formatted input file named
integer_defined.color, an integer value into the variable n, which
specifies the number of distinct color definitions in the file, and
then it reads in n sets of three element IDL integer color definitions.
It then converts those n integer definitions into their corresponding
three element VRML ratio color definitions, and stores them in the
output file named color_ratios.in . The color_ratios.in
output file can be called by a variation of the NEAR C to VRML
Interface program, NEAR-CVIwc.c (wc stands for with color), if you so
choose to add one distinct color to each distinct polygonal face of
your triangulated plate model.
|
|
 |
This application was named by Cindy Marron, may she rest in peace, former Office
Manager for the Department of Mathematics and Computer Science at Saint Xavier
University.
|
During the summer of 1999, I worked as an intern for the first time at Argonne National
Laboratory in the Mathematics and Computer Science Division. I was assigned to the
Computational Differentiation ADIC project, which stands for Automatic
Differentiation of C. 
Automatic Differentiation (AD)
programming techniques are designed to generate, the algorithmic
derivative(s) of a function that has been coded into a computer
program. These derivative programs can be invoked to generate derivative values that
are both accurate to machine precision, and of the same order of accuracy as that
of their original. 
My research on Optimizing the Computational
Differentiation Mehtod enabled me to conclude that at that present, automatic differentiation
was less cost efficient than finite differentiation. Using both differentiation
methods, two operating system platforms (the Sun Sparcstation 5 and IBM RS/6000), and varying
the design variables from 1 to 7, the runtime performance of CSCMDO' vs CSCMDO.AD was obtained (see Table 1).
Of these methods, the first employed was a best-case one-sided
finite difference approximation (Best-case FD), of volume with respect to geometric design
variables, the second "CSCMDO.AD(1)", was an invocation of the sensitivity-enhanced stand-alone program obtained
via automatic differentiation that included a quality validation of the generated grid
(though the time to generate it is unaccounted for), and the third method
"CSCMDO.AD(2)", was an invocation of the AD program (generation time also
unaccounted for), excluding the validation step of the second method.
Comparison of Runtime Performance: CSCMDO' vs CSCMDO.AD
Volume with respect to Design Variable Quantities
| Number of Design Variables |
|
| Platform | Differentiation Method |
1 | 2 | 3 | 4 | 5 | 6 | 7 |
| IBM | Best-Case FD | 39 | 59 | 79 | 99 | 118 | 138 | 158 |
| IBM | CSCMDO.AD {1} | 130 | 174 | 223 | 281 | 324 | 526 | 560 |
| IBM | CSCMDO.AD {2} | 69 | 84 | 103 | 119 | 144 | 192 | 212 |
| SUN | Best-Case FD | 57 | 86 | 115 | 144 | 172 | 201 | 230 |
| SUN | CSCMDO.AD {1} | 149 | 205 | 278 | 351 | 465 | 546 | 637 |
| SUN | CSCMDO.AD {2} | 83 | 104 | 131 | 160 | 184 | 213 | 242 |
|
| Table 1.
The timing results, presented here in user seconds, were
obtained with version 2.5.8 of the GNU C compiler with the
-02 -ffast-math compiler flags, and clearly indicate that
the finite difference method out-performs automatic
differentiation, and that the IBM clearly out-performs
the SUN on a method-for-method basis. It is important to
note that the finite difference estimates, though less
costly, are numerically unpredictable due to their
built-in instability, and that the automatic differentiation
estimates, are accurate to machine precision.
|
Efforts (which will probably need to
incorporate parallel processing techniques), can possibly be made to increase the cost-efficiency
of automatic differentiation routines. As such, in the future, it might be possible for automatic
differentiation techniques to attain a first-place cost-efficiency position in the field of
computational differentiation, and therefore to be implemented as the method of choice in ongoing
efforts to render cost-efficient and accurate-to-machine-precision solutions for non-linear
differential and partial differential equation models.
 |
|
|
During the summer 1997, I was hand-picked by Laureen Summers to participate in the Achieving Competency in Computing,
Engineering, and Space Science (ACCESS) internship program which was sponsored jointly by
the American Association for the Advancement of Science (AAAS) and the National Aeronautics and
Space Administration (NASA). This project was part of President Clinton's larger Project on
Science and Disability, and I have dyslexia.
I worked for the first time at Goddard Space Flight
Center, in the Laboratory for Astronomy and Solar Physics with Dr. Randall K. Smith, who was my
mentor. I performed research work in the area of Thermal Dynamics, and using the concept of finite
differentiation, wrote a computer program in IDL which simulated thermal conduction as a
function of both emmisivity and radiation (as opposed to one or the other). Randy Smith and myself
were interviewed for the article, "Aptitude and Attitude Outweigh Disabilities", that appeared in
the July 1997 issue of Science Magazine.
 |
|
I presented this material three times in oral form,
first at Goddard as a ten minute research dissertation pertaining to the role that classical
thermal conduction might play in the evolution of supernova remnants within the interstellar
medium. The second was a forty minute presentation to the faculty and students of the SXU
Mathematics and Computer Science Department, which was a necessary requirement (in addition to a
research paper), for the completion of Senior Seminar Mathematics. The third such presentation of
my research project was completed at the Eighth Annual Symposium for Undergraduates in Science,
Engineering and Mathematics at Argonne National Laboratory during November of 1997.
The title of
my presentation and paper was "Testing Numerical Instabilities in Numerical Hydrocodes", and was
based upon an analysis of simulated data that was generated with my Computerized Mathematical
Model. The model was engineered to simulate the evolution and cooling of a supernova remnant
within a time dependent one-dimensional magneto hydrodynamically expanding shock wave, relative
to the time dependent distance of the substance's propgation front from its propagation origin.
Both Tim McClanahan and Randy Smith have had a tremendous influence upon me, as did Dr. Arunas
Dagys who was my Mathematics Senior Seminar Advisor, and Dr. Kathleen Alaimo who was my History
Senior Seminar Advisor. Working with these four has proven to be an invaluable experience,
hopefully, the effects of which will remain with me for some time to come. At some point in time,
I would like to return the experiences that they have shared with me to other students, perhaps
in the capacity of an advisor or mentor. My long-term goal is to become a United States Astronaut,
and this has been my dream since I was a small child.
I would also like to continue doing
research for NASA/Goddard and/or Argonne National Laboratory. Also, it would be nice to be able
to get a paper or two published and to obtain a patent for my software.
As you can see, I like to keep myself busy. I almost never watch television
(though I do watch the X-Files and CNN regularly).
I prefer to make my own headlines, if you know what I mean.
This page was last updated 4-28-2000.
