Computer Science and Software Engineering Capstone Presentations
Fall Quarter
December 18, 2020
|
Daria Voblikova "Quasar
Spectra Analysis" (UWB School of STEM Physics & CSS) Faculty Advisor: Dr. Arkady
Retik |
|
Abstract The main goal of this capstone project was to help
Astrophysics researchers improve the normalization algorithm of quasar
spectra. The algorithm is a part of previous and future research of
Extremely-High-Velocity Outflows in Quasars. The project is an initial step
towards an Astrophysical Computational Center at UW Bothell. The CSS and
Physics teams work together to create an infrastructure for data analysis and
collaboration in Astrophysics domain. The first step in improving the algorithm was to
understand what the research was meant for and the initial implementation of
the algorithm. It took a bit of effort and meetings with the Astronomy
Professor to understand what the algorithm had to accomplish. Once I had a better understanding of the physics
part of the project, I first created a test to improve the development
process. Then I could improve the algorithm based on the Physics researchers'
requirements to produce more accurate results. Finally, a big part of the
project was refactoring the algorithm so it's clear
for future students and researchers and open for extensibility and easily
adaptable to their needs. Because the final goal is a Computational Center,
the refactoring of the algorithm was an important part of the project. Also, during this collaborative research project, I
provided Git and Python mentorship for teammates, including small tutorials,
helping with merge conflicts, and guiding on the refactoring and debugging
process. I also met virtually with my teammates and Physics Professor to
discuss and collaborate on the project. The final algorithm provided a 97%
accuracy rate against the previous 90%. This rate is crucial for future work
as it should reduce the number of cases researchers would have to manually
inspect from ≈1400 to ≈420. The algorithm was developed and
tested with Python, with the use of scientific libraries such as NumPy,
SciPy, Matplotlib, etc. It was reviewed by the Physics Advisor of the
project, Professor Rodriguez Hidalgo, who gave it positive feedback and
approval. |
|
|
Updated December 15, 2020