Sparse Solver Benchmark:

An Undergraduate Research Project

This project was a 10-week long process, where I had the opportunity to work alongside a professor at UC Davis. The research involved his ElasticRods program, which utilized the “SuiteSparse” library to factor and solve sparse matrices. Although this library made the program functional, there were lingering questions as to if it was the most efficient library to use. In my work, I managed to extend the ElasticRods program to function using another library called “catamari,” create a benchmark test to compare factor and solve times between the two libraries and create a script to run the test on several matrices with different variables, while organizing all the collected data. In summary, we found “catamari” ran the factorization and refactorization much faster and scaled in speed when using multiple threads, whereas “SuiteSparse” was unable to do this. This project provided me a valuable learning experience at UC Davis, as I could learn and enhance skills with C++ and Python. Additionally, I learned how to increase efficiency in my personal workspace to program faster and easier. My time performing undergraduate research was an amazing opportunity for me to enhance my skills and develop something with real-world applications.