Exploring Fast Matrix Multiplication Algorithms

My summer research project was to find a more efficient of matrix multiplication algorithm by locating a relatively big mathematical object called a uniquely solvable puzzle (USP). We implemented and iteratively improved algorithms for checking and searching for USPs using dynamic programming techniques in C/C++.  We also enabled our code to run on the Union College Jupiter cluster by using a high performance computing library.  By the end of the summer, we had found a USP with dimensions 6 by 14 that should yield a run time of O(n^2.44) for the matrix multiplication algorithm.