In ParaMathics, we develop novel systems, compilers, and mathematical methods that significantly improve the performance of large-scale and big data applications on parallel and cloud platforms. We leverage information from the domain and the problem structure to invent frameworks that transform the space of optimization that compilers and systems can explore. Our frameworks are designed to optimize a large class of algorithms and numerical kernels and thus support simulations of numerous applications such as in computer graphics, physics, vision, and machine learning. Learn more