GPU-accelerated path tracker for polyhedral homotopy

Abstract

The polyhedral homotopy method of Huber and Sturmfels is a particularly efficient and robust numerical method for solving system of (Laurent) polynomial equations. A central component in an implementation of this method is an efficient and scalable path tracker. While the implementation issues in a scalable path tracker for computer clusters or multi-core CPUs have been solved thoroughly, designing good GPU-based implementations is still an active research topic. This paper addresses the core issue of efficiently evaluate a multivariate system of Laurent polynomials together with all its partial derivatives. We propose a simple approach that maps particularly well onto the parallel computing architectures of modern GPUs. As a by-product, we also simplify and accelerate the path tracker by consolidating the computation of Euler and Newton directions.