Fast Parallel Algorithm on GPU for High Order Solution of Eikonal and Transport Equation

Eikonal and Transport equation have wide range applications in the seismic wave simulation and inverse problem. In this paper, we use factorized methods to solve the source singularity problem, adopt the Lax-Friedrichs Hamiltonian to construct the iterative solution, and base on the WENO technique to get the high order approximation of the solution. For GPU parallel implementation, we adopt the Jacobi iteration method which is more suitable for parallel computing. we also presents the optimization methods of CUDA codes. Finally, give some examples to check the feasibility and efficiency of our methods.