Speeding up a few orders of magnitude the Jacobi method: high order Chebyshev-Jacobi over GPUs

This work provides a practical acceleration technique for solving elliptic PDEs, which is relevant for computational scientists and engineers using finite difference methods.

The authors achieve a remarkable speedup in solving elliptic PDEs with finite differences by combining the Chebyshev-Jacobi method with high-order discretizations and GPU parallelization, resulting in orders of magnitude faster convergence.