1: \begin{abstract}

2: We present a scalable and efficient iterative solver for high-order

3: hybridized discontinuous Galerkin (HDG) discretizations of hyperbolic

4: partial differential equations. It is an interplay between domain

5: decomposition methods and HDG discretizations. In particular, the method is a fixed-point

6: approach that requires only independent element-by-element local

7: solves in each iteration. As such, it is well-suited for current and

8: future computing systems with massive concurrencies. We rigorously

9: show that the proposed method is exponentially convergent in the number

10: of iterations for transport and linearized shallow water

11: equations. Furthermore, the convergence is independent of the solution

12: order. Various 2D and 3D numerical results for steady and

13: time-dependent problems are presented to verify our theoretical

14: findings.

15: 

16: %%   The solver has an

17: %% exponential convergence rate and the iterations are independent of the

18: %% polynomial order considered. Moreover, during each iteration the

19: %% calculations are independent for each element and requires only

20: %% communication between the neighbours. Hence the solver is naturally

21: %% suitable for large scale parallel simulations of hyperbolic PDE's

22: %% using higher order discontinuous Galerkin methods. In the analysis and

23: %% numerical simulations two kinds of HDG fluxes are considered and the

24: %% performance of the solver with respect to both of them is

25: %% studied. Numerical simulations supporting the theoretical results are

26: %% given for 2D and 3D steady state/time dependent transport equations

27: %% and linearized shallow water equations.

28: 

29: \end{abstract}

30: