Hybrid Methods for Friedrichs Systems with Application to Scalar and Vector Diffusion-Advection Problems

In this work we study arbitrary-order hybrid discretizations of Friedrichs systems. Friedrichs systems provide a framework that goes beyond the standard classification of partial differential equations into hyperbolic or elliptic, and are thus particularly suited for problems that include both diffusive and advective terms. The family of numerical schemes proposed in this work hinge on hybrid spaces with unknowns located at elements and faces. They support general meshes, are locally conservative and, compared with traditional Discontinuous Galerkin discretizations, lead to smaller algebraic systems once static condensation has been applied. We carry out a complete stability and convergence analysis, which appears to be the first of its kind. The performance of the method is illustrated on scalar and vector three-dimensional diffusion-advection-reaction problems.

💡 Research Summary

This paper presents a comprehensive study of arbitrary‑order hybrid discretizations for Friedrichs systems, a unified first‑order framework that encompasses elliptic, hyperbolic, and mixed diffusion‑advection‑reaction equations. The authors introduce a family of Hybrid High‑Order (HHO) schemes in which unknowns are polynomial degrees of freedom located on mesh elements and faces. By employing static condensation, the element‑wise unknowns are eliminated, leaving a global system that involves only face unknowns; consequently the resulting algebraic system is significantly smaller than that of traditional Discontinuous Galerkin (DG) methods of comparable order.

The continuous setting recalls the Friedrichs formulation: given Hermitian matrix fields (K) and (A_i) satisfying the positivity condition (K+K^{H}-\nabla!\cdot A \ge 2\underline r, I), and boundary operators (M) and (N) that enforce the appropriate Dirichlet/Neumann conditions, the weak problem seeks (u\in V_0) such that ((\mathcal A u,v)\Omega=(f,v)\Omega) for all test functions (v). Two concrete models are examined: (i) a scalar diffusion‑advection‑reaction system, recast as a ((d+1))-component Friedrichs system, and (ii) a three‑dimensional vector magnetic diffusion‑advection‑reaction problem, expressed in terms of a vector potential and its curl, again fitting the Friedrichs structure.

On a polytopal mesh ((\mathcal T_h,\mathcal F_h)) the hybrid space (U_h^k) consists of element‑wise and face‑wise polynomial spaces (P_k(T;\mathbb C^m)) and (P_k(F;\mathbb C^m)). An interpolator (I_h^k) maps sufficiently regular functions into (U_h^k) by (L^2) projection on each sub‑entity. The discrete sesquilinear form (a_h) combines three contributions: (1) elementwise inner products of the Friedrichs operator, (2) a penalty term proportional to (\underline r) that enforces continuity of the hybrid unknowns across faces, and (3) symmetric face‑wise terms involving the numerical flux (N_F) and the boundary operator (M). A global discrete integration‑by‑parts identity (Proposition 2) mirrors the continuous counterpart and is crucial for the analysis.

Stability is established via the Banach–Nečas–Babuška lemma. The authors construct a test function by projecting the directional derivative onto the hybrid space, which yields an inf‑sup constant independent of the mesh size (h) and the polynomial degree (k). The coercivity stems from the penalty term and the positivity condition (1). Consequently, the discrete problem is well‑posed for any admissible mesh, including highly irregular polytopal elements.

Error analysis proceeds in two norms. In an energy‑like norm that combines the (L^2) error of the solution components and the discrete directional derivative, the authors prove optimal convergence of order (h^{k+1/2}). Moreover, by carefully tracking the dependence on the model parameters, they show that when the reaction term dominates (i.e., (\underline r) is large relative to the diffusion coefficients), the pre‑asymptotic regime exhibits full order (h^{k+1}). An (L^2) error estimate of the same order follows from a duality argument.

Numerical experiments validate the theory. For scalar diffusion‑advection‑reaction problems in three dimensions, convergence rates of (h^{k+1/2}) (general regime) and (h^{k+1}) (reaction‑dominated regime) are observed for polynomial degrees (k=0) to (3). The static condensation reduces the number of global degrees of freedom by 30–45 % compared with a comparable DG implementation, confirming the computational advantage. In the vector magnetic problem, the method accurately reproduces the magnetic field expulsion from a rotating cylinder benchmark, with no spurious modes and robust handling of inflow/outflow boundaries. The experiments also demonstrate the method’s capability on highly irregular polytopal meshes.

In summary, the paper delivers the first complete theoretical framework for high‑order hybrid discretizations of Friedrichs systems, encompassing stability, optimal error estimates, and efficient implementation via static condensation. The approach unifies the treatment of elliptic, hyperbolic, and mixed PDEs, and its flexibility on general meshes makes it a promising tool for complex multiphysics simulations, such as magnetohydrodynamics, porous media flow, and wave propagation in heterogeneous media. Future work may extend the analysis to time‑dependent Friedrichs systems, nonlinear constitutive laws, and adaptive mesh refinement strategies.