Abstract:Nonconservative evolution problems describe irreversible processes and dissipative effects in a broad variety of phenomena. Such problems are often characterised by a conservative part, which can be modelled as a Hamiltonian term, and a nonconservative part, in the form of gradient flow dissipation. Traditional numerical approximations of this class of problem typically fail to retain the separation into conservative and nonconservative parts hence leading to unphysical solutions. In this work we propose a mixed variational method that gives a semi-discrete problem with the same geometric structure as the infinite-dimensional problem. As a consequence the conservation laws and the dissipative terms are retained. A priori convergence estimates on the solution are established. Numerical tests of the Korteweg-de Vries equation and of the two-dimensional Navier-Stokes equations on the torus and on the sphere are presented to corroborate the theoretical findings.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to maintain the separation of the conservative part and the dissipative part in the infinite - dimensional Hamiltonian system during the numerical discretization process, thereby ensuring the physical consistency of the solution. Specifically, the paper focuses on non - conservative evolution problems with gradient - flow dissipation. Such problems usually include a conservative part (which can be modeled by a Hamiltonian term) and a non - conservative part (in the form of gradient - flow dissipation). Traditional numerical methods are usually unable to preserve this separation, resulting in unphysical solutions. To solve this problem, the paper proposes a hybrid variational method, which can generate a semi - discrete problem with the same geometric structure as the infinite - dimensional problem. This ensures that the conservation laws and dissipative terms are preserved and establishes a priori convergence estimates. Through this method, the paper verifies the numerical tests of the Korteweg - de Vries equation and the two - dimensional Navier - Stokes equation on the torus and the sphere, further supporting the theoretical results. ### Specific Problem Description Consider the time - dependent partial differential equation, which is characterized by a conservative part (expressed through a Hamiltonian term) and a gradient - flow dissipative part. The problem is as follows: We look for \(u\) defined on the Hilbert space \(V\) such that: \[ \partial_t u = J(u)\delta H(u)+G(u)\delta S(u),\quad \text{in}\ \Omega,\ t > 0 \] where the initial condition is \(u(0) = u_0\in V\), along with appropriate boundary conditions. In this evolution equation (1.1), \(J(u)\) represents the conservative term and is a Poisson operator; \(G(u)\) represents the dissipative term and is a positive/negative semi - definite self - adjoint operator. The function \(H: V\rightarrow\mathbb{R}\) is the Hamiltonian, and \(S: V\rightarrow\mathbb{R}\) is the quantity of production/dissipation. ### Main Contributions of the Paper 1. **Maintaining Geometric Structure**: The proposed variational method can maintain the geometric structure in the infinite - dimensional problem, especially the separation of the conservative part and the dissipative part. 2. **A Priori Convergence Estimates**: It establishes a priori convergence estimates for the solutions of the semi - discrete problem. 3. **Numerical Verification**: Through the numerical tests of the Korteweg - de Vries equation and the two - dimensional Navier - Stokes equation, the validity of the theoretical results is verified. ### Key Formulas - Evolution Equation: \[ \partial_t u = J(u)\delta H(u)+G(u)\delta S(u) \] - Weak Form: \[ \langle\partial_t u, v\rangle = -\langle\delta H(u), J(u)v\rangle+\langle\delta S(u), G(u)v\rangle,\quad \forall v\in V \] Through these methods, the paper successfully solves the problem of how to maintain the separation of the conservative and dissipative parts in the Hamiltonian system during the numerical discretization process, ensuring the physical consistency of the numerical solution.

Conformal variational discretisation of infinite dimensional Hamiltonian systems with gradient flow dissipation

Variational multiscale modeling with discretely divergence-free subscales

Hamilton--Jacobi equations for controlled gradient flows: cylindrical test functions

High order Finite Difference/Discontinuous Galerkin schemes for the incompressible Navier-Stokes equations with implicit viscosity

Variational multiscale modeling with discretely divergence-free subscales: Non-divergence-conforming discretizations

Discontinuous Galerkin schemes for hyperbolic systems in non-conservative variables: quasi-conservative formulation with subcell finite volume corrections

Weighted Energy-Dissipation approach to semilinear gradient flows with state-dependent dissipation

High-order conservative and accurately dissipative numerical integrators via auxiliary variables

Stochastic Variational Principles for Dissipative Equations with Advected Quantities

A variational approach to the non-newtonian Navier-Stokes equations

A variational perspective on the dissipative Hamiltonian structure of the Vlasov-Fokker-Planck equation

Development of discontinuous Galerkin methods for hyperbolic systems that preserve a curl or a divergence constraint

Energetic Variational Neural Network Discretizations of Gradient Flows

A mixed variational discretization for non-isothermal compressible flow in pipelines

Dissipative Solutions and Semiflow Selection for the Complete Euler System

Finite element discretization of the steady, generalized Navier-Stokes equations with inhomogeneous Dirichlet boundary conditions

The dissipative Generalized Hydrodynamic equations and their numerical solution

High-order energy stable discrete variational derivative schemes for gradient flows

Discovering Artificial Viscosity Models for Discontinuous Galerkin Approximation of Conservation Laws using Physics-Informed Machine Learning

Symplectic and variational formulations of compressible and incompressible Navier-Stokes equation