Discrete Entropy Stability for Conservation Laws

30 concepts17 papers1 proposal

When simulating high-speed fluid flows—such as the air rushing over a re-entering space capsule or water crashing through a broken dam—numerical algorithms must navigate a treacherous mathematical landscape. The partial differential equations (PDEs) governing these systems, known as hyperbolic conservation laws, routinely develop sharp discontinuities called shocks.

In smooth regions of a flow, standard high-order numerical methods perform beautifully. However, when a shock forms, these standard schemes often suffer from severe unphysical oscillations or crash entirely. The root cause of this instability is a failure to respect basic physics at the numerical level: while standard schemes are designed to conserve mass, momentum, and energy, they often fail to satisfy the second law of thermodynamics—specifically, the continuous entropy inequality—in their discrete form. Without a rigorous guarantee that mathematical entropy behaves correctly across discontinuities, numerical solutions can become physically impossible, leading to code failure.

To solve this problem, a powerful framework has emerged: discrete entropy stability. By redesigning numerical schemes to strictly enforce a discrete entropy inequality, researchers can build high-order simulations that are mathematically guaranteed to remain stable and physically consistent, even in the presence of violent shocks and under-resolved turbulence.

Symmetrization via Entropy Variables

To understand why discrete entropy stability works, we must first look at how we represent the state of a fluid. Typically, fluid equations are solved using conserved variables (such as density, momentum, and total energy). However, this representation obscures the underlying thermodynamic structure of the equations.

The breakthrough intuition is to transform these equations by choosing a different set of primary degrees of freedom: entropy variables. These variables are derived mathematically as the gradient of a convex entropy function with respect to the conserved variables.

Formulating fluid dynamics equations using entropy variables transforms them into a symmetric hyperbolic system. As Praveen Chandrashekar demonstrated in Discontinuous Galerkin method for Navier–Stokes equations using kinetic flux vector splitting, this mathematical symmetry is a prerequisite for proving the thermodynamic stability of high-order numerical methods, such as Discontinuous Galerkin (DG) frameworks.

By utilizing entropy variables as the primary degrees of freedom rather than standard conserved variables, a numerical scheme can rigorously satisfy entropy inequalities. As Andreas Hiltebrand et al. showed in Entropy stability and well-balancedness of space-time DG for the shallow water equations with bottom topography, this representational choice ensures that the simulation remains physically consistent and stable by guaranteeing energy dissipation across shocks and complex flows, such as hydraulic jumps in shallow water.

Balancing Act: Entropy Conservation and Scaled Dissipation

How do we construct a numerical scheme that satisfies this discrete entropy inequality? The strategy relies on a precise two-part architecture:

  1. Entropy-Conservative Fluxes: In perfectly smooth regions of the flow, entropy should be conserved exactly. We design a baseline numerical flux that produces zero net entropy.
  2. Entropy-Scaled Dissipation: At a shock, physical entropy must increase (or mathematical entropy must decay). We add a targeted numerical diffusion operator to the conservative flux to act as artificial friction, suppressing spurious oscillations.
[Total Numerical Flux] = [Entropy-Conservative Flux] - [Entropy-Scaled Dissipation]
                                 │                              │
                    (Conserves entropy in smooth zones)   (Dissipates entropy at shocks)

The magic lies in how this dissipation is scaled. Rather than using arbitrary or manually tuned artificial viscosity—which can blur out fine physical details—the dissipation terms are scaled by entropy-consistent eigenvectors of the system's Jacobian matrix.

Diego Rojas et al. demonstrated the power of this approach in On the robustness and performance of entropy stable collocated discontinuous Galerkin methods. By combining two-point entropy-conservative fluxes with interface dissipation to mimic continuous physical entropy analysis, they achieved nonlinear stability in simulations of under-resolved turbulence and shocks without requiring manual tuning. Similarly, Harish Kumar et al. utilized this combination in Entropy Stable Numerical Schemes for Two-Fluid Plasma Equations to maintain physical consistency and prevent unphysical oscillations near plasma discontinuities.

This framework also allows researchers to resolve a classic trade-off in fluid dynamics: the tension between preserving kinetic energy in turbulent eddies and dissipating entropy at shocks. In Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations, Praveen Chandrashekar showed that entropy consistency conditions can be used to uniquely determine the free parameters in kinetic energy preserving (KEP) schemes. Integrating these two properties allows a scheme to resolve both high-Mach shock waves and fine-scale turbulent structures simultaneously, preventing the artificial dissipation of energy in smooth regions while ensuring physical consistency across discontinuities.

Handling Extreme Regimes and Source Terms

When simulations enter highly non-equilibrium regimes or involve complex multi-physics coupling, maintaining entropy stability requires additional safeguards.

In kinetic-based simulations like the Lattice Boltzmann Method (LBM), ensuring that the system obeys the H-theorem (the kinetic guarantee that entropy always increases) provides a firm thermodynamic foundation for stability. Iliya V. Karlin demonstrated in Invariance correction to Grad’s equations: where to go beyond approximations? that enforcing these entropic constraints allows for the stable modeling of turbulent and non-equilibrium microflows.

However, pure entropy preservation has mathematical limits. As Dave Packwood proved in Allowed and forbidden regimes of entropy balance in lattice Boltzmann collisions, entropy-preserving collisions become mathematically impossible in highly non-equilibrium states. In these extreme regimes, "positivity limiters" must step in. These limiters are not mere numerical corrections; they serve as essential thermodynamic components that introduce necessary entropy production to keep particle populations positive.

To handle these severe gradients without losing high-order accuracy, modern frameworks often blend different methods. For example, Huajun Zhu et al. developed an approach in Entropy Stable DGSEM Schemes of Gauss Points Based on Subcell Limiting that pairs a high-order entropy-stable scheme in smooth regions with a robust, low-order scheme in "troubled" subcells containing strong shocks.

Furthermore, when fluids couple with external fields—such as electromagnetic forces in relativistic plasma—the resulting source terms can threaten stability. Deepak Bhoriya et al. addressed this in High-order finite-difference entropy stable schemes for two-fluid relativistic plasma flow equations by proving that coupling source terms like the Lorentz force are entirely entropy-neutral. Because these terms do not contribute to the production or decay of entropy, the stability of the entire coupled system can be guaranteed simply by ensuring that the individual fluid and field components satisfy their respective discrete entropy inequalities.

Broader Impact and Generalizations

The utility of discrete entropy stability extends far beyond classical aerodynamics and hydraulics. Its mathematical principles are transforming alternative simulation paradigms, machine learning, and even statistical physics:

By anchoring numerical algorithms to the second law of thermodynamics, discrete entropy stability provides a rigorous, physics-first foundation for solving the most challenging non-linear conservation laws across science and engineering.

Key Foundational Work

Invariance correction to Grad’s equations: where to go beyond approximations? (2005) — contributed 2 concept elements to this direction.
Nonequilibrium entropy limiters in lattice Boltzmann methods (2007) — contributed 1 concept elements to this direction.
History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 17 papers
  2. Version 1 · 2026-10-04 · 26 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_entropy_variable_formulation_f7e2b09f,
  title = {entropy variable formulation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/entropy-variable-formulation/rdr_f7e2b09f4af64d50a3f7e60a7747e8d3},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 2, frozen text):

@misc{lacuna_entropy_variable_formulation_f7e2b09f_v2,
  title = {entropy variable formulation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/entropy-variable-formulation/rdr_f7e2b09f4af64d50a3f7e60a7747e8d3/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake