EPO: A Unified Framework for Entropy Stability, Positivity, and Oscillation Suppression

Abstract: High-order finite volume and discontinuous Galerkin methods are often stabilized by separate nonlinear devices for admissibility, entropy control, and oscillation suppression. This separation hides a simple geometric fact: all three act on the same cellwise candidate state. We propose a general framework (termed EPO) unifying fully discrete entropy stability, positivity/bound preservation, and spurious oscillation elimination. Starting from a candidate update, we scale along the ray anchored at its updated cell average. The admissible-state constraint, the entropy constraint, and the oscillation-suppressing constraint each define an admissibility radius on that ray, and the applied limiter is their minimum. The decisive analytical ingredient is a {\em weak entropy stability} at the level of the updated cell average. A two-point Lax--Friedrichs/Riemann-average entropy inequality yields local cell-average entropy budgets, and the same radial scaling mechanism behind Zhang--Shu positivity preservation lifts these weak budgets to strong quadrature-based entropy inequalities. The framework is therefore not a summation-by-parts, split-form, or flux-differencing construction: EPO acts on a candidate finite volume or discontinuous Galerkin update and converts weak average information into fully discrete nodal entropy stability. {\em The construction also works for any prescribed finite family of convex entropy pairs. Each pair yields its own entropy radius, and taking the minimum enforces fully discrete entropy stability for all of them simultaneously.} We prove the preservation of cell averages, invariant-set preservation, local and global strong entropy inequalities, stagewise budgets for strong-stability-preserving (SSP) Runge--Kutta methods, an SSP multistep variant that retains the designed high-order temporal accuracy, and extensions on rectangular and unstructured triangular meshes.

• The paper introduces a unified EPO framework that enforces entropy stability, positivity, and oscillation suppression using a single cellwise scaling limiter.
• The methodology extends the Zhang–Shu invariant domain limiter to achieve weak-to-strong closure for both cell-average and nodal entropy inequalities.
• The framework is locally conservative and compatible with high-order FV/DG schemes and SSP timestepping, ensuring robust performance even on unstructured meshes.

EPO: A Unified Local Limiter for Entropy Stability, Positivity, and Oscillation Suppression

Introduction and Context

The "EPO: A Unified Framework for Entropy Stability, Positivity, and Oscillation Suppression" (2604.00301) presents a comprehensive approach to enforcing three central types of nonlinear stability in high-order finite volume (FV) and discontinuous Galerkin (DG) methods for hyperbolic conservation laws: admissibility/positivity (invariant domain preservation), entropy stability, and spurious oscillation suppression. Traditionally, these constraints are handled by distinct, often incompatible, limiting devices. EPO (Entropy-Positivity-Oscillation) proposes a rigorous, local, and geometric unification, deploying a single cellwise scaling limiter operating on the same convex ray anchored at the updated cell average. This construction draws on and extends the foundational Zhang–Shu cell-average-based scaling employed in invariant domain preservation to also incorporate entropy stability and oscillation suppression, in a manner that is entirely representation-agnostic with respect to FV and DG discretizations.

Technical Construction of the EPO Framework

Given a candidate update—polynomial, modal/nodal array, or reconstructed profile—with cell average $\bar{\mathbf{U}}_j^{\star}$ and high-order representation $\mathbf{U}_j^{\star}$, EPO considers the affine segment (scaling ray) $$\mathcal{S}_{\bar{\mathbf U}_j^{\star}}(\theta; \mathbf{U}_j^{\star}) = \bar{\mathbf U}_j^{\star} + \theta (\mathbf{U}_j^{\star} - \bar{\mathbf U}_j^{\star})$$, for $0 \leq \theta \leq 1$, which preserves the conservative cell average by construction.

Three independent constraints—admissibility, entropy, oscillation—each define within $[0,1]$ a maximal admissible scalar radius along this ray. Their intersection gives the final limiting coefficient: $$\theta_{j}^{\mathrm{EPO}} = \min\bigl\{\theta_{j}^{\mathrm{PE}}, \theta_{j}^{\mathrm{O}}\bigr\}$$ where:

• $\theta_j^{\mathrm{P}}$: maximal scaling such that all nodal points remain in the required physical/invariant domain,
• $\theta_j^{\mathrm{PE}}$: maximal (positivity-first) scaling such that the strong quadrature-based entropy inequality is attained (for one or multiple prescribed convex entropy pairs),
• $\theta_j^{\mathrm{O}}$: maximal scaling such that a convex oscillation suppressing (COS) set is respected.

A key analytical ingredient is weak entropy stability at the cell-average level for the candidate update. Using a local two-point Lax–Friedrichs/Riemann-average entropy inequality, a cell-averaged entropy budget (inequality) is obtained. The same convex scaling machinery then "lifts" this weak budget to enforce a strong (nodal/quadrature-based) entropy inequality, analogous to the positivity machinery but applied to the entropy functional.

For the oscillation module, EPO directly integrates the COS framework developed in [COS(DG); JCP 2026] as a convex cellwise set, with geometric, scale-invariant sensor coefficients.

Main Analytical Results and Properties

• Weak-to-Strong Closure: EPO establishes that weak stability properties (at the cell-average level) for admissibility and entropy can be elevated to strong, nodal/quadrature-based control simply via radial convex limiting. This holds for both FV and DG methods.
• Multi-entropy-pair Enforcement: Unlike flux-differencing or split-form entropy stable schemes that are typically associated with a single convex entropy pair, EPO can enforce strong, fully discrete entropy stability for any prescribed finite family of convex entropy pairs simultaneously—each yielding a separate radius, with the final limiter their minimum.
• Locality and Conservativity: The construction is local (cellwise, not relying on global coupling), conservative (cell-average is preserved), and algebraically simple for practical implementation.
• Compatibility with SSP Timestepping: The framework is compatible with both explicit SSP Runge–Kutta and SSP multistep time discretizations. For SSPRK, stagewise strong entropy stability budgets can be maintained. For multistep methods, the entropy module can be applied once per step, retaining designed temporal accuracy.
• Extension to 2D and Unstructured Meshes: While the core EPO geometry is one-dimensional, the analytical structure extends to 2D rectangular and triangular meshes using decomposition strategies based on special quadratures and cell-average decomposition.

Numerical and Theoretical Implications

• Guaranteed Fully Discrete Entropy Stability: For any convex entropy functional(s), EPO yields strong, quadrature-based entropy inequalities at each time step, locally and globally (up to telescoping boundary fluxes).
• Positivity and Invariant Domain Preservation: The scaling ensures all nodal points stay within the user-prescribed admissible set, regardless of mesh, degree, or spacetime representation.
• Spurious Oscillations Suppression: The oscillation module is non-intrusive, non-dissipative away from shocks, robustly controls post-shock and contact oscillations, and is provably non-expansive in $L^2$ and entropy metrics.
• Simultaneous, Unified Limiting: By co-limiting all three constraints via a single convex projection, interactions and incompatibilities across separate modules are avoided. This unification also simplifies both theory and implementation.
• Flexibility and Compatibility: The EPO construction is compatible with high-order base schemes that are already entropy stable; in this case, the entropy limiter degenerates to a no-op, and only positivity and oscillation limiters are activated as needed.

Quantitative and Contradictory Claims

• Strong-Order Accuracy Maintained: For smooth solutions, strong-order accuracy is systematically retained even with full EPO limiting.
• Entropy Stability for Arbitrary Convex Family: The ability to enforce strong entropy stability for any finite collection of convex entropy-entropy flux pairs is unprecedented among locally postprocessed limiters.
• Repudiation of Necessity for SBP/Split-Form Constructions: EPO demonstrates that entropy stability does not require summation-by-parts or flux-differencing structures, contradicting a prevailing assumption that such algebraic forms are necessary for entropy control in high-order discretizations.

Numerical Validation

Comprehensive benchmarks on compressible Euler equations demonstrate sharp shock-capturing, positivity preservation under severe underresolved conditions (Sedov, Leblanc), monotonic decay of the global discrete entropy, and the absence of nonphysical oscillations—consistently matching or surpassing existing state-of-the-art positivity and entropy frameworks in robustness and accuracy. Third-order convergence is observed on smooth test cases. Numerically, the entropy limiter is found to be non-activating on smooth flows, while playing a critical role in damping post-shock oscillations and preventing unphysical undershoots in highly nonlinear regimes.

Theoretical and Practical Impact; Directions for Future Work

• Bridging Device-Centric and Constraint-Centric Perspectives: EPO reorients the construction of nonlinear stabilizations toward constraint-centric, geometric optimization rather than separate device-centric modules, promising integrability, modularity, and extensibility.
• Applicability to Advanced Physics and Coupled Multiphysics: The algebraic locality and representation-agnostic construction facilitate generalization to balance laws, nonconservative systems, and multiphysics coupling, including MHD and chemically reactive flows.
• Pathway to Certified High-Order Entropy Stable Integrators: The framework opens avenues for fully discrete, high-order accurate, and entropy-stable time integrators for complex problems on generic meshes.
• Potential for Automated, Robust Limiters: The modularity and convexity-driven logic support adaptation to automated limiter selection and tuning in large-scale multiphysics codes.

Conclusion

"EPO: A Unified Framework for Entropy Stability, Positivity, and Oscillation Suppression" (2604.00301) provides a mathematically rigorous, algorithmically simple, and computationally effective unification of core nonlinear stability constraints in high-order FV/DG methods for hyperbolic conservation laws. Its weak-to-strong closure, compatibility with multi-entropy pairs, and seamless extensibility to multi-dimensional and unstructured settings establish it as a robust foundation for the next generation of structure-preserving high-order numerical schemes.

Reference:

EPO: A Unified Framework for Entropy Stability, Positivity, and Oscillation Suppression (2604.00301)