Restoring Convergence Order in Explicit Runge-Kutta Integration of Hyperbolic PDE with Time-Dependent Boundary Conditions

Alfredo Pinelli; Giorgio Maria Cavallazzi; Miguel P\'erez Cuadrado

arxiv: 2604.09183 · v1 · submitted 2026-04-10 · 🧮 math.NA · cs.NA· physics.comp-ph

Restoring Convergence Order in Explicit Runge-Kutta Integration of Hyperbolic PDE with Time-Dependent Boundary Conditions

Giorgio Maria Cavallazzi , Miguel P\'erez Cuadrado , Alfredo Pinelli This is my paper

Pith reviewed 2026-05-10 16:47 UTC · model grok-4.3

classification 🧮 math.NA cs.NAphysics.comp-ph

keywords explicit Runge-Kuttahyperbolic PDEorder reductionboundary conditionsfinite differencetruncation errorSSP-RK3convergence order

0 comments

The pith

Redesigning the first two boundary derivative operators restores the nominal convergence order of explicit Runge-Kutta methods for hyperbolic PDEs.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper aims to show that order reduction in explicit Runge-Kutta integrations of hyperbolic initial-boundary value problems arises from the interaction between the time-stepping stages and conventional spatial closures near time-dependent boundaries. By deriving a decomposition of the truncation error that depends on the Runge-Kutta coefficients, it obtains algebraic conditions the boundary weights must satisfy to cancel the problematic terms. A coefficient R(b,c,A) indicates when such cancellation is possible, and for the SSP-RK3 method this leads to explicit formulas that are solved numerically to produce new five-point stencils. When these replace the standard Taylor expansions at the boundary while keeping a high-order interior scheme, the full third-order accuracy is recovered on linear and nonlinear test problems.

Core claim

For an arbitrary explicit s-stage RK method applied to linear advection, the one-step truncation error at the boundary-adjacent nodes is shown to admit a tableau-dependent decomposition whose cancellation yields explicit algebraic conditions on the boundary weights. A solvability coefficient R(b,c,A) determines whether a spatial compensation mechanism exists; the result is specialised to SSP-RK3, for which closed-form conditions are derived. Constrained differential evolution then identifies 5-point closures that, coupled to a 5th-order upwind interior stencil, recover third-order convergence from the degraded second-order behaviour of classical Taylor closures.

What carries the argument

A tableau-dependent decomposition of the boundary truncation error together with the solvability coefficient R(b, c, A) that governs whether compensating boundary weights can be found.

If this is right

Explicit algebraic conditions on boundary weights for any s-stage explicit RK method enable design of compensating stencils.
For SSP-RK3 the conditions are closed-form and allow optimization of five-point boundary closures that achieve third-order convergence.
The recovered order holds for linear advection, manufactured Burgers solutions, and split two-dimensional advection.
Incorporating an eigenvalue penalty in the stencil optimization balances accuracy recovery against stability constraints.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The purely spatial approach could be combined with other time integrators or applied to variable meshes as suggested by the analysis.
It highlights the limitations of purely temporal modifications like weak stage order for addressing spatial boundary mismatches in finite differences.
Similar compensation might be derived for other spatial discretizations such as finite elements near boundaries.

Load-bearing premise

The analysis performed for the linear advection equation on uniform meshes applies to the nonlinear and two-dimensional problems examined in the numerical tests.

What would settle it

If the convergence order observed when using the optimized boundary stencils on the linear advection problem with time-dependent boundaries falls short of the interior order, the cancellation mechanism would be shown not to work as claimed.

read the original abstract

Explicit Runge-Kutta (RK) integration of hyperbolic initial-boundary value problems with time-dependent Dirichlet data often displays order reduction: the observed convergence order falls below the nominal order because the stage structure interacts with asymmetric near-boundary spatial closures. This paper develops a purely spatial remedy that preserves the time integrator while redesigning only the first two boundary-adjacent derivative operators. For an arbitrary explicit $s$-stage RK method applied to linear advection, the one-step truncation error at the boundary-adjacent nodes is shown to admit a tableau-dependent decomposition whose cancellation yields explicit algebraic conditions on the boundary weights. A solvability coefficient $R(\mathbf{b},\mathbf{c},A)$ determines whether a spatial compensation mechanism exists; the result is specialised to SSP-RK3, for which closed-form conditions are derived. Constrained differential evolution then identifies 5-point closures that, coupled to a 5th-order upwind interior stencil, recover third-order convergence from the degraded second-order behaviour of classical Taylor closures. A stability-aware variant augments the optimisation with an eigenvalue penalty, exposing the trade-off between order recovery and CFL robustness. Validation covers linear advection, manufactured-solution Burgers flow, and dimensionally split two-dimensional advection. The analysis clarifies why weak-stage-order temporal fixes do not resolve the finite-difference boundary problem, and indicates how the framework extends to non-uniform meshes.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper derives tableau-specific algebraic conditions on boundary stencils to cancel leading truncation errors for explicit RK on linear advection, then optimizes and tests the resulting closures on nonlinear and multi-D cases.

read the letter

The main contribution is a truncation-error decomposition for the one-step error at boundary-adjacent nodes when an explicit s-stage RK method is applied to linear advection. This decomposition produces explicit algebraic conditions on the near-boundary weights, governed by a solvability coefficient R(b,c,A) that depends on the RK tableau. For SSP-RK3 they obtain closed-form conditions, then use constrained differential evolution to find 5-point stencils that restore third-order convergence when paired with a fifth-order interior upwind operator. A stability-penalized variant trades some order for better CFL robustness. The tests on linear advection, manufactured Burgers solutions, and dimensionally-split 2-D advection all show the recovered order, and the paper explains why weak-stage-order temporal fixes do not solve the spatial boundary problem. This is a practical, purely spatial remedy that leaves the time integrator unchanged. The derivation and the optimization procedure are the genuinely new pieces. The numerical evidence across the three test regimes is consistent and supports the claim that the optimized closures work in practice. The soft spot is that the cancellation mechanism is derived only for constant-coefficient linear advection; the same stencils are reused on the nonlinear and multi-dimensional problems with no further analysis showing that the leading error term still cancels once the flux is nonlinear. The results are empirical rather than proven in those settings, though the observed orders are clean. This work is aimed at people who implement high-order finite-difference schemes for hyperbolic conservation laws and need to handle time-dependent Dirichlet data without order loss. A reader who already uses SSP-RK3 or similar explicit integrators will find the explicit conditions and the stencil examples directly usable. The combination of the linear analysis, the optimization, and the validation is solid enough to merit referee time, even if reviewers will likely press on the extension to nonlinear problems.

Referee Report

2 major / 2 minor

Summary. The manuscript develops a purely spatial remedy for order reduction in explicit Runge-Kutta integration of hyperbolic IBVPs with time-dependent Dirichlet data. For linear advection it derives a tableau-dependent decomposition of the one-step truncation error at boundary-adjacent nodes whose cancellation produces explicit algebraic conditions on the boundary weights, introduces the solvability coefficient R(b,c,A), specializes the conditions to SSP-RK3, and uses constrained differential evolution to obtain 5-point closures that, when paired with a 5th-order interior stencil, recover third-order convergence. The same closures are then tested numerically on manufactured-solution Burgers flow and dimensionally-split 2-D advection; a stability-aware variant of the optimization is also presented.

Significance. If the spatial compensation mechanism extends beyond the linear constant-coefficient setting, the work would be a useful contribution to high-order finite-difference methods for hyperbolic problems. The explicit derivation of the truncation-error conditions and the solvability coefficient for arbitrary explicit RK tableaux constitute a clear theoretical advance, while the optimization procedure and the inclusion of an eigenvalue penalty for CFL robustness are practical strengths. The paper correctly distinguishes the boundary-order issue from weak-stage-order temporal remedies.

major comments (2)

[§2] §2: The one-step truncation error decomposition and the solvability coefficient R(b,c,A) are derived exclusively under the assumption of linear constant-coefficient advection. The manuscript re-uses the resulting 5-point closures on the nonlinear Burgers equation and on dimensionally-split 2-D advection without supplying an analogous truncation-error analysis showing that the same cancellation still occurs once the flux is nonlinear or the problem is multi-dimensional.
[§5] §5 (numerical validation): Order recovery is reported for the manufactured-solution Burgers test, yet the absence of a nonlinear error analysis leaves open the possibility that the observed third-order rates arise from the specific manufactured solution or from other discretization details rather than from the linear-derived compensation mechanism.

minor comments (2)

[§2] The notation for the boundary stencil weights and the precise definition of the solvability coefficient R(b,c,A) would be clearer if collected in a single display equation or table early in §2.
Figure captions for the stability diagrams should explicitly state the CFL number and the comparison baseline (classical Taylor closure) used in each panel.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the careful reading and constructive comments on our manuscript. We address each major comment below with the strongest honest defense possible, without misrepresenting the scope of the analysis or the numerical results presented.

read point-by-point responses

Referee: [§2] §2: The one-step truncation error decomposition and the solvability coefficient R(b,c,A) are derived exclusively under the assumption of linear constant-coefficient advection. The manuscript re-uses the resulting 5-point closures on the nonlinear Burgers equation and on dimensionally-split 2-D advection without supplying an analogous truncation-error analysis showing that the same cancellation still occurs once the flux is nonlinear or the problem is multi-dimensional.

Authors: We agree that the truncation-error decomposition and the solvability coefficient R(b,c,A) are derived only for linear constant-coefficient advection. The 5-point closures obtained from that analysis are then applied without modification to the nonlinear Burgers and dimensionally-split 2-D tests. A full nonlinear truncation-error analysis would require expanding the flux Jacobian and tracking how the leading boundary error terms interact with the RK stages; this extension is not performed in the manuscript. The numerical evidence nevertheless shows consistent recovery of third-order accuracy on the manufactured Burgers solution and on the 2-D problem. We will add a short clarifying paragraph in §2 and §5 stating that the theoretical cancellation conditions are linear-specific while the nonlinear and multi-dimensional results remain empirical. revision: partial
Referee: [§5] §5 (numerical validation): Order recovery is reported for the manufactured-solution Burgers test, yet the absence of a nonlinear error analysis leaves open the possibility that the observed third-order rates arise from the specific manufactured solution or from other discretization details rather than from the linear-derived compensation mechanism.

Authors: The manufactured solution is constructed to satisfy the nonlinear Burgers equation exactly, so the observed convergence rates measure the combined spatial-temporal discretization error. The same boundary closures recover the design order on the linear advection problem (where the analysis applies directly) and on the 2-D split advection problem. While we acknowledge that a nonlinear error analysis would eliminate any residual doubt, the consistency of the order recovery across three distinct problem classes makes it unlikely that the result is an artifact of the particular manufactured solution. No additional numerical tests or analysis will be added at this time. revision: no

standing simulated objections not resolved

A complete nonlinear truncation-error analysis demonstrating that the same cancellation occurs for general nonlinear fluxes.

Circularity Check

0 steps flagged

No significant circularity; derivation from truncation-error analysis is independent

full rationale

The central algebraic conditions on boundary weights are obtained by direct one-step truncation-error decomposition for the linear constant-coefficient advection equation, yielding the solvability coefficient R(b,c,A) and explicit constraints for SSP-RK3. These constraints are then imposed on a constrained differential-evolution search that produces the 5-point stencils; the subsequent numerical tests on Burgers and 2-D advection serve as external validation rather than inputs to the derivation. No step reduces by construction to a fitted parameter, self-definition, or self-citation chain. The paper remains self-contained against its own error-analysis benchmark for the linear case.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The approach assumes the existence of a spatial fix via weight conditions for the given RK methods, with optimization introducing fitted parameters for practical implementation.

free parameters (1)

boundary stencil weights = determined by constrained differential evolution
The specific coefficients for the 5-point boundary closures are found via optimization to satisfy the derived algebraic conditions while considering stability.

axioms (2)

domain assumption The truncation error admits a tableau-dependent decomposition
This is the key step in the analysis for linear advection.
standard math Standard finite difference Taylor expansions hold near boundaries
Used to analyze the one-step error.

pith-pipeline@v0.9.0 · 5563 in / 1433 out tokens · 74955 ms · 2026-05-10T16:47:52.139345+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

3 extracted references · 3 canonical work pages

[1]

Biswas, D

[1]A. Biswas, D. I. Ketcheson, S. Roberts, B. Seibold, and D. Shirokoff,Explicit Runge– Kutta methods that alleviate order reduction, SIAM J. Numer. Anal., 63 (2025), pp. 1398–

work page 2025
[2]

Biswas, D

[2]A. Biswas, D. I. Ketcheson, B. Seibold, and D. Shirokoff,Weak stage order conditions for Runge–Kutta methods, 2023, https://arxiv.org/abs/2310.02817. [3]A. Biswas, D. I. Ketcheson, B. Seibold, and D. Shirokoff,Algebraic structure of the weak stage order conditions for Runge–Kutta methods, SIAM J. Numer. Anal., 62 (2024), pp. 48–72. [4]S. A. Gershgorin,...

work page arXiv 2023
[3]

Lubich and A

[6]C. Lubich and A. Ostermann,Runge–Kutta methods for parabolic equations and convolution quadrature, Math. Comput., 64 (1995), pp. 601–627. [7]K. Mattsson and M. H. Almquist,A solution to the stability issues with block norm sum- mation by parts operators, J. Comput. Phys., 253 (2014), pp. 418–442. [8]K. Mattsson and J. Nordstr ¨om,Summation by parts ope...

work page 1995

[1] [1]

Biswas, D

[1]A. Biswas, D. I. Ketcheson, S. Roberts, B. Seibold, and D. Shirokoff,Explicit Runge– Kutta methods that alleviate order reduction, SIAM J. Numer. Anal., 63 (2025), pp. 1398–

work page 2025

[2] [2]

Biswas, D

[2]A. Biswas, D. I. Ketcheson, B. Seibold, and D. Shirokoff,Weak stage order conditions for Runge–Kutta methods, 2023, https://arxiv.org/abs/2310.02817. [3]A. Biswas, D. I. Ketcheson, B. Seibold, and D. Shirokoff,Algebraic structure of the weak stage order conditions for Runge–Kutta methods, SIAM J. Numer. Anal., 62 (2024), pp. 48–72. [4]S. A. Gershgorin,...

work page arXiv 2023

[3] [3]

Lubich and A

[6]C. Lubich and A. Ostermann,Runge–Kutta methods for parabolic equations and convolution quadrature, Math. Comput., 64 (1995), pp. 601–627. [7]K. Mattsson and M. H. Almquist,A solution to the stability issues with block norm sum- mation by parts operators, J. Comput. Phys., 253 (2014), pp. 418–442. [8]K. Mattsson and J. Nordstr ¨om,Summation by parts ope...

work page 1995