[论文解读] A Patankar predictor-corrector approach for positivity-preserving time integration

Kamila Nurkhametova, Reid J. Gomillion+2 more|arXiv (Cornell University)|Feb 17, 2026

Numerical methods for differential equations被引用 0

一句话总结

该论文提出了一种模块化的 Patankar 风格预测-校正框架，强制时间积分生产-销毁系统的正性与守恒，应用于 SDIRK 方法并在 ODE 与 PDE 模型上进行测试。

ABSTRACT

Many natural processes, such as chemical reactions and wave dynamics, are modeled as production-destruction (PD) systems that obey positivity and linear conservation laws. Classical time integrators do not guarantee positivity and can produce negative or nonphysical numerical solutions. This paper presents a modular correction strategy that can be applied to implicit Runge-Kutta schemes, in particular SDIRK methods. The strategy combines stage-wise clipping with a ratio-based scaling that enforces invariants and is guaranteed to yield nonnegative, conservative solutions. We provide a theoretical analysis of the corrected schemes and characterize their worst-case order of accuracy relative to the underlying base method. Numerical experiments on stiff ODE systems (Robertson, MAPK, stratospheric chemistry) and a nonlinear PDE (the Korteweg-De Vries equation) demonstrate that the corrected SDIRK methods preserve positivity and invariants without significant loss of accuracy. Importantly, corrections applied only to the final stage are sufficient in practice, while applying them at all stages may distort dynamics in some cases. For explicit Runge-Kutta schemes, the correction maintained positivity but reduced convergence to first order. These results show that the proposed framework provides a simple and effective way to construct positivity-preserving integrators for stiff PD systems.

研究动机与目标

• Motivate the need for positivity and linear invariants in production-destruction systems modeled by ODEs and PDEs.
• Develop a modular predictor-corrector framework to enforce positivity and conservation without modifying base solvers.
• Provide a theoretical analysis of corrected schemes and characterize their worst-case accuracy loss relative to the base method.
• Demonstrate robustness and accuracy of the approach on stiff benchmark problems and a nonlinear PDE.
• Identify practical limitations and regimes where the method is most effective.

提出的方法

• Base solver: use predictor SDIRK methods with nonnegative weights to obtain a first approximation of stage values and the solution.
• Clipping: apply a clipping operation to any negative components of stage values to enforce nonnegativity.
• Scaling: introduce a ratio-based diagonal scaling to adjust the stages and preserve linear invariants.
• Corrector: form an averaged graph-Laplacian-like operator from clipped stages and solve a corrected linear system to obtain a nonnegative, invariant-preserving update.
• Guarantee: show the corrected update corresponds to an M-matrix, ensuring nonnegativity and preservation of invariants.
• Extensions: discuss stronger sign-structure assumptions that simplify the corrector and potential applicability to more general linear-invariant systems.

实验结果

研究问题

• RQ1 Can a predictor-corrector post-processing strategy enforce positivity and conservation when applied to RK/SDIRK time integrators for production-destruction systems?
• RQ2 How does the Patankar-based correction affect the order of accuracy and computational cost compared to the base method?
• RQ3 What are the practical limitations when applying corrections at all stages or using explicit RK schemes?
• RQ4 Do the corrected schemes preserve key invariants and positivity for stiff ODEs and nonlinear PDEs?
• RQ5 Under what structural assumptions do simplified corrections arise?

主要发现

• The proposed correction framework enforces positivity and linear invariants for SDIRK-based time integrators without modifying the base solver.
• Corrections via final-stage clipping and a ratio-based scaling yield nonnegative, conservative solutions and preserve invariants under the graph-Laplacian structure.
• Numerical experiments on stiff ODEs (Robertson, MAPK, stratospheric chemistry) and the KdV PDE demonstrate positivity preservation and invariant conservation with minimal loss of accuracy.
• Corrections applied only to the final stage are often sufficient in practice, while correcting all stages can distort dynamics in some cases.
• For explicit RK methods, the correction preserves positivity but may reduce convergence to first order.
• The framework provides a simple, effective way to construct positivity-preserving integrators for stiff production-destruction systems.