Fully Discrete Error Analysis of the CSLDG Scheme with Numerical Characteristic Tracing and Spline Reconstruction
Abstract
A semi-Lagrangian discontinuous finite element scheme based on the characteristic Galerkin method (CSLDG) is investigated.The scheme directly discretizes an integral invariant model derived from the coupling of the transport equation and its adjoint equation.we have established, for the first time, error estimates for the fully discrete CSLDG scheme over a general domain Ωin \mathbb{R}^d.The error is decomposed into three parts: the ideal numerical solution error (the difference between the exact solution and the spatial discretization), the time-dependent test function error (the deviation of the test function caused by the numerical characteristic backtracking via an ODE solver), and the upstream region approximation error (the integration region deviation induced by numerical characteristic backtracking and spline boundary reconstruction).Under either compact-support boundary conditions or periodic boundary conditions with the velocity field and the solution sharing the same spatial periodicity, without requiring incompressibility,and assuming the standard CFL condition (Δt=O(h)) linking the time step and the mesh size, the following results are proved:Using piecewise P^K polynomial spaces, the L^2 error of the ideal numerical solution attains a suboptimal order O(h^K); If the ODE solver order satisfies D \ge 2K+2+d, the overall error retains the suboptimal order even after incorporating the time-dependent test function error;When the upstream region approximation error is further taken into account, for d \ge 2 the overall error remains suboptimal O(h^K) provided the spline reconstruction degree satisfies r \ge \max(K+1+d, d+2);
Related
- NUMERICAL ANALYSIS OF A LAX-WENDROFF TYPE LATTICE BOLTZMANN SCHEME FOR THE ADVECTION EQUATION
- Grothendieck ring of varieties and at-the-characteristic numerical invariants
- A semi-discrete scheme for horizontal mean curvature flow in the Heisenberg group
- Compact implicit high-resolution numerical scheme for multicomponent transport problems with Langmuir sorption
- A micro-macro numerical scheme for a self-organized hydrodynamics model enriched by a kinetic contribution
BibTeX
@misc{xie2026fully,
title = {{Fully Discrete Error Analysis of the CSLDG Scheme with Numerical Characteristic Tracing and Spline Reconstruction}},
author = {Zhengrong Xie},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.01448},
eprint = {2611.01448},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.01448},
}
RIS
TY - UNPB
TI - Fully Discrete Error Analysis of the CSLDG Scheme with Numerical Characteristic Tracing and Spline Reconstruction
AU - Zhengrong Xie
PY - 2026
DA - 2026/11/01
PB - arXiv
DO - 10.48550/arxiv.2611.01448
UR - https://arxiv.org/abs/2611.01448
AB - A semi-Lagrangian discontinuous finite element scheme based on the characteristic Galerkin method (CSLDG) is investigated.The scheme directly discretizes an integral invariant model derived from the coupling of the transport equation and its adjoint equation.we have established, for the first time, error estimates for the fully discrete CSLDG scheme over a general domain Ωin \mathbb{R}^d.The error is decomposed into three parts: the ideal numerical solution error (the difference between the exact solution and the spatial discretization), the time-dependent test function error (the deviation of the test function caused by the numerical characteristic backtracking via an ODE solver), and the upstream region approximation error (the integration region deviation induced by numerical characteristic backtracking and spline boundary reconstruction).Under either compact-support boundary conditions or periodic boundary conditions with the velocity field and the solution sharing the same spatial periodicity, without requiring incompressibility,and assuming the standard CFL condition (Δt=O(h)) linking the time step and the mesh size, the following results are proved:Using piecewise P^K polynomial spaces, the L^2 error of the ideal numerical solution attains a suboptimal order O(h^K); If the ODE solver order satisfies D \ge 2K+2+d, the overall error retains the suboptimal order even after incorporating the time-dependent test function error;When the upstream region approximation error is further taken into account, for d \ge 2 the overall error remains suboptimal O(h^K) provided the spline reconstruction degree satisfies r \ge \max(K+1+d, d+2);
ER -