Corrigendum to: ``Discontinuous Galerkin methods for the one-dimensional Vlasov-Poisson system'' (Q6985366)

The authors revisit earlier analyses of discontinuous Galerkin (DG) methods for the Vlasov-Poisson system in one and higher dimensions. In those works, optimal error estimates relied on a specially designed projection \(\Pi_h\) that depends on the sign of the electric field \(E\) (or \(E\cdot n\)). However, they later realized that the electric field may change sign over time, making \(\Pi_h\) discontinuous in time and invalidating the original analysis for the linear part of the problem. While flow-dependent projections are currently the only known approach for proving optimal error estimates for DG methods on hyperbolic problems, handling time- and space-varying flow speeds is technically difficult. Existing techniques that address this issue -- for instance, a recent one-dimensional transport analysis -- cannot be extended to the Vlasov-Poisson system because of its structural complexity and the challenge of obtaining localized a priori estimates. As a result, the authors cannot provide a corrected proof of optimal convergence rates. Instead, they present revised results with suboptimal, yet still non-asymptotic, error bounds that remain valid for all mesh sizes \(h\leq 1\). By replacing the crafted projection with the standard \(L^2\)-orthogonal projection, the analysis becomes simpler and requires less regularity from the exact solution. They note that the original optimal estimates would still apply if the electric field happened to maintain a constant sign in time. Finally, they restate the corrected results for all Poisson discretizations in a unified theorem and include updated versions of the corollaries from the earlier work.