ENO reconstruction and ENO interpolation are stable (Q1955525)

The authors of this article state some stability results for an essentially non-oscillatory (ENO) reconstruction and ENO interpolation procedure. They prove that both of these methods are stable in the following sense: the sign of the jumps of the reconstructed point values at cell interfaces remains the same as the sign of the jumps in the underlying cell averages across cell interfaces. To this end, suitable algorithms for a high order ENO stencil (both for the reconstruction and interpolation approach) are defined which depend on the smoothness of the data. Thus spurious oscillations are avoided and the so-called ``Sign Property'', the mentioned property that the jump in the ENO reconstruction resp. interpolation ``cannot have an opposite sign to the jump in the underlying cell averages'', can be proven. Furthermore, upper bounds for the relative jumps in the ENO reconstruction resp. interpolation are given. Although these results hold for arbitrary meshes and any order of reconstruction, some parts of the proofs depend on the ENO stencil procedure, hence other reconstruction procedures (as for example WENO) ``will generally not satisfy a similar sign property''.