Abstract | ||
---|---|---|
High frequency errors are always present in numerical simulations since no difference stencil is accurate in the vicinity of the $$\pi $$-mode. To remove the defective high wave number information from the solution, artificial dissipation operators or filter operators may be applied. Since stability is our main concern, we are interested in schemes on summation-by-parts (SBP) form with weak imposition of boundary conditions. Artificial dissipation operators preserving the accuracy and energy stability of SBP schemes are available. However, for filtering procedures it was recently shown that stability problems may occur, even for originally energy stable (in the absence of filtering) SBP based schemes. More precisely, it was shown that even the sharpest possible energy bound becomes very weak as the number of filtrations grow. This suggest that successful filtering include a delicate balance between the need to remove high frequency oscillations (filter often) and the need to avoid possible growth (filter seldom). We will discuss this problem and propose a remedy. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1007/s10915-019-01116-9 | Journal of Scientific Computing |
Keywords | Field | DocType |
Numerical filters, Stability, Accuracy, Summation-by-parts, High frequency oscillations, Semi-bounded, Transmission problem | Applied mathematics,Boundary value problem,Oscillation,Dissipation,Mathematical analysis,High wave number,Stencil,Filter (signal processing),Operator (computer programming),Mathematics,Energy stability | Journal |
Volume | Issue | ISSN |
82 | 1 | 0885-7474 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Tomas Lundquist | 1 | 0 | 0.34 |
Jan Nordström | 2 | 218 | 31.47 |