Local time stepping for a mass‐consistent and time‐split advection scheme
Michael A. Baldauf
Abstract
Michael A. Baldauf
Abstract
An advection scheme that is locally mass‐conserving, positive‐definite, and has strongly reduced splitting errors in deformational flows is presented in a recent work by Bott. Additionally, a row‐oriented substepping is proposed to deal with Courant numbers C larger than one. However, this leads to either a nonlocal communication pattern or a violation of numerical reproducibility under domain decomposition in a (massively) parallel computing environment. Instead, a local time stepping is proposed here, which acts only on those grid points where C exceeds one. It is shown that this local time stepping maintains the above‐mentioned properties of the Bott scheme and that it achieves numerical reproducibility, together with a slight improvement in efficiency.
OpenAlex reports 6 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
An advection scheme that is locally mass‐conserving, positive‐definite, and has strongly reduced splitting errors in deformational flows is presented in a recent work by Bott. Additionally, a row‐oriented substepping is proposed to deal with Courant numbers C larger than one. However, this leads to either a nonlocal communication pattern or a violation of numerical reproducibility under domain decomposition in a (massively) parallel computing environment. Instead, a local time stepping is proposed here, which acts only on those grid points where C exceeds one. It is shown that this local time stepping maintains the above‐mentioned properties of the Bott scheme and that it achieves numerical reproducibility, together with a slight improvement in efficiency.
Key concepts: Advection, Time stepping, Grid, Scheme (mathematics), Domain decomposition methods, Computer science, Work (physics), Mathematics