Local time stepping for the shallow water equations in MPAS

Capodaglio, Giacomo; Petersen, Mark

doi:10.1016/j.jcp.2021.110818

Cited by 9 publications

(13 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As such, the results from Dawson et al. (2013) are comparable to the theoretical maximum speedup given by Equation , which LTS3 achieves (Capodaglio & Petersen, 2022).…”

Section: Resultssupporting

confidence: 53%

“…In general, we see that the best speedups can be achieved on meshes with fewer fine cells relative to the number of coarse cells. It was shown in Capodaglio and Petersen (2022) that LTS3 is able to achieve this theoretical maximum speedup in practice for a given mesh and configuration of the LTS regions.…”

Section: Methodsmentioning

confidence: 87%

“…LTS requires that cells in a given mesh be sorted into different groups, some are coarse cells, some are fine cells, and some are interface cells. Each type of cell requires a different amount of work, so load balancing among MPI processes is non‐trivial; an efficient way to address this has been a primary concern of Capodaglio and Petersen (2022). For effective load‐balancing, each MPI rank is given three sets of cells, one set of fine cells, one set of coarse cells, and one set of interface cells.…”

Section: Methodsmentioning

confidence: 99%

“…LTS3 was originally developed for use in the MPAS framework by Hoang et al. (2019), then implemented in the MPAS framework by Capodaglio and Petersen (2022). While a full review of the LTS3 scheme is outside the scope of this paper, and can be found in Hoang et al.…”

Section: Methodsmentioning

confidence: 99%

“…(2019) are particularly relevant to the MPAS framework as they were developed specifically to solve the shallow water equations (SWEs) using TRiSK horizontal discretizations. These time‐stepping schemes have been implemented by the authors in the shallow water core of MPAS‐O and it has been shown that they can solve the SWEs up to 70% faster than higher‐order global methods (Capodaglio & Petersen, 2022).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Storm Surge Modeling as an Application of Local Time‐Stepping in MPAS‐Ocean

Capodaglio

Petersen

et al. 2023

J Adv Model Earth Syst

Self Cite

View full text Add to dashboard Cite

Variable-resolution climate models have become increasingly popular in the past decade (e.g., Skamarock et al., 2012;Danilov et al., 2017;Korn, 2017), as they offer the ability to create high-resolution regions within global domains, with fine control over the extent and transitions in grid-cell size. It is well known that the size of the largest time-step that can be used by an explicit time-stepping scheme is bounded above by the size of the smallest cell in the mesh according to the Courant-Friedrichs-Lewy (CFL) condition. This restriction is of particular interest on meshes where the cell size varies greatly. To optimize the computational cost of running a model on a variable-resolution mesh, one would like to select a small time-step on regions of high resolution (regions defined by small cells), and a large time-step on regions of low resolution (defined by large cells). For simplicity, variable-resolution climate model components typically use global time-stepping schemes, where a uniform time-step is used on the entire computational domain. As a result, one is forced to use a small time-step that is restricted by the CFL condition influenced by the smallest cell in the mesh even on large cells that would admit a larger time-step in the absence of smaller cells. This approach results in unnecessary computational cost

show abstract

“…As such, the results from Dawson et al. (2013) are comparable to the theoretical maximum speedup given by Equation , which LTS3 achieves (Capodaglio & Petersen, 2022).…”

Section: Resultssupporting

confidence: 53%

Section: Methodsmentioning

confidence: 87%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations