Domain decomposition approach for parallel improvement of tetrahedral meshes

Abstract

, +44 1745 535213 Highlights  CdCl 2 anneal treatment resulted in S diffusing to the back contact.  High Zn levels created mixed cubic/hexagonal structure at the p-n junction.  Increased Zn in Cd 1-x Zn x S supressed S diffusion into CdTe.

Scalable generation of large-scale unstructured meshes by a novel domain decomposition approach

“…Galtier & George [15] Delaunay Type IV Negatively impacted by the time-consuming domain decomposition step. Larwood, et al [16] Delaunay Type IV Chen, et al [4] Delaunay Type IV Said, et al [17] Delaunay Type I Limited scalability was reported. Ito, et al [18] AFT Type V Negatively impacted by the sequential mesh improvement procedure.…”

“…Therefore, the clock time for preparing a large-scale unstructured mesh is usually comparable with or more than the time consumed for conducting parallel simulations. To fully exploit the computing power of the present and future emerging parallel computers, an environment where both the simulation and its mesh generation codes are parallelized is now highly demanded [2][3][4][5][6][7] by research developers as well as industry end-users.…”

Parallel and automatic isotropic tetrahedral mesh generation of misaligned assemblies

“…For instance, it takes weeks or more to prepare a block-structured mesh for an external flow simulation over a complete aircraft model, even by an engineer with expertise in applying state-of-the-art meshing tools (Baker 2005). Unstructured mesh generation does not require a painful block-decomposition process, but preparing a suitable input for the unstructured meshing pipeline remains a major performance bottleneck in many cases (Shimada 2011;Chen et al 2015Chen et al , 2017, in general, this input contains a geometry that defines the meshing domain and a sizing control that defines the distribution of element scales over the meshing domain (Deister et al 2004;Kania and Pirzadeh 2005;Pirzadeh 2010;Quadros et al 2010). It remains a very active research topic to automate the preparation of this input.…”

“…As a result, mesh generation has become a prominent factor that determines the efficiency and scalability of parallel simulations (Weatherill et al 2002;Freitas et al 2013;Löhner 2013Löhner , 2014Yilmaz and Ozturan 2015;Laug et al 2017;Chen et al 2017). Automating and parallelizing the mesh generation pipeline becomes a long-standing goal pursued by both the mesh generation and HPC communities (Zagaris et al 2009;Loseille et al 2015).…”

“…The second algorithm is developed for parallel mesh generation. In general, the existing parallel mesh generation approaches could be classified into the algorithm-parallel ones and the problem-parallel ones (Cougny and Shephard 1999;Chrisochoides 2006;Foteinos and Chrisochoides 2011;Chen et al 2017Chen et al , 2018. Presently, the problem-parallel approaches are preferred in many studies because of their better capability of reusing state-of-theart sequential mesh generation codes (Chrisochoides and Nave 2003;Chen et al 2012).…”