ITAPS Component Services

Several mesh and geometry component services have been developed that are useful to applications in their own right and can also used as building blocks by integrated services. We give an overview of several of the services that have seen the most development to date. More information can be obtained by following the links to the left.

Mesh Quality Improvement.  Mesh quality is critical for obtaining optimal efficiency and accuracy in numerical simulations. The Mesh Quality Improvement Toolkit (Mesquite) is a highly flexible, efficient, and comprehensive software package that incorporates state-of-the-art algorithms to optimize the shape of mesh elements by relocating vertices in both two and three dimensions. Close collaborations with the TOPS center has resulted in a new feasible Newton solver for Mesquite that can find the optimal location of millions of vertices in just a few minutes \cite{FeasNewt}. Mesquite interactions with SciDAC-1 applications include: updating meshes on deforming geometry for the SLAC design optimization effort, improvement of hybrid and unstructured meshes in several meshing codes (Cubit, Overture, NWGrid, GRUMMP), and smoothing of geodesic meshes in support of the SciDAC climate group at CSU. In addition, Mesquite has been used in many non-SciDAC applications including smoothing of arteriovenous graft meshes, Alegra ALE rezone, and CSAR propellant burn. The code has been released as open source and downloaded over 400 times in the past two years.

Front Tracking. Front tracking can significantly improve the accuracy of simulations which have sharply defined moving surfaces of discontinuity. These surfaces are given explicit computational degrees of freedom, supplementing the continuous solution values at regular grid points. Our method uses the locally grid-based method which employs an accurate Lagrangian algorithm for front propagation but adopts the boxed Eulerian reconstruction for topological bifurcation. This method is both accurate and robust and was found to be the best of several methods tested, including level sets and volume of fluids \cite{DuFixGli05}. During SciDAC-1, the front tracking algorithms implemented in {\sl FronTierLite} underwent several important innovations and was modularized into a physics-independent scientific software package. In addition, a conservative scheme was implemented in all dimensions; its efficiency is still to be tuned. The FronTierLite code has been used in many application areas including the simulation of the diesel jet injection, the study of the fluid mixing, the MHD simulation of fusion pellet injection, target design for advanced accelerators, and the simulation of type Ia supernova. FronTierLite is now in public release, with over 1000 hits per month and 10's of downloads per month.

Mesh Swapping. Mesh swapping improves the quality of unstructured meshes by changing the local topology of the mesh. This technique is especially helpful in three dimensions, where the combination of swapping and vertex relocation has been shown to dramatically improve mesh quality and simulation efficiency \cite{frei-cfog:ijnme97,fo-g:cost-benefit}. As part of SciDAC-1, a state-of-the-art swapping algorithm \cite{grummp} was re-implemented as a stand-alone library using the mesh and geometry interfaces \cite{tstt-swap-tool}. In two dimensions, edge swapping is supported, while in three dimensions, both edge and face swapping are supported. In addition, users can define their own mesh quality metrics, allowing great flexibility in usage. This swapping software is already in use in one widely-distributed SciDAC meshing code (GRUMMP), which is used in numerous non-SciDAC applications in fluid and solid mechanics, medicine, and astrophysics, with over 500 downloads per year.

Unstructured Mesh h-Adaptation. The mesh adaptation service modifies a given input mesh to be consistent with a specified target mesh size field. That is, refinement and coarsening operators are applied to tetrahedral mesh elements until they match the specified target sizes \cite{shep05,wan05} which can include anisotropic \cite{li04,li05} mesh configurations. These procedures have been developed to work in conjunction with a CAD definition of the geometric domain so that as the mesh is refined its geometric approximation is automatically improved \cite{li05,shep05}. A preliminary version of this service has been used in the accelerator and fusion application areas and has recently been made available through the ITAPS software web site.

Support for Distributed Meshes. Many of the SciDAC applications require ITAPS services operate on meshes that are distributed over the processors of a parallel computer. To meet these needs, the ITAPS team has developed two tools that support distributed mesh operations. The first is a parallel mesh generation process that distributes a CAD model over multiple processors allowing each processor to mesh surfaces and volumes in parallel yielding a valid conforming mesh of the entire domain. The second tool is the addition of a partition model and parallel mesh migration procedures to our flexible mesh database (FMDB)  to support dynamic load balancing in the parallel adaptive loops service. Both of the distributed mesh support tools have been used on the SLAC accelerator modeling project. The parallel partitioning and migration are part of FMDB which is available for download.