Numerical Grid-generation Algorithms are becoming more critical due to the need for detailed modeling of the geometry of the physical domain and as part of solution-adaptive algorithms. Current planar grid generation algorithms are not robust because they frequently produce skewed or folded grids. These problems can be corrected by trial and error modifications (e.g., to the source term of a Poisson equation), but the procedure is far from automatic. Mathematical analysis will be used to clarify the limitations of current algorithms and to suggest new algorithms. The best algorithms will be incorporated into user-friendly automatic grid-generation codes which will have: a common interface; namelist input; interactive graphics; a library of geometries; and test problems. The codes will be designed to run on parallel and vector computers.
