Automatic and Adaptive Mesh Generation

Algorithms for numerical field computation have always been strongly influenced by the need to adapt the structure of node points to the problem configuration, and specifically to the configuration of the field. In the typical field computation, there are portions of the problem domain in which the field varies only very slowly and accordingly, widely spaced node points are acceptable. Alternatively, there are other portions of the problem domain in which the field varies quite rapidly, and where it must be known quite accurately. In these places, the node points must be closely spaced.