Structured Grid Computational Pattern Name Structured Grid Problem

A problem in 2, 3 or 4 dimensions, easily described with a structured grid which can be implemented as a n-dimensional array. Each update on the data is local in the physical space. How do you design an efficient solver which provides these updates and scales to many processors? Context In science and engineering it is often necessary to solve a problem represented by a grid of points and the value of a point in space is updated by a function of other nearby points. The value could, for example, represent the strength of an electric potential, the temperature of an oven or the color of a pixel in an image. Data is arranged in a regular multidimensional grid, usually in two or three dimensions. Computation proceeds as a sequence of grid update steps. The location of nodes in the physical space directly maps to their location in data array (this is in contrast to the Unstructured Grid where the location of the element in the array cannot be directly inferred from the physical location). The updates are logically concurrent, but in practice are implemented as a sequential swee