Disclosed is a method for representing or storing a sparse matrix and a method of calculating with the sparse matrix. Each line of the matrix is stored as a list of predetermined size, of the none zero values and indices indicating the position in the line of the stored value. The matrixes may be processed using parallel processors with a CPU allocating the computation of the values in each line to a thread for each processor. The computation may be an iterative process to solve a linear algebra problem. The matrixes and method of calculation may be used as part of a computer simulation of a moving system, such as the simulation of an incompressible fluid flow in smoothed particle hydrodyanimics.
展开▼
