Hi there. The question I wanted to ask is: Why are matrix methods so widely used for numerical solution of partial differential equations? Many times I've found that storing a whole matrix requires much more memory than just doing an iteration scheme to propagate the solution. Sometimes I think it is easier to deal with boundary conditions in matrix form, specially in second order differential equations. But sometimes storing a matrix in a program requires to store a lot of elements that are zero. What I want to know is if there is any advantage to parallelize a program when the system of equations are written in matrix form, or if there is any other reason why matrix methods are so widely used. Thanks in advance.