Creating a sparse tridiagonal matrix (for 2D finite difference second order Laplacian)

Hi,

I'm trying to solve a 2D finite difference problem. I simplified the problem to solving a Ax = B equation. My issue is building the A matrix in Matlab rapidly.

The matrix I need to create is a sparce matrix with 5 diagonal lines. Also, at some lines, the whole line will be filled with 0 except for a 1 at the diagonal for my Dirichlet condition.

How can I write it in Matlab so it can create this matrix rapidly? I'm simulating a 400x400 environment which means a 16000x16000 sparce matrix.

Let's say A is my sparce matrix, is there a way I can call my diagonal matrix with a vector? Like A(1:5) would select the 5 first diagonal values and the same for not centered diagonal.

Answer 1

[msaitta]

spdiags will do all of those things for you.