Solveur precond

It’s a matrix-free implementation of Minres solver used in scipy.sparse and the function Minres of Matlab. This function is adapted to solve the following problem. \(P^{-1}Ax=P^{-1}b\).

Where \(A\) is the optical flow matrix, \(b\) the right hand term and \(P\) is the chosen preconditioner already defined in Precond file.

If we are handling \(N\times M\) images, then the problem solved has \(2\times N\times M\) as size. The file contains also a function that computes the matrix-vector product of matrix of the problem and a given vector.

Note: The Laplace operator of certain image is computed using a convolution instead of computing it with a matricial product with the Laplace matrix.