On the inversion of the Vandermonde matrix