Hey numpy people! Do you know the better way to compute with numpy: r**M where r is a positive real and M a 2x2 (rank 2) matrix, symmetric or not, with real or not eigenvalues. Help or comment are appreciated, Jean-Bernard