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