From your more recent question, I found:
I am still not sure what a is (so I will assume it is a constant like your m1 and m). So instead of those for loops you could just:
A2 = zeros(size(m));
A2(m1 >= 0 & m >= 0) = ...
alpha( a, m1(m1 >= 0 & m >= 0), m(m1 >= 0 & m >= 0) );
And similarly for the other conditions.
Now... I realize that this produces a (P+M+1)x(P+M+1) matrix instead of a ((P+M+1)*(P+N+1))x((P+M+1)*(P+N+1)) matrix. Once you have these values for the smaller matrix you can populate the larger matrix by essentially copying/replicating... you don't want to have to do the same computations (P+M+1)*(P+N+1) times if you only need to do it once and then just replicate.