It is very rare that you need to have matrices of different names (as I explain in the article prior to the one you reference).
Sometimes people THINK they need to do this because they are trying to wrap SAS/IML statements within a macro loop, which is rarely necessary. Instead, you can use the SAS/IML language itself to perform whatever the macro loop is doing.
You didn't post an example, but one common programming technique is to have all of the x vars in an nx10 matrix, where each column is a variable.
Then the following statements form a matrix m where each column is what you want:
y = j(n,1,0) || x; /* temp: append column of zeros */
m = y[,2:11] - y[,1:10];
free y;
Another alternative, if you don't need the whole matrix, is to loop:
do i = 1 to 10;
if i = 1 then m=x[,i];
else m = x[,i] - x[,i-1];
/* do something with m, which is m_i in your notation */
end;