Scale a matrix by rows or columns, or a multidimensional tensor along a specified dimension. If a is a vector of length
rows (b), returndiag (a) *b (but computed much more efficiently). Similarly, if b is a vector of lengthcolumns(a), return a* diag(b).If b is a multidimensional array and a a vector, c will have the same shape as b, with C
(i,:,...) =a(i)*b(i,:,...).If a is a multidimensional array and b a vector, c will have the same shape as a, with C
(:,...,i) =a(:,...,i)*b(i).If ind is supplied, a should be an array and b a vector of length
size (a,index). The result is then C(:,...,i,:,...) =a(:,...,i,:,...)*b(i)where i indexes the ind-th dimension.