Note that there is only one implementation of this object so there aren't any - different kernels to choose from when you create instances of the matrix object. + different kernels to choose from when you create instances of the tuple object. So for example, you could declare a tuple of 3 ints using the following statement: dlib::tuple<int,int,int> t; @@ -1303,18 +1303,6 @@ could declare a matrix of 2 rows and 3 columns using the following statement: dlib::matrix<float,2,3> m;
-- It should also be noted that matrix multiplication is fastest when the two matrices - being multiplied are not complex matrix_exp objects returned from other expressions - (such as other matrix multiplies). This is because the matrix multiply operator will - evaluate each element of the matrices it is multiplying many times, and a matrix_exp - computes its elements' values each time they are queried. However, the matrix multiply - operator is the only one that evaluates its argument's elements multiple times so you can - stack up all the other operators however you want without any performance penalty. If - you want to multiply two complex matrix_exp expressions together you can easily convert them into - fully evaluated temporary matrix objects by using the tmp() function. For example, to - multiply four matrices together you should use an expression such as result = tmp(a*b)*tmp(c*d); -