I am trying to understand some matlab code of an algorithm . I have a vector 1x4 of error values. The idea is to check if the errors are low enough or not. The way it is done in the algorithm is by multiplying the transpose of the error vector by the error vector ( A' * A , where A is the error matrix). this is equivalent to summing up the squared values of the error vector. Then, the result is checked to determine if the errors are low or not.

Does this procedure has a specific name ? I dont really understand why it is done this way. they could have simply taken the average of the absolute value of the errors.