Fast method to compute the scalar product of gradient and given vector (Q1116641)

The paper gives a new method for computing \(g(u)^ t\cdot v\), where g is the gradient of a rational function of n variables f and v is a given vector. It is shown that the amount of work needed for computing the scalar product grows as the number of operations to evaluate f(x), while the standard method for the scalar product computation needs an amount of work which grows as the product of the number of operations to evaluate f(x) and the number of variables, i.e. the dimension of vector x.