The Fourier amplitude sensitivity test (FAST) is a quantitative method [3]. It computes the contribution of each parameter to variations of the target function. It is called the “main effect” and defined as

S, = varx.|£ (y|x >’ (2)

j var(y)

Variations in numerator and denominator of (2) are multidimensional integrals over appropriate spaces. Their computation is very expensive. In the FAST, they are replaced by the one-dimensional integrals over the some curve exploring the space

x (s) = K (sin o,-s) (3)

In the next stage of the method, the target function y(s) = y(x(s),x2(s),…,xn(s)) is expanded in the

+да

Fourier series y(s) = ^ {A. cos js + Bj sin js}, and then the spectrum Л2 = Aj + B^ of the Fourier

J=-n

series expansion is used for calculation of the main effects.

The FAST algorithm can quantify influence of the parameters but for this it requires more calculations of the target function than the Morris method. If the system depends on very large number of parameters then it would be reasonable first to apply the Morris method and then to quantify the influence of only the most important parameters by the FAST algorithm.