Accuracy analysis of automatic units by minimax linearization

The paper is conserned with finding ratiates which are confidence intervals of a nonlinear function of random variables from the specified lower values of confidence probability and the first and second moments of the arguments. The estimation is performed by using the linear function whose value is equal to the largest (smallest) value of the nonlinear function to be approximated. The form of the linear function is shown for fractional rational (nonlinear) functions.