D. Sprevak et al., SCREENING OF VARIABLES FOR THE EMPIRICAL MODELING OF SEMICONDUCTOR-DEVICES, IEE proceedings. Science, measurement and technology, 143(5), 1996, pp. 319-326
A strategy for screening the important effects in a multivariable resp
onse function is presented. An optimisation algorithm is developed, wh
ich searches for the optimum of the function in the subspace of the im
portant variables, The application of these methods is illustrated wit
h a known nonlinear 23 variable function, The technique is applied to
the design of an NMOS process which depends on 35 factors.