Dimension reduction procedure for the simplicial approximation approach to design centering

The simplicial approximation approach to design centering1 is based on developing a polyhedral approximation to the feasible region in the parameter space. Although the original method has been modified so as to allow arbitrary statistical distributions,2 it still becomes computationally expensive as the dimension of the parameter space becomes large. In this paper, a technique is proposed for reducing the dimension of the parameter space and thereby significantly increasing the size of the circuit for which the simplicial approximation approach can be applied. This technique is particularly useful in integrated circuit applications where strong correlations between circuit elements exist, and the number of circuit elements can be large.