ABSTRACT
This chapter discusses how we can speed-up yield verification, and MC in general, by using advanced sampling techniques. Interestingly in school children learn about statistics without Monte-Carlo and apply basic analytical calculus instead and use combinatorial approaches. Nonrandom sampling schemes can lead to more stable estimates, e.g., for sample mean and standard deviation of the performances or for the correlations in a multivariate analysis. A simple orthogonal grid approach is limited, and a pure random approach has slow general convergence. One aspect in multi-dimensional sets is the correlation between the different variables. In statistics, bootstrap is something different but in some way also similar to what it is in circuit design.
