Abstract: The simulation information creating part 111 generates random numbers corresponding to the correlation coefficient data between the simulation model parameters, and also creates the simulation model library having correlated model parameters with variations, and the simulation net list for the objective characteristic, by Monte Carlo method using the typical simulation model library, the standard deviations for the plural simulation model parameters and the simulation net list for the typical objective characteristic. The simulation part 112 obtains samples having the objective characteristics with variations. The yield estimation part 120, estimates the yields by determining whether the samples satisfy the predetermined specification or not (Pass or Fail), wherein by repeating the determination by making the simulations again only for the samples on which the filter having the function of learning the boundary for decision of Pass or Fail did not determine as Pass.

Abstract: The simulation information creating part 111 generates random numbers corresponding to the correlation coefficient data between the simulation model parameters, and also creates the simulation model library having correlated model parameters with variations, and the simulation net list for the objective characteristic, by Monte Carlo method using the typical simulation model library, the standard deviations for the plural simulation model parameters and the simulation net list for the typical objective characteristic. The simulation part 112 obtains samples having the objective characteristics with variations. The yield estimation part 120, estimates the yields by determining whether the samples satisfy the predetermined specification or not (Pass or Fail), wherein by repeating the determination by making the simulations again only for the samples on which the filter having the function of learning the boundary for decision of Pass or Fail did not determine as Pass.