Abstract: The present disclosure relates to a method for probabilistic simulation of a microelectronic device, said method being implemented automatically by an electronic processing device and including the following steps: a) defining a plurality of first samples of the device from a probability distribution of at least one physical parameter of the device; b) for each first sample, determining, through an electrical simulation method, the value of at least one operating variable of the device; c) defining, by regression from values of the physical parameters and operating variables of the first samples simulated in step b), a mathematical model approximating the response of the electrical simulation method.

Abstract: A method for determining, for each of at least p physical parameters of one or several components of an electronic circuit or of a microelectromechanical system, a number n of experiment values of the physical parameter includes determining n vectors of dimension p, each component of each of the vectors corresponding to one of n initial values of one of physical parameters; and iteratively modifying at least some of the n vectors to bring to a maximum, at least locally, for each pair of vectors from among pairs of n vectors, the smallest average of the sum of distances between the vectors of said pair projected onto sub-spaces of dimension k, where k belongs to a set of integers ranging between 1 and p and at least comprising 1, 2, and p, the components of each of the n vectors corresponding, at the end of the iterations, to experiment values.

Abstract: A method for determining, for each of at least p physical parameters of one or several components of an electronic circuit or of a microelectromechanical system, a number n of experiment values of the physical parameter includes determining n vectors of dimension p, each component of each of the vectors corresponding to one of n initial values of one of physical parameters; and iteratively modifying at least some of the n vectors to bring to a maximum, at least locally, for each pair of vectors from among pairs of n vectors, the smallest average of the sum of distances between the vectors of said pair projected onto sub-spaces of dimension k, where k belongs to a set of integers ranging between 1 and p and at least comprising 1, 2, and p, the components of each of the n vectors corresponding, at the end of the iterations, to experiment values.