Determining an optimal linear dynamic system in the case where no probabilistic moments of its error exist

The problem of determining an optimal linear dynamic system which insures a minimum for the probability that the system error will go beyond tolerable limits at a specified time is formulated. Initial data are considered with which the system error has no mean or variance, or the system error is not a second order random process. As an extension of methods for determining optimal systems in terms of complex statistical criteria a method is given for solution of the problem posed and examples are given.