Abstract
A method which enables the solution of linear programming problems without knowledge a priori of both an objective function and restrictions is presented in the paper. It is based upon the well-known method of feasible directions which it also becomes after a finite number of steps. It consists in the step-by-step approximation of the gradient of the objective function with use of measurements on the investigated system. Each step of these measurements enables an improvement of the state of the System.
Notes
†Communicated by the Author.