Solution of the Discrete Resources Distribution Problem by Methods of Linear Programming

Abstract

Introduction. When planning projects, work packages, one often has to solve distribution-type problems associated with the optimal allocation of resources. Existing methods for solving such problems suggest the presence of an analytical dependence of a continuous type between the volumes of the distributed resource and performance indicators. However, optimization problems become inapplicable when the resource is discrete and dependencies are specified in tabular ways. Aim. To develop a mathematical model for solving the problem of the optimal distribution of resources of a discrete type with table-defined optimality criteria using linear programming methods. Describe the method of numerical solution of the problem using computer technology. Materials and methods. It is possible to solve the problem by forming an analytical dependence of a piecewise-continuous type between the volume of the distributed resource and the optimality criterion. This allows us to formulate an optimization problem solved by mathematical programming methods. It is possible to construct an analytical objective function by introducing additional parameters and restrictions. A numerical solution to the problem can be obtained both using mathematical packages of applied programs, such as, for example, “Mathcad”, and using programming systems. The paper describes the methodology for solving problems in the MS Excel environment using the add-on “Solver”. Results. A mathematical model is developed for solving a discrete distribution problem for the optimality criterion given in a table by integer linear programming methods. The technique of numerical solution in the MS Excel environment is described. Conclusion. Previously, such problems were solved by dynamic programming methods, which is more difficult in the computational plan. Conducted computational experiments showed high accuracy of model calculations and resistance to changes in the source data.