Sensitivity analysis of an agricultural linear programming model

The paper presents a method of sensitivity analysis for linear systems and its practical application, allowing directs interpretation of results. The subject of analysis is a set of large regional agricultural LP models meant for rationalization of regional agricultural policies in Polish conditions. Sensitivity analysis is performed with regard to perturbations in values of elements of the LP basis matrix for a given LP solution. Software developed makes it possible to analyze solutions obtained with MPS 360 or MPSX 370. Structural approach was applied consisting in block-triangularization of the matrix defining the set of equations corresponding to the optimum vertex. The approach lowers computational burden and yields qualitative sensitivity assessment, i.e. enables exclusion of influences of certain parameters on some variables.