ABSTRACT
Overview In most practical situations, the exact values of the model parameters are not known with absolute certainty. For example, model parameters are often obtained by statistical estimation procedures. The analysis of the eects of parameter changes on a given optimal solution of the model is called sensitivity analysis or post-optimality analysis. Changing the value of a parameter is called a perturbation of that parameter. In this chapter we will look at perturbations of several parameters, including objective function coecients, right hand side coecient of constraints (including the zeroes of the nonnegativity constraints), and the entries of the technology matrix. In economic terms, while the primal model gives solutions in terms of, e.g., the amount of profit obtainable from the production activity, the dual model gives information on the economic value of the limited resources and capacities. So-called shadow prices are introduced in the case of nondegenerate optimal solutions; in the case of degeneracy, right and left shadow prices are defined.
