The tolerance approach to sensitivity analysis of matrix coefficients in linear programming

Research developing a tolerance approach to sensitivity analysis of matrix coefficients in linear programming is described. The tolerance approach to standard linear programming considers simultaneous perturbations in more than one row or column of a linear program and allows for simultaneous and independent perturbations in the matrix coefficients. The concept of approximate tolerance is used to correct for the difficulty in computing the inverse of the perturbed basis when delineating the maximum tolerance under general perturbations. A decision maker can place a priori limits on the possible values over which certain parameters may vary. The tolerance approach delineates a maximum tolerance percentage within which all the elements in a row or column can vary simultaneously and independently from estimated value but still retain the same set of basic variables in an optimal solution.