Robust facility location (Q1416795)

A facility is to be located in a planar region minimising the sum of weighted distances to a finite set of points, but only an estimate of the weights is known. It is therefore proposed to find instead the most robust location, where robustness of a location is measured as the minimum deviation in weights from the estimate which would lead to a weighted sum of distances above an a-priori fixed allowable upper bound. When this deviation is measured by a monotone norm of the weight-differences-vector, robustness may be maximised by way of standard fractional programming techniques. In some special cases, like spatial distance measured by a polyhedral norm and weight-deviations measured through maximum norm, the problem may be solved using more direct finite enumeration algorithms.