ABSTRACT
Beginning informally, consider a familiar kind of constrained optimization problem. You need to decide where to have lunch. You have a consideration set of several conveniently named restaurants: Burgers, Pizza, Couscous, Caminetto, Salad, Sushi, Curry, Chinese, Asian Fusion, Schnitzel, Brasserie, and Greasy Spoon. You also have information on each of these restaurants that you consider relevant to your decision: price, distance, and health value. Let us assume that for each of these you have reliable scores on a (more or less) continuous scale. Price is in the local currency. Distance is measured in time or kilometers. Health value is based on a rating scheme published by your favorite source, Whelp.
