Combinatorial optimization problems with normal random costs

We consider combinatorial optimization problems with random jointly normal costs. Target-probability efficient solutions are those that grant a given target at a certain probability and no other solution can grant the same target at a higher probability. Since the costs belong to a location-scale family the efficient solutions have a simple geometric characterization in terms of mean and variance. Computing efficient solutions can be carried out by an iteration scheme calling for the solution of a constrained integer programming problem.