Mathematics of Operations Research

This paper considers the problem faced by a seller who has a single object to sell to one of several possible buyers, when the seller has imperfect information about how much the buyers might be willing to pay for the object. The seller's problem is to design an auction game which has a Nash equilibrium giving him the highest possible expected utility. Optimal auctions are derived in this paper for a wide class of auction design problems.

Let A be a binary matrix of size m × n, let c^T be a positive row vector of length n and let e be the column vector, all of whose m components are ones. The set-covering problem is to minimize c^Tx subject to Ax ≥ e and x binary. We compare the value of the objective function at a feasible solution found by a simple greedy heuristic to the true optimum. It turns out that the ratio between the two grows at most logarithmically in the largest column sum of A. When all the components of c^T are the same, our result reduces to a theorem established previously by Johnson and Lovasz.

We study convex optimization problems for which the data is not specified exactly and it is only known to belong to a given uncertainty set U, yet the constraints must hold for all possible values of the data from U. The ensuing optimization problem is called robust optimization. In this paper we lay the foundation of robust convex optimization. In the main part of the paper we show that if U is an ellipsoidal uncertainty set, then for some of the most important generic convex optimization problems (linear programming, quadratically constrained programming, semidefinite programming and others) the corresponding robust convex program is either exactly, or approximately, a tractable problem which lends itself to efficient algorithms such as polynomial time interior point methods.

NP-complete problems form an extensive equivalence class of combinatorial problems for which no nonenumerative algorithms are known. Our first result shows that determining a shortest-length schedule in an m-machine flowshop is NP-complete for m ≥ 3. (For m = 2, there is an efficient algorithm for finding such schedules.) The second result shows that determining a minimum mean-flow-time schedule in an m-machine flowshop is NP-complete for every m ≥ 2. Finally we show that the shortest-length schedule problem for an m-machine jobshop is NP-complete for every m ≥ 2. Our results are strong in that they hold whether the problem size is measured by number of tasks, number of bits required to express the task lengths, or by the sum of the task lengths.

The theory of the proximal point algorithm for maximal monotone operators is applied to three algorithms for solving convex programs, one of which has not previously been formulated. Rate-of-convergence results for the “method of multipliers,” of the strong sort already known, are derived in a generalized form relevant also to problems beyond the compass of the standard second-order conditions for oplimality. The new algorithm, the “proximal method of multipliers,” is shown to have much the same convergence properties, but with some potential advantages.

The problem of minimizing the total tardiness for a set of independent jobs on one machine is considered. Lawler has given a pseudo-polynomial-time algorithm to solve this problem. In spite of extensive research efforts for more than a decade, the question of whether it can be solved in polynomial time or it is NP-hard (in the ordinary sense) remained open. In this paper the problem is shown to be NP-hard (in the ordinary sense).

A real-valued function z whose domain is all of the subsets of N = {1, …, n) is said to be submodular if z(S) + z(T) ≥ z(S ∪ T) + z(S ∩ T), ∀S, T ⊆ N, and nondecreasing if z(S) ≤ z(T), ∀S ⊂ T ⊆ N. We consider the problem max_S⊂N {z(S): |S| ≤ K, z submodular and nondecreasing, z(Ø) = 0}.

Many combinatorial optimization problems can be posed in this framework. For example, a well-known location problem and the maximization of certain boolean polynomials are in this class.

We present a family of algorithms that involve the partial enumeration of all sets of cardinality q and then a greedy selection of the remaining elements, q = 0, …, K − 1. For fixed K, the qth member of this family requires O(n^q+1) computations and is guaranteed to achieve at least [Formula: see text] Our main result is that this is the best performance guarantee that can be obtained by any algorithm whose number of computations does not exceed O(n^q+1).

In this paper we propose a robust formulation for discrete time dynamic programming (DP). The objective of the robust formulation is to systematically mitigate the sensitivity of the DP optimal policy to ambiguity in the underlying transition probabilities. The ambiguity is modeled by associating a set of conditional measures with each state-action pair. Consequently, in the robust formulation each policy has a set of measures associated with it. We prove that when this set of measures has a certain “rectangularity” property, all of the main results for finite and infinite horizon DP extend to natural robust counterparts. We discuss techniques from Nilim and El Ghaoui [17] for constructing suitable sets of conditional measures that allow one to efficiently solve for the optimal robust policy. We also show that robust DP is equivalent to stochastic zero-sum games with perfect information.

Sherali and Adams (1990), Lovász and Schrijver (1991) and, recently, Lasserre (2001b) have constructed hierarchies of successive linear or semidefinite relaxations of a 0–1 polytope P ⫅ ℝⁿ converging to P in n steps. Lasserre's approach uses results about representations of positive polynomials as sums of squares and the dual theory of moments. We present the three methods in a common elementary framework and show that the Lasserre construction provides the tightest relaxations of P. As an application this gives a direct simple proof for the convergence of the Lasserre's hierarchy. We describe applications to the stable set polytope and to the cut polytope.

We develop a new approach to solving minimum-cost circulation problems. Our approach combines methods for solving the maximum flow problem with successive approximation techniques based on cost scaling. We measure the accuracy of a solution by the amount that the complementary slackness conditions are violated.

We propose a simple minimum-cost circulation algorithm, one version of which runs in O(n³log(nC)) time on an n-vertex network with integer arc costs of absolute value at most C. By incorporating sophisticated data structures into the algorithm, we obtain a time bound of O(nm log(n²/m)log(nC)) on a network with m arcs. A slightly different use of our approach shows that a minimum-cost circulation can be computed by solving a sequence of O(n log(nC)) blocking flow problems. A corollary of this result is an O(n²(log n)log(nC))-time, m-processor parallel minimum-cost circulation algorithm. Our approach also yields strongly polynomial minimum-cost circulation algorithms.

Our results provide evidence that the minimum-cost circulation problem is not much harder than the maximum flow problem. We believe that a suitable implementation of our method will perform extremely well in practice.