Showing papers in "Annals of Operations Research in 1991"

A survey of algorithmic methods for partially observed Markov decision processes

Abstract: A partially observed Markov decision process (POMDP) is a generalization of a Markov decision process that allows for incomplete information regarding the state of the system. The significant applied potential for such processes remains largely unrealized, due to an historical lack of tractable solution methodologies. This paper reviews some of the current algorithmic alternatives for solving discrete-time, finite POMDPs over both finite and infinite horizons. The major impediment to exact solution is that, even with a finite set of internal system states, the set of possible information states is uncountably infinite. Finite algorithms are theoretically available for exact solution of the finite horizon problem, but these are computationally intractable for even modest-sized problems. Several approximation methodologies are reviewed that have the potential to generate computationally feasible, high precision solutions.

Asymptotic analysis of stochastic programs

Abstract: In this paper we discuss a general approach to studying asymptotic properties of statistical estimators in stochastic programming. The approach is based on an extended delta method and appears to be particularly suitable for deriving asymptotics of the optimal value of stochastic programs. Asymptotic analysis of the optimal value will be presented in detail. Asymptotic properties of the corresponding optimal solutions are briefly discussed.

Minimum concave-cost network flow problems: applications, complexity, and algorithms

Abstract: We discuss a wide range of results for minimum concave-cost network flow problems, including related applications, complexity issues, and solution techniques. Applications from production and inventory planning, and transportation and communication network design are discussed. New complexity results are proved which show that this problem is NP-hard for cases with cost functions other than fixed charge. An overview of solution techniques for this problem is presented, with some new results given regarding the implementation of a particular branch-and-bound approach.

A Survey of solution techniques for the partially observed Markov decision process

Abstract: We survey several computational procedures for the partially observed Markov decision process (POMDP) that have been developed since the Monahan survey was published in 1982. The POMDP generalizes the standard, completely observed Markov decision process by permitting the possibility that state observations may be noise-corrupted and/or costly. Several computational procedures presented are convergence accelerating variants of, or approximations to, the Smallwood-Sondik algorithm. Finite-memory suboptimal design results are reported, and new research directions involving heuristic search are discussed.

A practical use of Jackson's preemptive schedule for solving the job shop problem

Abstract: In this paper, we present a polynomial algorithm for optimally adjusting heads and tails in the job shop problem. This algorithm is based on Jackson's preemptive schedule for the one-machine problem. We next use this algorithm to construct a new branch and bound method. Computational results show the superiority of this method over the classical ones.

Topological design of telecommunication networks-local access design methods

Abstract: Computer communication networks and telecommunication systems are growing at an explosive rate. Some of the major factors influencing this phenomenal growth rate have been technology driven, deregulation of the telecommunication industry and the breakup of AT&T, product and service introductions and competition, new application areas, price reductions and improved services. Corporations have discovered how to use telecommunication-based systems and computer networks as a strategic competitive weapon. Modern computer networks consist of backbone networks which serve as major highways to transfer large volumes of communication traffic, and local access networks which feed traffic between the backbone network and end user nodes. The design of the local access network is a complex process which builds on many difficult combinatorial optimization problems. This paper surveys many of the problems, presents the state of the art in solving them, and demonstrates a variety of solution procedures. The paper concludes with a list of open problems and areas open for further investigation.

A theory of rolling horizon decision making

Abstract: In this paper, we develop a theoretical framework for the common business practice of rolling horizon decision making. The main idea of our approach is that the usefulness of rolling horizon methods is, to a great extent, implied by the fact that forecasting the future is a costly activity. We, therefore, consider a general, discrete-time, stochastic dynamic optimization problem in which the decision maker has the possibility to obtain information on the uncertain future at given cost. For this non-standard optimization problem with optimal stopping decisions, we develop a dynamic programming formulation. We treat both finite and infinite horizon cases. We also provide a careful interpretation of the dynamic programming equations and illustrate our results by a simple numerical example. Various generalizations are shown to be captured by straightforward modifications of our model.

A dynamic programming algorithm for preemptive scheduling of a single machine to minimize the number of late jobs

Abstract: The scheduling problem 1|pmtn, r j|Σw jU j calls forn jobs with arbitrary release dates and due dates to be preemptively scheduled for processing by a single machine, with the objective of minimizing the sum of the weights of the late jobs. A dynamic programming algorithm for this problem is described. Time and space bounds for the algorithm are, respectively,O(nk 2W 2) andO(k 2W), wherek is the number of distinct release dates andW is the sum of the integer job weights. Thus, for the problem 1|pmtn, r j|ΣU j, in which the objective is simply to minimize the number of late jobs, the pseudopolynomial time bound becomes polynomial, i.e.O(n 3k 2).

Analysis of relaxations for the multi-item capacitated lot-sizing problem

Abstract: The multi-item capacitated lot-sizing problem consists of determining the magnitude and the timing of some operations of durable results for several items in a finite number of processing periods so as to satisfy a known demand in each period. We show that the problem is strongly NP-hard. To explain why one of the most popular among exact and approximate solution methods uses a Lagrangian relaxation of the capacity constraints, we compare this approach with every alternate relaxation of the classical formulation of the problem, to show that it is the most precise in a rigorous sense. The linear relaxation of a shortest path formulation of the same problem has the same value, and one of its Lagrangian relaxations is even more accurate. It is comforting to note that well-known relaxation algorithms based on the traditional formulation can be directly used to solve the shortest path formulation efficiently, and can be further enhanced by new algorithms for the uncapacitated lot-sizing problem. An extensive computational comparison between linear programming, column generation and subgradient optimization exhibits this efficiency, with a surprisingly good performance of column generation. We pinpoint the importance of the data characteristics for an empirical classification of problem difficulty and show that most real-world problems are easier to solve than their randomly generated counterparts because of the presence of initial inventories and their large number of items.

Sensitivity and stability analysis for nonlinear programming

Abstract: We give a brief overview of important results in several areas of sensitivity and stability analysis for nonlinear programming, focusing initially on “qualitative” characterizations (e.g., continuity, differentiability and convexity) of the optimal value function. Subsequent results concern “quantitative” measures, in particular optimal value and solution point parameter derivative calculations, algorithmic approximations, and bounds. Our treatment is far from exhaustive and concentrates on results that hold for smooth well-structured problems.

Batch sizing and job sequencing on a single machine

Abstract: We study a single-machine scheduling problem in which the items to be processed have to be batched as well as sequenced. Since processed items become available in batches, flow times are defined to be the same for all items in the same batch. A constant set-up delay is incurred between consecutive batches. For any fixed, but arbitrary item sequence, we present an algorithm that finds a sequence of batches such that the total flow time of the items is minimized; we prove that for a set ofn items, the algorithm runs inO(n) time. We show that, among all sequences, the one leading to the minimum flow time has the items in non-decreasing order of running times. Thus, the optimal algorithm for the combined problem, called thebatch-sizing problem, runs inO(n logn) time. We also prove that this algorithm yields an improved solution to a scheduling problem recently studied by Baker [1].

Applying the progressive hedging algorithm to stochastic generalized networks

Abstract: The introduction of uncertainty to mathematical programs greatly increases the size of the resulting optimization problems. Specialized methods that exploit program structures and advances in computer technology promise to overcome the computational complexity of certain classes of stochastic programs. In this paper we examine the progressive hedging algorithm for solving multi-scenario generalized networks. We present computational results demonstrating the effect of various internal tactics on the algorithm's performance. Comparisons with alternative solution methods are provided.

Stability and sensitivity-analysis for stochastic programming

Abstract: Stability and sensitivity studies for stochastic programs have been motivated by the problem of incomplete information about the true probability measure through which the stochastic program is formulated and in connection with the development and evaluation of algorithms. The first part of this survey paper briefly introduces and compares different approaches and points out the contemporary efforts to remove and weaken assumptions that are not realistic (e.g., strict complementarity conditions). The second part surveys recent results on qualitative and quantitative stability with respect to the underlying probability measure and describes the ways and means of statistical sensitivity analysis based on Gâteaux derivatives. The last section comments on parallel statistical sensitivity results obtained in the parametric case, i.e., for probability measures belonging to a parametric family indexed by a finite dimensional vector parameter.

Approaches to sensitivity analysis in linear programming

Abstract: A continuing priority in sensitivity and parametric analysis is to develop approaches that provide useful information, that are easy for a decision-maker to use, and that are computationally practical. Herein we review approaches to sensitivity analysis in linear programming and discuss how they meet the above needs. Special emphasis is given to sensitivity analysis of the objective function coefficients.

Throughput rate optimization in the automated assembly of printed circuit boards

Abstract: The electronics industry relies heavily on numerically controlled machines for the placement of electronic components on the surface of printed circuit boards (PCB). These placement (or mounting, or pick-and-place) machines automatically insert components into PCB’s, in a sequence determined by the input program. The most recent among them are characterized by high levels of accuracy and speed, but their throughput rates still appear to be extremely sensitive to the quality of the instructions. On the other hand, the effective programming of the machines becomes steadily more difficult in view of the increasing sophistication of the available technology. The development of optimization procedures allowing the efficient operation of such placement machines therefore provides an exciting challenge for the operations research community, as witnessed by, e.g., the recent papers by Ahmadi, Grotzinger and Johnson (1988), Ball and Magazine (1988), and Leipala and Nevalainen (1989).

Stability analysis for stochastic programs

Abstract: For stochastic programs with recourse and with (several joint) probabilistic constraints, respectively, we derive quantitative continuity properties of the relevant expectation functionals and constraint set mappings. This leads to qualitative and quantitative stability results for optimal values and optimal solutions with respect to perturbations of the underlying probability distributions. Earlier stability results for stochastic programs with recourse and for those with probabilistic constraints are refined and extended, respectively. Emphasis is placed on equipping sets of probability measures with metrics that one can handle in specific situations. To illustrate the general stability results we present possible consequences when estimating the original probability measure via empirical ones.

The max-cut problem and quadratic 0-1 optimization; polyhedral aspects, relaxations and bounds

Abstract: Given a graphG, themaximum cut problem consists of finding the subsetS of vertices such that the number of edges having exactly one endpoint inS is as large as possible. In the weighted version of this problem there are given real weights on the edges ofG, and the objective is to maximize the sum of the weights of the edges having exactly one endpoint in the subsetS. In this paper, we consider the maximum cut problem and some related problems, likemaximum-2-satisfiability, weighted signed graph balancing. We describe the relation of these problems to the unconstrained quadratic 0–1 programming problem, and we survey the known methods for lower and upper bounds to this optimization problem. We also give the relation between the related polyhedra, and we describe some of the known and some new classes of facets for them.

Recent trends in random number and random vector generation

Abstract: A survey of recent work in the areas of uniform pseudorandom number and uniform pseudorandom vector generation is presented. The emphasis is on methods for which a detailed theory is available. A progress report on the construction of quasirandom points for efficient multidimensional numerical integration is also given.

Minimizing the sum of the job completion times in the two-machine flow shop by Lagrangian relaxation

Abstract: A branch-and-bound algorithm is presented for the two-machine flow shop problem with the objective of minimizing the sum of the job completion times. Lower bounds and precedence constraints result from a Lagrangian relaxation of this problem. The Lagrangian subproblem turns out to be a linear ordering problem that is polynomially solvable for appropriate choices of the Lagrangian multipliers. The best choice within this class yields a lower bound that dominates previous bounds. In fact, the existing bounds correspond to particular choices of the multipliers. Several dominance criteria are given to restrict the search tree. Computational experiments show that the proposed algorithm outperforms the previously best method.

Computational experience with an interior point algorithm on the satisfiability problem

Abstract: We apply the zero-one integer programming algorithm described in Karmarkar [12] and Karmarkar, Resende and Ramakrishnan [13] to solve randomly generated instances of the satisfiability problem (SAT). The interior point algorithm is briefly reviewed and shown to be easily adapted to solve large instances of SAT. Hundreds of instances of SAT (having from 100 to 1000 variables and 100 to 32,000 clauses) are randomly generated and solved. For comparison, we attempt to solve the problems via linear programming relaxation with MINOS.

On the average cost optimality equation and the structure of optimal policies for partially observable Markov decision processes

Abstract: We consider partially observable Markov decision processes with finite or countably infinite (core) state and observation spaces and finite action set. Following a standard approach, an equivalent completely observed problem is formulated, with the same finite action set but with anuncountable state space, namely the space of probability distributions on the original core state space. By developing a suitable theoretical framework, it is shown that some characteristics induced in the original problem due to the countability of the spaces involved are reflected onto the equivalent problem. Sufficient conditions are then derived for solutions to the average cost optimality equation to exist. We illustrate these results in the context of machine replacement problems. Structural properties for average cost optimal policies are obtained for a two state replacement problem; these are similar to results available for discount optimal policies. The set of assumptions used compares favorably to others currently available.

Generalized linear multiplicative and fractional programming

Abstract: This paper addresses itself to the algorithm for minimizing the sum of a convex function and a product of two linear functions over a polytope. It is shown that this nonconvex minimization problem can be solved by solving a sequence of convex programming problems. The basic idea of this algorithm is to embed the original problem into a problem in higher dimension and apply a parametric programming (path following) approach. Also it is shown that the same idea can be applied to a generalized linear fractional programming problem whose objective function is the sum of a convex function and a linear fractional function.

Approximate scenario solutions in the progressive hedging algorithm: a numerical study

Abstract: This paper describes how the scenario aggregation principle can be combined with approximate solutions of the individual scenario problems, resulting in a computationally efficient algorithm where two individual Lagrangian-based procedures are merged into one. Computational results are given for an example from fisheries management. Numerical experiments indicate that only crude scenario solutions are needed.

Recurrence conditions for Markov decision processes with Borel state space: a survey

Abstract: This paper describes virtually all the recurrence conditions used heretofore for Markov decision processes with Borel state and action spaces, which include some forms of mixing and contraction properties, Doeblin's condition, Harris recurrence, strong ergodicity, and the existence of bounded solutions to the optimality equation for average reward processes The aim is to establish (when possible) implications and equivalences between these conditions

A variational approach to the Steiner network problem

Abstract: Supposen points are given in the plane. Their coordinates form a 2n-vectorX. To study the question of finding the shortest Steiner networkS connecting these points, we allowX to vary over a configuration space. In particular, the Steiner ratio conjecture is well suited to this approach and short proofs of the casesn=4, 5 are discussed. The variational approach was used by us to solve other cases of the ratio conjecture (n=6, see [11] and for arbitraryn points lying on a circle). Recently, Du and Hwang have given a beautiful complete solution of the ratio conjecture, also using a configuration space approach but with convexity as the major idea. We have also solved Graham's problem to decide when the Steiner network is the same as the minimal spanning tree, for points on a circle and on any convex polygon, again using the variational method.

Statistical verification of optimality conditions for stochastic programs with recourse

Abstract: Statistically motivated algorithms for the solution of stochastic programming problems typically suffer from their inability to recognize optimality of a given solution algorithmically. Thus, the quality of solutions provided by such methods is difficult to ascertain. In this paper, we develop methods for verification of optimality conditions within the framework of Stochastic Decomposition (SD) algorithms for two stage linear programs with recourse. Consistent with the stochastic nature of an SD algorithm, we provide termination criteria that are based on statistical verification of traditional (deterministic) optimality conditions. We propose the use of “bootstrap methods” to confirm the satisfaction of generalized Kuhn-Tucker conditions and conditions based on Lagrange duality. These methods are illustrated in the context of a power generation planning model, and the results are encouraging.

On parametric nonlinear programming

Abstract: In this tutorial survey we study finite dimensional optimization problems which depend on parameters. It is our aim to work out several basic connections with different mathematical areas. In particular, attention will be paid to unfolding and singularity theory, structural analysis of families of constraint sets, constrained optimization problems and semi-infinite optimization.

Approximation results in parallel machines stochastic scheduling

Abstract: We consider scheduling a batch of jobs with stochastic processing times on parallel machines. We derive various new formulae for the expected flowtime and weighted flowtime under general scheduling rules. Smith's Rule, which orders job starts by decreasing ratio of weight to expected processing time provides a natural heuristic for this problem. We obtain a bound on the worst case difference between the expected weighted flow time under Smith's Rule and under an optimal policy. For a wide class of processing time distributions, this bound is of oderO(1) and does not increase with the number of jobs.

Parametric methods in integer linear programming

Abstract: In contrast to methods of parametric linear programming which were developed soon after the invention of the simplex algorithm and are easily included as an extension of that method, techniques for parametric analysis on integer programs are not well known and require considerable effort to append them to an integer programming solution algorithm.