Simulation of stochastic activity networks using path control variates

This article details several procedures for using path control variates to improve the accuracy of simulation-based point and confidence-interval estimators of the mean completion time of a stochastic activity network (SAN). Because each path control variate is the duration of the corresponding directed path in the network from the source to the sink, the vector of selected path controls has both a known mean and a known covariance matrix. This information is incorporated into estimation procedures for both normal and nonnormal responses. To evaluate the performance of these procedures experimentally, we examine the bias, variance, and mean square error of the controlled point estimators as well as the average half-length and coverage probability of the corresponding confidence-interval estimators for a set of SANs in which the following characteristics are systematically varied: (a) the size of the network (number of nodes and arcs); (b) the topology of the network; (c) the percentage of activities with exponentially distributed durations; and (d) the relative dominance of the critical path. The experimental results show that although large improvements in accuracy can be achieved with some of these procedures, the confidence-interval estimators for normal responses may suffer serious loss of coverage probability in some applications.     

Scheduling on a single machine with a single breakdown to minimize stochastically the number of tardy jobs

Jobs with known processing times and due dates have to be processed on a machine which is subject to a single breakdown. The moment of breakdown and the repair time are independent random variables. Two cases are distinguished with reference to the processing time preempted by the breakdown (no other preemptions are allowed): (i) resumption without time losses and (ii) restart from the beginning. Under certain compatible conditions, we find the policies which minimize stochastically the number of tardy jobs.     

Cumulative search-evasion games

Cumulative search-evasion games (CSEGs) are two-person zero-sum search-evasion games where play proceeds throughout some specified period without interim feedback to either of the two players. Each player moves according to a preselected plan. If (X_t, Y_t,) are the positions of the two players at time t, then the game's payoff is the sum over t from 1 to T of A(X_t, Y_t, t). Additionally, all paths must be “connected.” That is, the finite set of positions available for a player in any time period depends on the position selected by that player in the previous time period. One player attempts to select a mixed strategy over the feasible T-time period paths to maximize the expected payoff. The other minimizes. Two solution procedures are given. One uses the Brown-Robinson method of fictitious play and the other linear programming. An example problem is solved using both procedures.     

A proof for the longest-job-first policy in one-machine scheduling

We consider a one-machine scheduling problem with earliness and tardiness penalties. All jobs are assigned a common due date and the objective is to minimize the total penalty due to job earliness and tardiness. We are interested in finding the optimal combination of the common due-date value and the job sequence. Despite the fact that this problem in general is very hard to solve, we prove that there exists at least a common property for all optimal solutions: The first job in an optimal sequence is one of the longest jobs. We also prove that this property holds for a general class of unimodal penalty functions.     

Applications of renewal theory in analysis of the free-replacement warranty

Under a free-replacement warranty of duration W, the customer is provided, for an initial cost of C, as many replacement items as needed to provide service for a period W. Payments of C are not made at fixed intervals of length W, but in random cycles of length Y = W + γ(W), where γ(W) is the (random) remaining life-time of the item in service W time units after the beginning of a cycle. The expected number of payments over the life cycle, L, of the item is given by M_Y(L), the renewal function for the random variable Y. We investigate this renewal function analytically and numerically and compare the latter with known asymptotic results. The distribution of Y, and hence the renewal function, depends on the underlying failure distribution of the items. Several choices for this distribution, including the exponential, uniform, gamma and Weibull, are considered.     

Hyperbolic integer programming

The hyperbolic integer program is treated as a special case of a hyperbolic program with a finite number of feasible points. The continuous hyperbolic program also belongs to this class since its solution can be obtained by considering only the extreme points of the feasible set. A general algorithm for solving the hyperbolic integer program which reduces to solving a sequence of linear integer problems is proposed. When the integer restriction is removed, this algorithm is similar to the Isbell-Marlow procedure. The geometrical aspects of the hyperbolic problem are also discussed and several cutting plane algorithms are given.     

Multiechelon repairable-item provisioning in a time-varying environment using the randomization technique

Multiechelon repairable-item provisioning systems are considered under a time-varying environment. Such conditions could arise, for example, in a military context where a shift from peacetime operation to wartime operation takes place; or, in a civilian setting where a public transit system decides to increase its hours of operation or frequency of service. Exact Markovian models, incorporating a finite population of repairable components and limited repair capacity (nonample service), are treated, with transient solutions obtained using the randomization technique. The exact models are compared with the approximate Dyna-METRIC model which assumes an infinite population of components and ample repair capacity.     

Stochastically minimizing the makespan in flow shops

In this article, we are concerned with scheduling stochastic jobs in a flowshop with m machines and zero intermediate storage. We assume that there are n - 2 identically distributed and 2 fast stochastic jobs. Roughly, the main result states that the makespan is stochastically minimized by placing one of the fast jobs first and the other last.     

A new polynomial algorithm for maximum value flow with an efficient parallel implementation

A new algorithm is presented for finding maximal and maximum value flows in directed single-commodity networks. Commonly algorithms developed for this problem find a maximal flow by gradually augmenting (increasing) a feasible flow to a maximal flow. In the presented algorithm, at the beginning of each step or iteration, the flow on arcs is assigned to flow capacity. This may lead to an infeasible flow violating flow conservation at some nodes. During two passes of a MAIN step, consisting of a forward pass and a backward pass, the flow is reduced on some arcs to regain feasibility. The network is then pruned by omitting saturated arcs, and the process is repeated. The parallel implementation of the algorithm applies the two main steps at the same time to the same network. The outputs of the two steps are compared and the processing continues with the higher feasible flow. The algorithm is simple, intuitive, and efficient. © 1993 John Wiley & Sons, Inc.