Single‐machine scheduling with deadlines to minimize the total weighted late work

We consider scheduling a set of jobs with deadlines to minimize the total weighted late work on a single machine, where the late work of a job is the amount of processing of the job that is scheduled after its due date and before its deadline. This is the first study on scheduling with the late work criterion under the deadline restriction. In this paper, we show that (i) the problem is unary NP‐hard even if all the jobs have a unit weight, (ii) the problem is binary NP‐hard and admits a pseudo‐polynomial‐time algorithm and a fully polynomial‐time approximation scheme if all the jobs have a common due date, and (iii) some special cases of the problem are polynomially solvable.     

Suppression of enemy air defenses (SEAD) as an information duel

Blue strike aircraft enter region ? to attack Red targets. In Case 1, Blue conducts (preplanned) SEAD to establish air superiority. In the (reactive) SEAD scenario, which is Case 2, such superiority is already in place, but is jeopardized by prohibitive interference from Red, which threatens Blue's ability to conduct missions. We utilize both deterministic and stochastic models to explore optimal tactics for Red in such engagements. Policies are developed which will guide both Red's determination of the modes of operation of his engagement radar, and his choice of Blue opponent to target next. An index in the form of a simple transaction kill ratio plays a major role throughout. Published 2002 Wiley Periodicals, Inc. Naval Research Logistics 49: 723–742, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/nav.10046     

Extension of Bolzano search to rectangles which preserves rectangles as iterates

Multivariable elimination algorithms, which may be regarded as generalizations of various one-dimensional search procedures, have not found wide application. A probable reason may be the generally very irregularly shaped regions of uncertainty that evolve iteratively in the procedures. Hence hope for practical salvage of this class of algorithms seems to lie in controlling the shape of the successively smaller regions of uncertainty. In this article an extension of Bolzano search to rectangles which preserves rectangular iterates is given. Since the result is essentially geometrical in nature, a geometric proof of this procedure is given. Hopefully the proof procedure will be of independent interest. A numerical illustration of the procedure is given for a game problem, such problems lending themselves to this method.     

A flow-shop problem with sequence-dependent additive setup times

The flow-shop scheduling problem with sequence-dependent additive setup times is considered as a special case of the general problem, and a polynomially bounded approximate method is developed to find a minimum makespan permutation schedule. The approximate algorithm is shown to yield optimal results for the two-machine case. A version of Sule's model is defined that produces the first approximation of the optimal solution for this problem. Computational experience along with numerical examples are provided to test the effectiveness of the method.     

A decision support system for the procurement of military equipment

The use of life-cycle costs in procurement is limited by the accuracy of cost estimates and assessments of availability risks under the conditions generating costs. Critical to these problems are the needs and responses of equipment to maintenance and repair. Using transition probabilities to define the relationships between serviceability and maintenance and repairs, this article develops a comprehensive decision support system for military procurement. It uses a dynamic programming model to determine the least-cost set of maintenance and repair decisions, where adjustments are included for warranty conditions, time value of money, the opportunity costs of equipment failure, and the end-of-cycle salvage values or disposal costs. The system allows users to define working and failed states, create their own rules or indicators of availability from estimates of state probabilities and establish thresholds of risk acceptability.     

An assumption-free convergence analysis for a perturbation of the scaling algorithm for linear programs,with application to the L1 estimation problem

This article is concerned with the scaling variant of Karmarkar's algorithm for linear programming problems. Several researchers have presented convergence analyses for this algorithm under various nondegeneracy types of assumptions, or under assumptions regarding the nature of the sequence of iterates generated by the algorithm. By employing a slight perturbation of the algorithm, which is computationally imperceptible, we are able to prove without using any special assumptions that the algorithm converges finitely to an ε-optimal solution for any chosen ε > 0, from which it can be (polynomically) rounded to an optimum, for ε > 0 small enough. The logarithmic barrier function is used as a construct for this analysis. A rounding scheme which produces an optimal extreme point solution is also suggested. Besides the non-negatively constrained case, we also present a convergence analysis for the case of bounded variables. An application in statistics to the L₁ estimation problem and related computational results are presented.