The robust inventory control problem was proposed and solved by Bertsimas and Thiele (2006). Their results are very interesting in that the problem can be solved easily and also the solution possesses nice properties of those found in the traditional stochastic inventory control problem. However, their formulation is shown to be incorrect, which invalidates all of the results given there. In this paper, we propose an alternative formulation of the problem which uses a different but practically applicable uncertainty set. Under the newly proposed model, all of the useful properties given in Bertsimas and Thiele (2006) will be shown to be valid.

Content Delivery Network (CDN) has evolved to overcome a network bottleneck and improve user perceived Quality of Service (QoS). A CDN replicates contents from the origin server to replica servers to reduce the overload of the origin server. CDN providers would try to achieve an acceptable performance at the least cost including the storage space or processing power. In this paper, we introduce a new optimization model for the CDN design problem considering the user perceived QoS and single path (non-bifurcated) routing constraints and analyze the computational complexity for some special cases.

We investigate an optimal retirement time and consumption/investment policy of a wage earner who expects to find a better investment opportunity after retirement by being freed from other work and participating fully in the financial market. We obtain a closed form solution to the optimization problem by using a dynamic programming method under general time-separable von Neumann-Morgenstern utility. It is optimal for the wage earner to retire from work if and only if his wealth exceeds a certain critical level which is obtained from a free boundary value problem. The wage earner consumes less and takes more risk than he would without anticipation of a better investment opportunity.

This paper studies the assortment optimization problem in online retailing by using a multinomial logit model in order to take consumer choice behavior into account. We focus on two unique features of online purchase behavior: first, there exists increased amount of uncertainty (e.g., size and color of merchandize) in online shopping as customers cannot experience merchandize directly. This uncertainty is captured by the scale parameter of a Gumbel distribution; second, online shopping entails unique shopping-related disutility (e.g., waiting time for delivery and security concerns) compared to offline shopping. This disutility is controlled by the changes in the observed part of utility function in our model. The impact of changes in uncertainty and disutility on the expected profit does not exhibit obvious structure: the expected profit may increase or decrease depending on the assortment. However, by analyzing the structure of the optimal assortment based on convexity property of the profit function, we show that the cardinality of the optimal assortment decreases and the maximum expected profit increases as uncertainty or disutility decreases. Therefore, our study suggests that it is important for managers of online retailing to reduce uncertainty and disutility involved in online purchase process.

A flow shop scheduling problem involves scheduling jobs on multiple machines in series in order to optimize a given criterion. The flow time of a job is the amount of time the job spent before its completion and the stretch of the job is the ratio of its flow time to its processing time. In this paper, a hybrid genetic algorithm (HGA) approach is proposed for minimizing the total stretch in flow shop scheduling. HGA adopts the idea of seed selection and development in order to reduce the chance of premature convergence that may cause the loss of search power. The performance of HGA is compared with that of genetic algorithms (GAs).