This paper presents a robust canonical duality–triality theory for solving nonconvex programming problems under data uncertainty. This theory includes a robust canonical saddle-point theorem and robust canonical optimality conditions, which can be used to identify both robust global and local extrema of the primal problem. Two numerical examples are presented to illustrate that the robust Triality theory is particularly powerful for solving nonconvex optimization problems with data uncertainty.

In this paper we study the Markov-modulated M/M/ $$\infty $$ queue, with a focus on the correlation structure of the number of jobs in the system. The main results describe the system’s asymptotic behavior under a particular scaling of the model parameters in terms of a functional central limit theorem. More specifically, relying on the martingale central limit theorem, this result is established, covering the situation in which the arrival rates are sped up by a factor N and the transition rates of the background process by $$N^\alpha $$ , for some $$\alpha >0$$ . The results reveal an interesting dichotomy, with crucially different behavior for $$\alpha >1$$ and $$\alpha <1$$ , respectively. The limiting Gaussian process, which is of the Ornstein–Uhlenbeck type, is explicitly identified, and it is shown to be in accordance with explicit results on the mean, variances and covariances of the number of jobs in the system.

We consider a class of exit time stochastic control problems for diffusion processes with discounted criterion, where the controller can utilize a given amount of resource, called “fuel”. In contrast to the vast majority of existing literature, concerning the “finite fuel” problems, it is assumed that the intensity of fuel consumption is bounded. We characterize the value function of the optimization problem as the unique continuous viscosity solution of the Dirichlet boundary value problem for the correspondent Hamilton–Jacobi–Bellman (HJB) equation. Our assumptions concern the HJB equations, related to the problems with infinite fuel and without fuel. Also, we present computer experiments for the problems of optimal correction and optimal tracking of a simple stochastic system with the stable or unstable equilibrium point.

We study \(G/GI/N(+GI)\) queues with alternating renewal service interruptions in the Halfin–Whitt regime. The systems experience up and down alternating periods. In the up periods, the systems operate normally as the usual \(G/GI/N(+GI)\) queues with non-idling first-come–first-served service discipline. In the down periods, arrivals continue entering the systems, but all servers stop functioning while the amount of service that each customer has received will be conserved and services will resume when the next up period starts. For models with abandonment, interruptions do not affect customers’ patience times. We assume that the up periods are of the same order as the service times but the down periods are asymptotically negligible compared with the service times. We establish the functional central limit theorems for the queue-length processes and the virtual-waiting time processes in these models, where the limit processes are represented as stochastic integral convolution equations driven by jump processes. The convergence in these limit theorems is proved in the space \({\mathbb D}\) endowed with the Skorohod \(M_1\) topology.

The following paper describes a method for solving certain non-convex minimization problems. These problems contain properties where the feasible region may be represented as a union of convex cones. This set of cones is generated by the intersection of finitely many half-spaces passing through a defined origin. Globally the problem is non-convex, but it is locally convex within the interior of the generated cones whereas the objective function tends to infinity elsewhere. The following algorithm results in a polyhedral representation of all of the cones and an interior point in each of them. In order to solve the problem globally, a convex optimization algorithm is required to be initialized at each interior point and to converge to the local minimum over the corresponding cone. The best of these solutions is the global optimum. A study of the time complexity is presented herewith along with a detailed computational analysis.

A method of convex combined expectations of the least absolute deviation and least squares about the so-called regularized gap function is proposed for solving nonlinear stochastic variational inequality problems (for short, NSVIP). The NSVIP is formulated as a weighted expected residual minimization problem (in short, WERM) in this way. Moreover, we present a discrete approximation of WERM problem by applying the quasi-Monte Carlo method when the sample space is compact, and a compact approximation approach for the case that the sample space is noncompact. The limiting behaviors of optimal solutions of the discrete approximation problem and the compact approximation are also analyzed, respectively.

In this paper, a sequential approach, recently addressed by Mueller-Gritschneder et al. [SIAM J Optim 20(2):915–934, 2009] is studied. After a brief review, some weaknesses of the mentioned approach are pointed out. These weaknesses are due to a redundant assumption and the inability to generate the entire Pareto front. We present two improvements for the mentioned approach, firstly by eliminating the assumption of equality between the Pareto front and the weak Pareto front and secondly by presenting a new definition of the boundary of the Pareto front and establishing some theoretical results about it.

We propose an exact algorithm to find the optimal solution to the grey pattern problem which is a special instance of the quadratic assignment problem. A very effective branch and bound algorithm is constructed. The special structure of the problem is exploited and optimal solutions for many problem instances established. One instance based on 64 facilities, organized in an 8 by 8 pattern, that was extensively researched in the literature was optimally solved in 15 s of computer time while existing papers report solution times of hours.

Convoy movement planning problems arise in a number of important logistical contexts, including military planning, railroad optimization and automated guided vehicle routing. In the convoy movement problem (CMP), a set of convoys, with specified origins and destinations, are to be routed with the objective of minimizing either makespan or total flow time, while observing a number of side constraints. This paper characterizes the computational complexity of several restricted classes of CMPs. The principal objective is to identify the most parsimonious set of problem features that make the CMP intractable. A polynomial-time algorithm is provided for the single criterion two-convoy movement problem. The performance of a simple prioritization–idling approximation algorithm is also analyzed for the K-convoy movement problem. Error bounds are developed for the makespan and flow time objectives.

We consider a discrete-time risk model with delayed claims and a dividend payment strategy. It is assumed that every main claim will induce a by-claim which may be delayed for one time period with a certain probability. In the evaluation of the expected present value of dividends, the interest rates are assumed to follow a Markov chain with finite state space. Dividends are paid to the shareholders according to an admissible strategy. The company controls the amount of dividends in order to maximize the cumulative expected discounted dividends prior to ruin minus the expected discounted penalty value at ruin. We obtain some properties of the optimal dividend-payment strategy, and offer high efficiency algorithms for obtaining the optimal strategy, the optimal value function and the expectation of time of ruin under the optimal strategy. Our method is mainly to transform the value function. Numerical examples are presented to illustrate the transformation method and the impact of the penalty for ruin on the optimal strategy, the corresponding expected present value of dividends and the expectation of time of ruin.