Improving Sample Average Approximation Using Distributional Robustness

Sample average approximation is a popular approach to solving stochastic optimization problems. It has been widely observed that some form of robustification of these problems often improves the out-of-sample performance of the solution estimators. In estimation problems, this improvement boils down to a trade-off between the opposing effects of bias and shrinkage. This paper aims to characterize the features of more general optimization problems that exhibit this behaviour when a distributionally robust version of the sample average approximation problem is used. The paper restricts attention to quadratic problems for which sample average approximation solutions are unbiased and shows that expected out-of-sample performance can be calculated for small amounts of robustification and depends on the type of distributionally robust model used and properties of the underlying ground-truth probability distribution of random variables. The paper was written as part of a New Zealand funded research project that aimed to improve stochastic optimization methods in the electric power industry. The authors of the paper have worked together in this domain for the past 25 years.

Research on Emergency Material Allocation Mode of Sudden Disaster with Improved NSGA-II

Background:In order to solve the problems of redundancy, unfairness, low satisfaction and high cost of emergency material allocation caused by unreasonable allocation effectively in the case of sudden disasters, and minimize the economic cost, punishment cost and maximizing the satisfaction rate of disaster victims, a 3-level network emergency material allocation mode based on big data is proposed in this paper.Methods:Taking the loss degree and the dynamic change of material demand in the disaster stricken areas as constraints, the demand forecasting, scheduling optimization, targeted allocation and disaster victims' satisfaction model based on emergency relief materials is constructed. The Sample Average Approximation method and improved NSGA-II algorithm are designed to solve the problem.Results:Compared with the results obtained by the improved NSGA-II, the value is significantly reduced. From the fairness evaluation results of the two model distribution schemes, the model obtained by the improved NSGA-II is more suitable for the distribution of emergency supplies with fair distribution requirements.Conclusions:It can be concluded that the 3-level network allocation mode and improved NSGA-II can solve emergency relief materials allocation based on big data effectively. The next step is to design scheduling model with all feasible medical supplies allocation route to improve the practicability of the model.

Online Linear Programming: Dual Convergence, New Algorithms, and Regret Bounds

We study an online linear programming (OLP) problem under a random input model in which the columns of the constraint matrix along with the corresponding coefficients in the objective function are independently and identically drawn from an unknown distribution and revealed sequentially over time. Virtually all existing online algorithms were based on learning the dual optimal solutions/prices of the linear programs (LPs), and their analyses were focused on the aggregate objective value and solving the packing LP, where all coefficients in the constraint matrix and objective are nonnegative. However, two major open questions were as follows. (i) Does the set of LP optimal dual prices learned in the existing algorithms converge to those of the “offline” LP? (ii) Could the results be extended to general LP problems where the coefficients can be either positive or negative? We resolve these two questions by establishing convergence results for the dual prices under moderate regularity conditions for general LP problems. Specifically, we identify an equivalent form of the dual problem that relates the dual LP with a sample average approximation to a stochastic program. Furthermore, we propose a new type of OLP algorithm, action-history-dependent learning algorithm, which improves the previous algorithm performances by taking into account the past input data and the past decisions/actions. We derive an [Formula: see text] regret bound (under a locally strong convexity and smoothness condition) for the proposed algorithm, against the [Formula: see text] bound for typical dual-price learning algorithms, where n is the number of decision variables. Numerical experiments demonstrate the effectiveness of the proposed algorithm and the action-history-dependent design.

Asymptotic Analysis for a Stochastic Second-Order Cone Programming and Applications

Stochastic second-order cone programming (SSOCP) is an extension of deterministic second-order cone programming, which demonstrates underlying uncertainties in practical problems arising in economics engineering and operations management. In this paper, asymptotic analysis of sample average approximation estimator for SSOCP is established. Conditions ensuring the asymptotic normality of sample average approximation estimators for SSOCP are obtained and the corresponding covariance matrix is described in a closed form. Based on the analysis, the method to estimate the confidence region of a stationary point of SSOCP is provided and three examples are illustrated to show the applications of the method.

Robust Stochastic Models for Profit-Maximizing Hub Location Problems

This paper introduces robust stochastic models for profit -maximizing capacitated hub location problems in which two different types of uncertainty, including stochastic demand and uncertain revenue, are simultaneously incorporated into the problem. First, a two-stage stochastic program is presented in which demand and revenue are jointly stochastic. Next, robust stochastic models are developed to better model uncertainty in the revenue while keeping the demand stochastic. Two particular cases are studied based on the dependency between demand and revenue. In the first case, a robust stochastic model with a min-max regret objective is developed assuming a finite set of scenarios that describes uncertainty associated with the revenue under a revenue-elastic demand setting. For the case when demand and revenue are independent, robust stochastic models with a max-min criterion and a min-max regret objective are formulated considering both interval uncertainty and discrete scenarios, respectively. It is proved that the robust stochastic version with max-min criterion can be viewed as a special case of the min-max regret stochastic model. Exact algorithms based on Benders decomposition coupled with a sample average approximation scheme are proposed. Exploiting the repetitive nature of sample average approximation, generic acceleration methodologies are developed to enhance the performance of the algorithms enabling them to solve large-scale intractable instances. Extensive computational experiments are performed to consider the efficiency of the proposed algorithms and also to analyze the effects of uncertainty under different settings. The qualities of the solutions obtained from different modeling approaches are compared under various parameter settings. Computational results justify the need to solve robust stochastic models to embed uncertainty in decision making to design resilient hub networks.