INFORMS 2021 Program Book

INFORMS Anaheim 2021

TB40

2 - Entrepreneurial Mindset and Behavior for Product I ntroduction Decisions

information about the actual demand is revealed. We introduce the stochastic k- adaptable assignment balancing problem that generates k assignment options apriori with the objective of minimizing the maximum expected workload assigned to any sorter. The goal is to enable decision makers to adapt their operations to a plan that works best under the realized demand while maintaining a good level of consistency and stability in operations desired in practice. We compare exact and heuristic solution approaches and test them on real data obtained from a large parcel carrier. TB36 CC Room 210B In Person: Emerging Logistics Models General Session Chair: Mohammad Moshref-Javadi, Northeastern University, Boston, MA, 02115-5005, United States 1 - Fleet Resupply by Drones for Last-mile Delivery Juan C. Pina-Pardo, Pontificia Universidad Catolica de Valparaiso, Avenida Brasil 2241, Valparaiso, 2340000, Chile, Daniel F. Silva, Alice E. Smith, Ricardo Gatica This presentation introduces the Vehicle Routing Problem with Drone Resupply, which consists of finding a set of routes for a fleet of vehicles to deliver orders that become available throughout the day. Once the vehicles start their delivery routes, they do not need to return to the depot to collect newly released orders, as these orders are sent to them via drones. Assuming that the orders’ release times are known at the beginning of the day, we present a MILP formulation and an efficient two-stage heuristic approach for solving realistic-sized instances. The effects of depot location, customer distribution, drone capacity, and spread of orders’ release dates are investigated. 2 - Locational Pricing of Metro Mobility Services Yanchao Liu, Wayne State University, Detroit, MI, 48201 United States Mobility supply and demand are unevenly distributed across time and space, and they often do not match in quantity. Dynamic pricing has been used to restore the balance, but are also challenged by equity concerns. In this talk, we present a novel mathematical model to characterize the market dynamics and propose algorithms to compute the market equilibria and the fair fares. 3 - Drone Logistics for Uncertain Demands of Disaster-impacted Populations In this study, we present a stochastic optimization model to address the challenges associated with timely delivery of aid packages to disaster-affected regions via a fleet of drones while considering the set of demand locations is unknown. The main problem is to locate a set of drone platforms such that with a given probability, the maximum total cost (or disutility) under all realizations of the set of demand locations is minimized. We formulate and solve a time-space drone scheduling model for a set of scenarios to build up the total disutility distribution. We also propose an algorithmic solution approach which decomposes the problem into three tractable subproblems. TB37 CC Room 210C In Person: Technology, Innovation Management and Entrepreneurship General Session Chair: Sreekumar R. Bhaskaran, Southern Methodist University, Dallas, TX, 75275-0333, United States 1 - Where to Pop-up? Channel Operation Strategies under Price Harmonization Arunima Chhikara, University of Kansas, Lawrence, KS, United States, Avinash Geda, Nazli Turken, Janice E. Carrillo Price harmonization across different channels is a widely practiced marketing strategy. Contrary to the intuition that dual-channel firms utilize both channels under channel-specific pricing strategy, we find conditions when a single (online/offline) channel dominates the dual-channel policy under the price harmonization strategy. We find that for the price harmonization strategy, the optimal channel selection, and the optimal prices depend on market sizes, on- hand inventory, and salvage value. Our results are important to channel managers’ coordinated decisions when offering a product in their respective channels to optimize the overall profits at the retailer level. Zabih Ghelichi, University of Louisville, Louisville, KY, United States, Monica Gentili, Pitu B. Mirchandani

Sinan Erzurumlu, Babson College, Babson Park, MA, 02457, United States, Sreekumar R. Bhaskaran, Karthik Ramachandran Firms often face a choice between developing a risky, advanced, product and launching an on-hand product. While launching the on-hand product might bring much needed revenues, it could affect the profitability of the advanced product under development depending on the consumer experience with the launched on-hand product. We present evidence from behavioral studies that study how product managers in established and startup firms make these decisions. We particularly examine the impact of cash constraint, available options and trade-off; our findings reveal insights on how to position project continuation with respect to cash on hand, value of the project and the leanness of development process. TB40 CC Room 211B In Person: Recent Developments in Semidefinite Programming General Session Chair: Alex Wang, Carnegie Mellon University, Pittsburgh, United States 1 - Approximating Sparse Semidefinite Programs Kevin Shu, Georgia Institute of Technology, Atlanta, GA, United States It is well understood how to solve sparse semidefinite programs when the sparsity pattern corresponds to a chordal graph. We extend these results by analyzing a relaxation of the PSD cone defined for any graph, which we call the locally-PSD cone. We introduce a numerical invariant of a graph, which we call the additive distance, measuring how well the locally-PSD relaxation approximates the PSD cone. We then give bounds on the additive distance for a wide range of graphs, and show that in many cases, the approximation ratio for the relaxed program is bounded from above by 1+n/g^3, where n is the number of vertices, and g is the Alexander Joyce, Clemson University, Clemson, SC, United States Let F be a set defined by quadratic constraints. Understanding the structure of the lifted closed convex hull of C(F) is crucial to solve quadratically constrained quadratic programs related to F. In this talk, we discuss the relationship between C(F) and C(G), where G results by adding non-intersecting quadratic constraints to F. We prove that C(G) can be represented as the intersection of C(F) and some half spaces defined by the added constraints. The proof relies on a complete description of the asymptotic cones of sets defined by a single quadratic equality and a partial characterization of the recession cone of C(F). Our proof generalizes an existing result for bounded F with non-intersecting quadratic hollows. 3 - On the Central Path of Semidefinite Optimization: Degree And Worst-case Convergence Rate Ali Mohammad Nezhad, PhD, Purdue University, West Lafayette, IN, United States, Saugata Basu We investigate the complexity of the central path of semidefinite optimization through the lens of real algebraic geometry. To that end, we propose an algorithm to compute real univariate representations describing the central path and its limit point, where the limit point is described by taking the limit of central solutions, as bounded points in the field of algebraic Puiseux series. As a result, we derive an upper bound on the degree of the Zariski closure of the central path and a complexity bound for describing the limit point. Furthermore, by the application of the quantifier elimination to the real univariate representations, we provide an upper bound on the worst-case convergence rate of the central path. 4 - Restricted Simultaneous Diagonalizability with Applications to Quadratic Programming Alex Wang, Carnegie Mellon University, Pittsburgh, 15213, United States Quadratically constrained quadratic programs (QCQPs) are a fundamental class of NP-hard optimization problems that ask us to minimize a quadratic objective function subject to a number of quadratic constraints. In this talk, we introduce and investigate an extension of the simultaneously-diagonalizable-via-congruence (SDC) property—namely, the d-restricted SDC (d-RSDC) property—that will provide us with a tool for simplifying general QCQPs. Informally, a general QCQP can be “lifted” into a diagonal QCQP with only d-many additional variables if and only if it satisfies the d-RSDC property. We will present a number of sufficient conditions for this property to hold and complement our theoretical results with preliminary numerical experiments applying this property to QCQPs with a single quadratic constraint and additional linear constraints.Based on joint work with Rujun Jiang. https://arxiv.org/abs/2101.12141 length of the shortest cycle in the graph with at least 4 vertices. 2 - Convex Hull Results for Quadratic Programs with Non-Intersecting Constraints

109