INFORMS 2021 Program Book

INFORMS Anaheim 2021

TE04

landscape of competitiveness owing to its heterogeneity, multiformity, and dynamicity. To mitigate these issues, we design a novel firm profiling and competitiveness assessment system following the structured procedure defined by Information System Design Theory (ISDT). The efficacy and competency of our proposed system are validated by performance comparisons of multiple models in different settings as well as by formal statistical testing analyses. Overall, our solutions shed new light on interfirm competitiveness assessment problem and are potentially beneficial to academic scholars as well as practitioners and decision makers. TE04 CC Ballroom D / Virtual Theater 4 Hybrid Methodological Advances in MINLP Sponsored: Computing Society Sponsored Session Chair: Aleksandr Kazachkov, University of Florida, FL “ 1 - Formulations Comparison for Piecewise Convex Relaxation of the Sequential Convex MINLP Method Claudia D’Ambrosio, LIX, CNRS, École Polytechnique, Institut Polytechnique de Paris, Route de Saclay, Palaiseau, France, Antonio Frangioni, Claudio Gentile, Renan Spencer Trindade The Sequential Convex Mixed Integer Non Linear Programming (SC-MINLP) method is an exact approach to solve problems with non convexities arising in separable functions. It is based on lower and upper bounds computed by solving a sequence of piecewise convex relaxations and non convex restrictions, respectively. In this talk, we focus on the comparison of different ways to formulate the piecewise convex relaxations, inspired by classic piecewise linear approximations. While in the linear case the continuous relaxations of the formulations are equivalent, this is not true for the convex case. Computational results on a few classes of problems are presented as proof-of-concept. 2 - Intersection Cuts for QCQPs via Maximal Quadratic-free Sets Gonzalo Munoz, Universidad de O’Higgins, Rancagua, 2820000, Chile, Antonia Chmiela, Felipe Serrano The generation of strong linear inequalities for QCQPs has been recently tackled by a number of authors using the intersection cut paradigm a highly studied tool in integer programming whose flexibility has triggered these renewed efforts in non-linear settings. In this talk, we consider intersection cuts using the recently proposed construction of maximal quadratic-free sets. We describe the construction of these sets and show how to obtain closed-form formulas from them to compute intersection cuts using an arbitrary quadratic inequality being violated by a vertex of an LP relaxation. We evaluate this approach with extensive computational experiments. 3 - Solving the Pooling Problem at Scale with Extensible Quadratic Optimizer Galini GALINI is an open source solver for non-convex quadratic optimization problems formulated with Pyomo. We have also built a Python library to model pooling problems, a class of network flow problems with many engineering applications. We demonstrate GALINI’s extensible characteristics by using the pooling library to develop two GALINI plug-ins: 1) a cut generator plug-in that adds valid inequalities in the GALINI cut loop and 2) a primal heuristic plug-in that uses the mixed-integer linear restriction. We show that, thanks to the good upper bound provided by the mixed-integer linear restriction and the good lower bounds provided by the convex relaxation, GALINI obtains optimality gaps competitive with Gurobi 9.1 on large instances. 4 - Quantum-inspired Formulations for the Max K-cut Problem Ramin Fakhimi, Lehigh University, Bethlehem, PA, 18015, United States, Hamidreza Validi, Illya V. Hicks, Tamás Terlaky , Luis F. Zuluaga The max k-cut problem is a challenging combinatorial optimization problem with multiple well-known optimization formulations. However, its mixed-integer linear optimization (MILO) formulations and mixed-integer semidefinite optimization formulation are all time-consuming to be solved. Motivated by recent progress in classic and quantum solvers, we study a binary quadratic optimization (BQO) formulation and two quadratic unconstrained binary optimization (QUBO) formulations. First, we compare the BQO formulation with the MILO formulations. Further, we propose an algorithm that converts any feasible fractional solution of the BQO formulation to a feasible binary solution whose objective value is at least as good as that of the fractional solution. Finally, we find tight penalty coefficients for the proposed QUBO formulations. Ruth Misener, Imperial College London, South Kensington Campus, London, United Kingdom, Francesco Ceccon

4 - The Obnoxious Facilities Planar P-median Problem with Variable Sizes Pawel J. Kalczynski, California State University-Fullerton, Fullerton, CA, 92834-6848, United States, Zvi Drezner The obnoxious facility location problem is to locate facilities that have a negative impact on communities (being “obnoxious”), and being farther from communities is preferred. However, such facilities also serve the communities. Our goal is to minimize the system’s operating cost subject to a minimum distance requirement from communities. The multiple obnoxious facility problem is usually defined as locating several facilities maximizing the minimum distance between facilities and communities. However, not all facilities have the same impact on communities. We assume that the size of a facility depends on the volume of service provided by it. The problem is extremely non-convex. We designed a special starting solution for non-linear solvers that provides a much better objective (in some cases cutting it by half) in a very small fraction of the run time. TE03 CC Ballroom C / Virtual Theater 3 Hybrid Data Mining in Networks Sponsored: Artificial Intelligence Sponsored Session Chair: Sulyun Lee, The University of Iowa, Iowa City, IA, 52240, United States 1 - HIPED: Heterogeneous Interaction-based Dynamic Embedding of Patients Hankyu Jang, University of Iowa, Iowa City, IA, 52246, United States, Sulyun Lee, Hasib Hasan, Sriram Pemmaraju, Bijaya Adhikari Representation learning of patients gained attention recently due to its applicability in meaningful prediction tasks in the healthcare setting. In this work, we propose Heterogeneous Interaction-based Dynamic Embedding of Patients (HIDEP), an unsupervised embedding approach that learns latent representations of patients from their temporal, heterogeneous interactions at hospitals. We model patient interaction with medications, doctors, and rooms over time during the course of their inpatient visit to the hospital. We evaluate the learned patient embeddings on various prediction tasks such as early detection of healthcare- associated infection, patient transfer into medical intensive care unit, mortality risk prediction, and severity risk prediction. Our results show that HIDEP outperforms the state-of-the-art methods in all the prediction tasks. 2 - Predicting NFL Team Performance via Hierarchical Team Embeddings Sulyun Lee, University of Iowa, Iowa City, IA, 52240, United States, Kang Zhao, Changze Han Collaboration is a fundamental part of teams where individuals form groups to reach the common goals of assigned tasks. In networks such as academic collaboration networks, sports teaming networks, or online gaming networks, exploring the aspects of collaborations is essential in predicting team performances. This work focuses on collaborations of coaches in NFL teams, which is the American professional football leagues. We observed the collaborations among coaches in the same team and proposed a model that predicts the team performance given the coach information and collaborations. Specifically, we focused on the hierarchical collaborations among the NFL coaches, where coaches are at different levels in the command structures. 3 - Understanding the Spread of Misinformation on Social Media - the Effects of Topics and a Political Leader’s Nudge Xiangyu Wang, University of Iowa, Iowa City, IA, 52246-5104, United States, Min Zhang, Weiguo Fan, Kang Zhao The spread of misinformation on social media has become a major societal issue. In this work, we used the ongoing COVID-19 pandemic as a case study to investigate factors associated with the spread of multi-topic misinformation based on the Heuristic-Systematic Model. Among factors related to systematic processing of information, we showed that the topics of a misinformation story matter, with conspiracy theories being the most likely to be retweeted. As for factors related to heuristic processing of information, when citizens look up to their leaders during such a crisis, our results demonstrated that behaviors of a political leader, former U.S. President Donald Trump may have nudged people’s sharing of COVID-19 misinformation. Outcomes of this study help social media platform and users better understand and prevent the spread of misinformation on social media. 4 - Firm Profiling and Competitiveness Assessment: A Heterogeneous Occupation Network Embedding Approach Howard Zhong, ESCP Business School, Paris, France, Chaunren Liu Extensive efforts have been made by both academics and practitioners to understand interfirm competitiveness due to its profound significance in multiple key business objectives. However, it is not an easy task to fully depict the

139