INFORMS 2021 Program Book

INFORMS Anaheim 2021

MC28

2 - How do Producers Fare with Fair Trade? Yen-Ting (Daniel) Lin, University of San Diego, Olin Hall, School of Business Administration Univ, San Diego, CA, 92110, United States, Adem Orsdemir, Ying Zhang Fair trade certificate promotes a sustainable livelihood for producers in developing countries. It ensures that producers are paid at a fair wage and potentially protects producers from the volatility in the commodity market. We examine the impact of fair trade certificate on a company’s decisions, profitability and participating producers’ welfare. We also examine a firm’s choice between fair trade and direct trade, another common socially responsible sourcing strategy.

MC27 CC Room 206B In Person: Optimization Modeling Software General Session Chair: Steven P. Dirkse, GAMS Development Corporation, Fairfax, VA, 22031-4342, United States 1 - Recent Improvements to MathOptInterface Benoît Legat, MIT, Cambridge, MA, United States MathOptInterface (MOI) provides a intrinsically flexible and extensible API.Such design often comes at the cost of reduced performance over specialized interface.In this presentation, we discuss the challenges for MOI to both be generic and performant.We focus on two specific approaches to this: mutable arithmetics and matrix representations. 2 - New Connections to the AMPL Modeling Language: Spreadsheets and Callbacks Robert Fourer, AMPL Optimization Inc., Evanston, IL, 60201- 2308, United States, Filipe Brandão Optimization applications are often concerned as much with making connections as with building models. This presentation describes two connections recently implemented in the AMPL modeling language and system. A direct spreadsheet connection reads and writes xlsx-format files, defining correspondences between common spreadsheet layouts and AMPL’s data definitions. Support is included for “two-dimensional” spreadsheet tables in which one index labels the columns and one or more indices label the rows. A solver callback connection enables AMPL’s APIs to communicate with algorithms as they are running, uniting the ease of modeling in AMPL with the flexibility of programming to customize algorithmic behavior. This facility can be used to write specialized routines that report progress, change settings, and generate constraints that cut off fractional solutions. 3 - GAMS/Engine A New System for Solving Models on Centralized Compute Resources Steven P. Dirkse, GAM S. Development Corporation, Fairfax, VA, 22031-4342, United States, Frederik Proske, Hamdi Burak Usul Typically, personal computers have been powerful enough to quickly solve the model instances generated by GAMS. If not, users (or their expert IT staff) have implemented custom scheduling systems to run large optimization jobs on central compute resources.Increasingly, users want to run large jobs or large streams of jobs on the cloud. This enables them to access more powerful machines than typically found on a desktop and also to utilize a scalable pool of worker machines if their job stream benefits from this, but arranging for all this still requires the expert IT staff.To relax this requirement, we have developed GAMS Engine, a powerful GAMS job-scheduling system. Central to Engine is a modern REST API that provides an interface to a scalable Kubernetes based system of services, providing API, database, queue, and a configurable number of GAMS workers.

MC26 CC Room 206A

In Person: Sequential Decision Making and Machine Learning in Healthcare/Innovations in Procurement and Utilization of Deceased-donor Organs General Session Chair: Tinglong Dai, Johns Hopkins University, Baltimore, MD, 21212- 1708, United States 1 - Feature-Based Design Of Priority Queues: Digital Triage in Healthcare Simrita Singh, Kellogg School of Management at Northwestern University, Evanston, IL, 160071, United States We study data-driven classification where a classifier assigns jobs (e.g., patients or medical images) based on observed features to priority queues for human review. Traditional classifiers are designed to minimize misclassification loss functions but may underperform when integrated with queueing systems. We propose an integrated approach where the classifier and the queueing system are optimized to minimize the workflow’s average waiting cost. We demonstrate the value of our approach using an actual data set covering 560,486 patient visits to three emergency rooms over three years. 2 - Patient Prioritization in the Emergency Department Gizem Yilmaz, PhD Candidate, University of Chicago, Chicago, IL, 60615, United States, Daniel Adelman In the emergency department, triage nurses assign arriving patients to an urgency class. Although urgency class is an essential factor for patient prioritization, it is not the only one. We aim to understand the current prioritization rule for the ED bed allocation by leveraging the high-fidelity emergency data and machine learning literature. We build a model that predicts the next patient assigned to a bed and evaluate the viability of deploying this prediction model in operations to support prioritization decisions. The prediction model gives us insights into the current prioritization rule for the ED bed allocation. We leverage these insights to develop an improved prioritization rule. 3 - A Simple Incentive Mechanism to Alleviate the Burden of Organ Wastage in Transplantation Sait Tunc, Virginia Tech, Blacksburg, VA, 24061, United States Despite efforts to increase the supply of donated organs for transplantation, organ shortages persist. We study the problem of organ wastage in a queueing-theoretic framework. We establish that self-interested individuals set their utilization levels more conservatively in equilibrium than the socially efficient level. To reduce the resulting gap, we offer an incentive mechanism that recompenses candidates, who have accepted a pre-defined set of organs and returned to the waitlist for re- transplantation, for giving up their position in the waitlist and show that it increases the equilibrium utilization of organs while also improving social welfare. 4 - Does Transportation Mean Transplantation? Impact of New Airline Routes on Sharing of Cadaveric Kidneys Tinglong Dai, Johns Hopkins University, Baltimore, MD, 21212- 1708, United States, Guihua Wang, Ronghuo Zheng Every year, nearly 5,000 patients die while waiting for kidney transplants, and yet an estimated 3,500 procured kidneys are discarded. Such a polarized co-existence of dire scarcity and massive wastefulness has been mainly driven by insufficient pooling of cadaveric kidneys across geographic regions. Although numerous policy initiatives are aimed at broadening organ pooling, they rarely account for a key friction— efficient airline transportation, ideally direct flights, is necessary for long-distance sharing, due to the time-sensitive nature of kidney transplantation. Conceivably, transplant centers may be reluctant to accept kidney offers from far- off locations without direct flights. In this paper, we estimate the effect of the introduction of new airline routes on broader kidney sharing.

MC28 CC Room 207B In Person Technology Tutorial: Nonlinear Optimization Using Artelys Knitro Technology Tutorial 1 - Nonlinear Optimization Using Artelys Knitro Richard Waltz, Artelys, Los Angeles, CA, 90045-2603, United States

Nonlinear optimization is used in many applications in a broad range of industries such as economy, finance, energy, health, 3D modeling, and marketing. With four algorithms and great configuration capabilities, Artelys Knitro is the leading solver for nonlinear optimization and demonstrates high performance for large scale problems. This session will introduce you to Artelys Knitro, its key features and modeling capabilities, with a particular emphasis on the latest major improvements including recent advances in solving mixed-integer nonlinear optimization problems. We will also provide benchmarks highlighting the power of Knitro to efficiently solve large-scale, nonlinear models with hundreds of thousands of variables and constraints.”

65