The OR 2021 is the joint annual conference of the Operations Research Societies of Switzerland (SVOR), Germany (GOR e.V.) and Austria (ÖGOR).
The conference connects academics and practitioners from operations research, management science, data science and analytics. Designed around this year’s special theme - « Business Analytics for Data-Driven Decision Making » - the OR 2021 is an excellent platform to share the most recent research, to discuss trends in the industry and interact with leading experts.
The OR 2021 will be held online from August 31 to September 3, 2021.
GAMS will be part of the OR 2021 with some interesting talks about the new GAMS Engine and about Model Deployment with GAMS:
Algebraic modeling languages (AMLs) such as AIMMS, AMPL, and GAMS are established tools in operations research, and in many business applications. Traditionally, personal computers have been powerful enough to solve AML model instances in acceptable time, especially when using highly efficient commercial solvers. Exceptions can be observed, for example in the energy systems modelling field, where very large, memory bound LPs have to be solved, which are beyond the capabilities of ordinary desktop computers. Modelers in those areas have access to qualified IT staff, who can implement custom scheduling systems to run large optimization jobs on central compute resources, or increasingly in the cloud. However, with the general trend towards data based decision making, the computational requirements of AML models in other fields grow faster than the performance increase of PC hardware. Therefore, many users outside of the typical memory hungry fields now regularly hit the limits of personal computers and would benefit from the use of more powerful compute resources. To offer a solution to this problem, we have developed GAMS Engine, which is a powerful GAMS job-scheduling system. Central to Engine is a modern REST API, which provides the interface to a scalable Kubernetes based system of services, providing API, database, queue, and a configurable number of GAMS workers. In this presentation we will explain the system architecture, it’s benefits, and demonstrate how to communicate with the REST API.
In most cases, using GAMS in the typical fashion - i.e. defining and solving models and evaluating the results within the given interfaces is a sufficient way to deploy optimization models. The underlying field of mathematical optimization, in which the focus is not so much on visualization as on the problem structure itself, has remained a kind of niche market to this day. In the large and very extensive segment of business analytics, however, intuitive deployment and visualization is indispensable. Since these two areas increasingly overlap, interest in alternative deployment methods is also growing in the field of mathematical optimization. In this talk we present a new interface to deploy GAMS models. We show how to turn a model into an interactive web application in just a few steps. In addition, the generation, organization, and sensitivity analysis of multiple scenarios of an optimization model is addressed. We demonstrate how a model written in GAMS can be deployed with this application on either a local machine or a remote server. While data manipulation and visualization as well as scenario management can be done via the web interface, the model itself is not changed. Therefore, the Operations Research analyst can keep focusing on the optimization problem while end users have a powerful tool to work with the data in a structured way and interactively explore the results.
