There are numerous modeling systems for generating linear programs and numerous solvers for optimizing them. However, it is often impossible for modelers to combine their preferred modeling system with their preferred solver. Current modeling systems use their own proprietary model-instance formats that various solvers have been adapted to recognize. The existence of all of these formats suggests that one way to encourage modeling-system and solver compatibility is to use a standard representation of a problem instance. Such a standard must be simple to manipulate and validate, be able to express instance-specific and vendor-specific information, and promote the integration of optimization software with other software. In this paper we present LPFML, an XML Schema for representing linear-programming (LP) instances. In addition, we provide open-source C++ libraries that simplify the exchange of problem-instance and solution information between modeling systems and solvers. We show how our system is used to enable previously unavailable language-solver connections and how our design improves on the state of the art under three different scenarios relevant to communication between solvers and modeling systems.
- Information systems
- Linear programming
ASJC Scopus subject areas
- Information Systems
- Computer Science Applications
- Management Science and Operations Research