Building an Optimization Model: Exploring Your Options
When building an optimization model, one must choose from among two alternatives: Using Gurobi with a proprietary modeling language such as AMPL or GAMS, or using Gurobi with a full programming language such as C, C++, C#, Java, Python, VB, MATLAB or R. Using a modeling language can be an attractive choice, especially for non-programmers, given the perceived ease with which a model can be formulated. Using a programming language can be a much more powerful and flexible choice if you are interested in formulating a model and then deploying that model and/or integrating it into an application for others to use. This perceived trade-off led us to wonder, why should someone have to choose between easy or powerful? Is there a way to combine the ease of the modeling language with the power and flexibility of a programming language, without requiring significant programming skills on the part of the modeler?
Introducing the Gurobi Python Environment
The Gurobi Python Environment combines the benefits of a modeling language with the strengths a programming language. By embedding our set of high-level optimization modeling constructs in the very popular Python programming language, we’ve eliminated the need to choose between working in just a modeling language or just a programming language. The Gurobi Python environment allows you to tap into the vast ecosystem of Python, a mature, full-featured and easy-to-use language. With Python, you can take advantage of numerous available pre-written and tested packages (over 50,000 at last count) that can save you significant development time when creating new capabilities for your program. And based on our experience, since Python is a very readable and easy programming language to get started with, we think you’ll be most productive in the Gurobi Python environment—even if you are already familiar with another programming language. By using Python, we’ve created an environment that is:
- Flexible and powerful – Use it just for prototyping, or to create full-featured optimization applications.
- Easy to use – Create simple models, requiring a very basic understanding of the language. Create more complex models that are still concise, efficient and easy to express with just a bit more knowledge.
- Robust – Take advantage of Python’s full range of pre-built packages to support full application development, including exceptional data access capabilities. These capabilities are in part the result of a very large and rapidly growing user community.