Gurobi Optimization, LLC today announced the release of Gurobi 9.0, the latest version of its industry-leading mathematical programming solver. Gurobi 9.0 delivers breakthrough new capabilities for the Gurobi Optimizer, major new features for the Gurobi Compute Server, and notable performance improvements across linear programming (LP), mixed-integer programming (MIP), and mixed-integer quadratic programming (MIQP) problem types.
First introduced in 2008, Gurobi’s mathematical optimization solver is currently being used by more than 2,400 companies in over 40 industries to tackle their complex business problems and make better decisions.
“The launch of Gurobi 9.0 – which contains more new features than any prior release we’ve ever done and improves upon the already best-in-class, record-setting performance of the previous version – proves that Gurobi continues to be the industry leader when it comes to mathematical programming solver performance and innovation,” said Dr. Tobias Achterberg, Director of Research and Development at Gurobi Optimization.
“With every new release, we strive to significantly enhance the performance and expand the capabilities of our mathematical programming solver – and with Gurobi 9.0 we’ve done exactly that,” Achterberg added.
Gurobi Optimizer: Major New Features
The Gurobi 9.0 release includes the following new features for the Gurobi Optimizer:
- Non-Convex Quadratic Optimization: With the addition of a new bilinear solver, the Gurobi Optimizer now supports non-convex quadratic optimization. This groundbreaking new capability allows users to solve problems with non-convex quadratic constraints and objectives – enabling them to find globally optimal solutions to classic bilinear pooling and blending problems.
- Piecewise-Linear (PWL) Constraints: Users can now add PWL constraints directly as a modeling construct instead of having to represent them as a combination of linear and SOS constraints – thereby greatly simplifying their use.
- Function Constraints with Automatic PWL Translation: With Gurobi 9.0, users can add nine basic, nonlinear, univariate function constraints – including polynomial, exponential, logarithmic, and trigonometric functions – and the Gurobi Optimizer will perform an automatic piecewise-linear approximation of these functions.
- Python Matrix API: The Gurobi Optimizer’s Python interface has been extended to support matrix-oriented modeling using NumPy or SciPy matrices.
- MIP Scenario Analysis: This new feature allows users to construct multiple scenarios and run them simultaneously.
- New Solution Improvement Heuristic: Gurobi 9.0 provides a new, more effective heuristic for focusing on improving MIP solutions.
- Intermediate Solution Files – Users are now able to write solutions found during the MIP solution process to files as soon as they are discovered.
Gurobi Optimizer: Increased Performance
With Gurobi 9.0, the Gurobi Optimizer – which was tested using Gurobi’s test library consisting of thousands of real-world models – demonstrates substantial performance improvements across LP, MIP, and MIQP problem types compared to Gurobi 8.1. The performance improvements are as follows:
- LP: 7% faster in default settings.
- Barrier: 7% faster and an additional 4% faster with computers supporting AVX 512.
- MIP: 18% faster overall and 26% faster on difficult models that take more than 100 seconds to solve.
- MIQP: 24% faster.
Commenting on the new capabilities and improved performance of the Gurobi Optimizer in the latest release, Achterberg said: “With Gurobi 9.0, we have achieved record-breaking solver speeds and added new features that have never been previously available on a commercial solver. These improvements will bring tremendous benefits to our customers, enabling them to more effectively utilize mathematical optimization to make better business decisions and realize greater operational efficiency and cost savings.”
Gurobi Compute Server: New Functionality
The Gurobi 9.0 release adds new functionality for on-premises and private cloud deployment of the Gurobi Compute Server including:
- New Cluster Manager: The Cluster Manager is a new server component that can be installed with Gurobi Compute Server nodes. It provides better security with user authentication and API keys and also expands the capabilities of the cluster nodes with a unified management of interactive and non-interactive optimization tasks.
- New Non-Interactive Solve (Batch Mode) and APls: With the Cluster Manager, users can build an optimization model locally, submit it as a non-interactive optimization task, and then disconnect from the server. Once the batch is completed, the user can retrieve the results.
- New Web User Interface: With the Cluster Manager, users can easily monitor and administrate user accounts, cluster nodes, jobs, and batches. Users can also submit batches from files using a drag-and-drop interface.
- New Job/Batch History: With the Cluster Manager, users can now keep records of submitted jobs and batches, enabling them to monitor the cluster usage by user, application, or time range.
- Support for Lazy Constraints Callbacks: With Gurobi 9.0, applications that need to add lazy constraints on the fly from within the callback can also run on the Gurobi Compute Server or the Cluster Manager.
For more information on Gurobi 9.0, click here.
About Gurobi Optimization
Gurobi (http://www.gurobi.com) is in the business of helping companies make better decisions. In addition to providing the fastest math programming solver, as well as tools for distributed optimization and optimization in the cloud, the company is known for its outstanding support and flexible licensing.
The Gurobi Optimizer is a state-of-the-art solver for linear programming (LP), quadratic programming (QP), quadratically constrained programming (QCP), mixed-integer linear programming (MILP), mixed-integer quadratic programming (MIQP), and mixed-integer quadratically constrained programming (MIQCP). Gurobi was designed from the ground up to exploit modern architectures and multi-core processors, using the most advanced implementations of the latest algorithms. Founded in 2008, Gurobi Optimization is based in Beaverton, OR (+1 713 871 9341).