Identify the actions that will produce the best results, while operating within resource limitations and other relevant restrictions, using a powerful array of optimization, simulation and project scheduling techniques.
Quickly solve complex optimization problems.
Find optimal solutions to difficult problems faster than ever. SAS Optimization takes advantage of the SAS® Viya® distributed, in-memory engine to deliver optimization modeling results at breakthrough speeds. In-memory data persistence eliminates the need to load data multiple times during iterative analyses.
Drive better decision making.
Identify and apply the best responses to complex, real-world problems. State-of-the-art methods for mathematical optimization are integrated with a full suite of data preparation, exploration, analytics and reporting capabilities – all in one unified environment.
Empower users with their preferred programming language.
Python, Java, R and Lua programmers can take advantage of the wide range of solvers in SAS Optimization without having to learn SAS code. They can access powerful, trusted and tested SAS algorithms from the programming language they are most comfortable with.
See SAS® Optimization in action.
Mike Gilliland, Product Marketing Manager at SAS, demonstrates how you can use SAS Optimization to build and solve an optimization model that guides financial investment decisions.
- Powerful, intuitive algebraic optimization modeling language. Enables you to produce a range of models, including linear, mixed integer linear, nonlinear, quadratic and network optimization, as well as solve constraint satisfaction problems.
- A unified modeling language. Supports a wide range of optimization models with a single modeling and solution framework. You only need to learn one set of statements and commands to build a range of optimization and constraint satisfaction models.
- Powerful optimization solvers and presolvers. Provides a suite of optimization solvers – all streamlined for simplicity and tuned for performance. Aggressive presolvers reduce effective problem size so you can tackle large problems and solve them more quickly.
- Network flow optimization. Provides network algorithms, accessible from both PROC OPTMODEL and PROC OPTNETWORK, for investigating the characteristics of networks and finding the best answers to network-oriented problems.
- Multistart algorithm for nonconvex nonlinear optimization. Increases your chance of finding a globally optimal solution from among many locally optimal solutions. Selects multiple starting points and begins optimization in parallel from each; then reports the best solution from all starting points.
- Decomposition algorithm (automated Dantzig-Wolfe). Decomposes the overall problem into a set of component problems, each with an exclusive set of decision variables, which are solved in parallel. Parallel solution of the subproblems is coordinated with the overall solution process, reducing time to solution significantly.
- Local search optimization (LSO). LSO solver can be used with (generally nonlinear) optimization problems that don’t adhere to the assumptions that conventional optimization solvers make. Functions might be discontinuous, nonsmooth, computationally expensive to evaluate, based on black-box simulations, etc.
- Constraint programming. Solves constraint satisfaction problems using domain reduction/constraint propagation and a choice of search strategies, such as look ahead and backtracking.
- Accessible, cloud-enabled, in-memory engine. Takes advantage of the SAS Viya engine, which brings new enhancements to the SAS Platform, including high availability, fast in-memory processing, the ability to code from open source languages and native cloud support.
SAS is a Leader in The Forrester Wave™: Multimodal Predictive Analytics And Machine Learning (PAML) Platforms, Q3 2018.