Microsoft has made its .NET Optimization library available on DevLabs - a site that puts forward technical prototypes for you to try out and comment on. There is no suggestion that this is open source or even free software - as the blurb says you can try it out and comment on it. What is surprising is that Microsoft Solver Foundation has been around for some time (its reached version 3.0) and the relationship with the Enterprise Edition of the same software isn't particularly clear. When you download the DevLabs offering it proclaims itself as the Express Edition.
The main point is not to look a gift horse too closely in the mouth. This is an opportunity for you to find out if this optimization suite does what you want and it is issued, according to the website, under the Microsoft Limited Public License which allows you to create derivative works and sell them. Although at the moment the actual download installs with a stricter licence that says you can't create derived products or sell them. If you are planning to use the library for real it might be worth clarifying the position.
Solver Foundation is a .NET library for mathematical programming, modeling, and optimization. The library can be used from any .NET language including C# and F#. There is also an Excel Add-in that can be used to create models that work with spreadsheet data. You can also run it as a service using IIS and ASP.NET. Models can be specified using Optimization Modeling Language. It takes a declarative approach and generally you can create a model simply by specifying constraints, goals and the data of course.
After you have installed the library two new project types are created in Visual Studio 2010. It also works with Visual C# 2010 Express.
While you can create custom models it also includes built-in solvers for standard model types. It also integrates with other popular solvers that cost a lot more than free.
Version 3 adds the ability to create constrained non-linear programming models and integer quadratic programming models. Of course it already supported the staple of the optimization world linear programming and more complex procedures such as simulation and stochastic programming.
Download Solver Foundation
Somasegar's weblog on Solver