Decision Support Systems

I have built a DSS prototype, which employs analytic form of fuzzy real options. Because it does not require much computational power, it is implemented in MatLab and C#. C# and XAML are employed for the user interface design based on WPF (.NET 3.0 framework), while MatLab is used to implement the fuzzy real options model. Data type conversion between .NET and Matlab is explored. The power of WPF is demonstrated.

When an closed-form solution is unavailable and a numerical method has to be applied, two tests are conducted. One is based on Ubuntu Linux with Python and an interface to Boost library. Both objective oriented and functional paradigms are recommended.  It is found that Python is a very flexible programming language, but due to its nature as an interpretative language, the performance is unsatisfactory.

Hence, the core numerical method is reimplemented in C with GSL. Result is astonishing: very much faster than Python. When parallel programming environment is deployed on SharcNet, the performance can only be described as unbelievable: from 3 hours to 10 seconds. I sincerely hope that Compute Canada will be preserved in service forever.

A DSS around the algorithmic module in C is designed and under development. Java is used as the major programming language, linking to the module via JNI. An open source GIS module (GeoTools)  is employed to add mapping capacity under the browser/server architecture.