Math-oriented programming for .NET

The July End Bracket column of MSDN magazine brings some exciting news. Don Syme, who was instrumental in bringing generics to the CLR, has turned his attention to “attempting to achieve a synthesis between type-safe, scalable, math-oriented scripting and programming for .NET. I believe the industry needs to do more for scientists and mathematicians, where the demand for this kind of language is great.”

I couldn't agree more.

The .NET platform already delivers on its promise for increased productivity for most mainstream applications. But the scientific community has been left to fend for itself mostly. Having built a complete math library for the platform (version 1.1 has just been released), I've had to contend with several features of the .NET platform that, from our point of view, showed a lack of attention for numerical programming scenarios.

Don 'gets' what engineers and scientists need: a scripting language within an environment suitable for exploratory analysis (like Matlab and Mathematica) as well as a solid infrastructure to build real-world end-user applications. The .NET Framework makes the latter a whole lot simpler.

On the scripting front, both F# and Jim Hugunin's IronPython offer promise. How well these languages scale to real-world applications remains to be seen. The difficulties in creating a performant Matlab compiler lead me to believe that more of the trade-offs may need to be decided in favor of a static language.

posted on Monday, June 27, 2005 10:47 AM