Hi, I'm looking for a .NET matrix library that is both functionally rich, easy to use from C# and fast. I have had a look at the following libraries:.NET Matrix Library (NML 2.2) by Bluebit SoftwareNMath by CenterSpace SoftwareMath.NET by Christoph Daniel Ruegg.NET Builder for Matlab by The MathWorks Inc.Extreme Optimization Mathematics Library for .NET by Extreme OptimizationMapack by Lutz RoederI personally found .NET Matrix Library and NMath to be super fast and reasonably priced, .NET Builder for Matlab overly expensive, Math.NET and Mapack too slow and simple and I have no real opinion to the Extreme Optimization library yet. Personally I favor NMath (it has richer functionality than .NET Matrix Library), but I wonder if I am missing important candidates.Ideally, I need a c# class library with the following features: basic matrix operations, linear algebra functionality, random number generation, interpolation functions, statistical methods, financial and optimization routines. (I know I'm demanding quite a bit...) Has anybody got experience with one or the other software package? Any recommendations or comments whatsoever are very welcome.

Hi,You should add to your to your list dnAnalytics -- http://www.dnanalytics.net/It's open-source with core computation implementations both in managed code and as wrappers around the super-optimised native Intel and AMD libraries. The scope might not be quite as broad as you require yet -- it seems like they currently cover the main BLAS and LAPACK functionality.Regards,Govert van Drimmelenhttp://exceldna.typepad.com(Free and easy .Net for Excel)

I use Mapack by Lutz Roeder and have done so for several years. It works fine for me. It could be improved by:- intrinsic support for complex matrices (although this is more of a .NET problem which will maybe go away in future versions?)- seperation of the basic matrix functions and the linear algebra functionalityIn general what would be one better is if there was an IMatrix interface defined and implemented by all these matrix libraries so we could move seamlesley between them.And why not have a compiler (or compiler hints) define the internal layout of the array (i.e. morton ordering, row ordering etc.) After all if the one and only operation I am going to do is the followingC = A' B then Having A in row major form, and B in column major form would be an distinct performance advantage.But I think that is blue sky thinking/ dreaming!!Microsoft's experimental language F# also has a matrix library, although no linear algebra is included!

Can't you plug into Matlab somehow???

Interfacing C# to compiled Matlab routines (COM Compiler) works great.

Add http://www.alglib.net to the list. It is a small collection of scientific software library

Take a look at ILNumerics:ILNumericsILNumerics feature listDominik