Christoph Ruegg

Consider Magma linear algebra native provider

2

http://icl.cs.utk.edu/magma/

From: https://mathnetnumerics.codeplex.com/workitem/5663 > nested tight loops, we can remove index calculation based on the outer loop.

Math.NET Numerics should leverage diagonal matrices and provide the trivial decompositions directly.

Proper sparse matrix multiplication and related (contributed)

1

Integrate excellent contribution by @wo80 via https://mathnetnumerics.codeplex.com/discussions/543114

Verify TargetedPatchingOptOut has no effect and drop all of them if confirmed

1

We currently apply `TargetedPatchingOptOut` attributes in a few places to mark functions that should be inlined. However, it seems this does not actually have any effect since user assemblies (i.e....

From: https://mathnetnumerics.codeplex.com/workitem/5633 > We could implement SciMark 2.0 to test the various versions of our library and compare them. Perhaps even with respect to other libraries that are out there....

A section in the docs for those with a background in NumPy, R or MATLAB with examples on what common code elements of these environments translate to in Math.NET Numerics...

Rework the Code Samples

2

We have both a C# and F# sample projects that are also distributed as NuGet packages. However, the samples have not really been maintained for a while: - We should...

From: https://mathnetnumerics.codeplex.com/workitem/5659 > As I am not familiar with preconditioning: is there any guide, when to use what preconditioner with a specific solver? Or when to use preconditioning at all?...

There are currently 3 test cases of `SkewnessConsistentWithR_e1071` in `RunningStatisticsTests` commented out. They compare the skewness of our `RunningStatistics` with the `e1071` R-package on a synthetic NIST dataset designed to...