Exact and robust summation/dot product subroutines

[mikolalysenko]

Would be nice to have some subroutines for doing fast summation and dot products, and maybe more speculatively exact linear solvers (though this could be prohibitively slow/expensive). This article has some neat ideas, also there is the work by Shewchuk we could perhaps adapt:

• http://arxiv.org/pdf/1505.05571v1.pdf

[prabod]

I would like to contribute to this.How should i get started?

[rreusser]

I think the best place to start is with code, as you've done with the flow network algorithms. @jaspervdg summarized some guidelines here for polishing modules and there are lots of templates to follow.

If you're asking about specific technical details though, this isn't my specialty so I don't have immediate answers for you. If that's the case, then the best place to start is probably with a specific proposal for a module or modules. If you can list the functions you'd like to implement (functions, inputs, outputs), then everything else is just a matter of implementation 😄