Looking a little at Kmath, how am I supposed to us...

# mathematics

[Hampus Londögård]

Looking a little at Kmath, how am I supposed to use the LinearSpace like EjmlLinearSpaceFSCC.
More exactly the extension-functions like Matrix<Float>.toEjml()

I've tried to use the with (EjmlLinearSpaceFSCC) { code... } syntax but it doesn't work. Not sure how else I'd get those to work?

[altavir]

This extension is mostly for inner use. It is used under the hood for converting external matrices to EJML. basically with (EjmlLinearSpaceFSCC) { code... } should work. Could you explain what is the problem?

[Hampus Londögård]

I'm currently doing a very simple example (which I have the code for in Ejml, but wanted to try Kmath as it seems interesting).

This is the current code failing:

[Hampus Londögård]

This is just a sample.
In the transform I really need sparsity as the output is very sparse
tI don't really need the sparsity right now, but the output of

[altavir]

Let me try it, I have not tried it since @Iaroslav Postovalov introduced codegen.

[Hampus Londögård]

@altavir my bad... I had <T>

[Hampus Londögård]

😆

[altavir]

Ah, OK. Then it should not work. Basically, you do not need to call toEjml because it is converted automatically under the hood. Just do the operations you need

[Hampus Londögård]

That's good to know! 🙂
If I wish to find all distinct values in a `Matrix`I must either use rows or elements where both has PerformancePitfall
Is there another recommended way to fetch elements? Simply a for-loop?

[altavir]

The rows does for-loop. So it should be OK for you. But it is in general depends on implementation. Elements is more time-consuming since it creates an index-element pair on each step. So it is not a good idea to use it in performance-critical code. The warning does not necessary mean that it is slow, it means that you should probably check it.

[Hampus Londögård]

Yes, the code made that clear (awesome!) - that's why I wanted to validate 😄

Thank you!