<@U0HUJ25V1> <@U0BN58B38> there are three differen...

# mathematics

[altavir]

@thomasnield @kyonifer there are three different layers to answer to the question https://kotlinlang.slack.com/archives/C4W52CFEZ/p1546628579173600.
First, my perspective. Since I work with statistical simulation tools, I of course need distributions and samplers. Most important point, I will need multivariate distribution. Kotlin-statistics is mostly univariate so it probably won't suffice. And I will work in that direction anyway

From Thomas perspective. I already have some tools wrapping commons-rng (which in my opinion is much better suited for that than ojalgo). I can port basic functionality to kmath (multiplatform api, but only for jvm implementation for now) in a few hours. You can just put a feature request on tracker.

From JetBrains perspective. They want a single multiplatform (this is important!) library supported for community or at least single interface with multiple implementations. I do not think koma could play this role because it is to pythonic and too monolithic for that. I am not sure that kmath should do also (it is too experimental for now). We need some kind of middle ground and we need to work together to get the approval from JetBrains (this is not my speculation, this is the official position). For now, the only way I see that does not lead to wounding someones pride is to fill https://github.com/mipt-npm/KEEP-math as soon as we can.