-
Notifications
You must be signed in to change notification settings - Fork 57
Description
Is your feature request related to a problem? Please describe.
Currently, the vector and matrix representations in FSharp.Stats are tightly coupled with the rest of the library. This can make it harder to focus the library purely on statistical functionality and maintain a clear separation of concerns.
Describe the solution you'd like
We propose separating the vector and matrix components into a lightweight, array-centric math library, named FsMath under the FsLab.org umbrella. FsMath would provide high-performance, general-purpose numerical primitives that integrate seamlessly with FSharp.Stats (zero friction).
This approach would:
- Improve maintainability by decoupling numerical primitives from statistical algorithms.
- Provide performance gains through a streamlined, specialized math core.
- Allow FSharp.Stats to focus on statistical functionality and reach feature completeness for statistical use cases.
- Enable cleaner integration into F# workflows that also depend on other common libraries.
Additional context
FSharp.Stats will remain the main hub for data science in F#, but will be supported by FsMath as the dedicated math layer.
FsMath repository: here