core.matrix support? about c2 HOT 3 CLOSED

Mike,

I've been following the core.matrix discussion closely on the Clojure mailing list and am planning on digging into it soon for an internal project we're working on related to shredded document reconstruction.

As for C2 "integration", I'm not sure what exactly there is to do.
If core.matrix vectors and matrices implement the seq protocols, you can do all of the usual things to generate hiccup/c2 DOM structures.
Color-encodings are popular for matrix visualizations, and there is certainly a ton of stuff that could be added to C2 to help with that (color scales, automatic quantization for discrete color scales, &c.).

Did you have anything specific in mind that would be helpful to you?

from c2.

Hi Kevin,

The core.matrix structures don't necessarily support seq - since they could easily be Java matrix classes that don't implement either ISeq, Seqable or Iterable.

They are however guaranteed to support clojure.core/eseq (element sequence) since that is backed by protocols that are extended to all relevant matrix / vector types.

As far as integration goes, I can think of several ideas:

• Enable code that deals with data sets to work with core.matrix data (e.g. the code in https://github.com/lynaghk/c2/blob/master/src/cljx/c2/maths.cljx)
• Usage for multi-dimensional data visualisations
• Acceleration of data processing (e.g. matrix transforms for visualisation projections?)
• Colour space transformations

Of course these are all just ideas...... mostly I'm just glad to see it is on your radar. I think it will be very beneficial for the Clojure community and help enormously with interoperability if foundational stuff like matrix / vector maths doesn't need to be reinvented for every single project. That's the main thing I'm trying to achieve here with core.matrix.

from c2.

Consider it on my radar = )

I'll open source anything core.matrix-related if we pick it up on projects and write something with broad utility.
However, I'm going to close this issue since there's nothing specific here that I can address.

from c2.