eigen decomposition not yet implemented?

[michaellindon]

I'm trying to perform an eigen decomposition of a simple matrix

(eigen [[10 2] [2 10]])

I get an error

ExceptionInfo TODO: Not yet implemented: (mp/eigen m options) for class mikera.matrixx.Matrix clojure.core/ex-info (core.clj:4617)

is this really the case? How else can I get the eigen decomposition of a matrix.

[mikera]

Use an implementation that has eigendecomposition implemented.

(set-current-implementation :vectorz) should work if you use the Vectorz implementation. I think Clatrix works too.

[michaellindon]

I was using :vectorz implementation when this error message was produced.

[mikera]

Ah sorry I see that Eigendecomposition is not fully implemented yet for vectorz. There are complexities around the handling of complex numbers.... do you need compex eigenvalues?

[michaellindon]

I'd be more than happy with just real eigenvalues implemented for now :) That's all I need for my purposes

[drone29a]

bump

I too would like eigendecomposition for real matrices. It looks like clatrix is abandoned.

Or, is there a better supported core.matrix implementation you'd recommend switching to?

[drone29a]

Ah, and I see the issue with the Vectorz backend is that the library itself doesn't yet provide eigendecomposition. Was hoping it was as easy as uncommenting out the wrapper code in vectorz-clj.

[daniel-yj-yang]

eigen is still not working

An alternative is to use svd from clojure.core.matrix.linear One can transform the results from svd to eigendecomposition

Or, using clatrix will be fine too.