INTERNAL:

v0.1

First internal release version of hyper.fit to accompany the Robotham & Obreschkow 2014 paper. Given to testers.

v0.2

Second internal release version of hyper.fit to accompany the Robotham & Obreschkow 2014 paper.

v0.2.1

Various minor fixes, improvements and document changes.

v1.0 (tried to release to CRAN, not accepted due to LaplacesDemon dependency)

Version posted to CRAN. Rejected!

v1.0.1

Fixed the passing of hyper.plot2d and hyper.plot3d outputs to the plot output for hyper.fit class objects.

Minor fixes and changes to the summary output.

v1.0.2

Added a direct unit vector search option to hyper.fit.

v1.0.3

Various fixes to version 1.0.2. Mostly to documentation.

RELEASE:

v1.0.0

Moved back to v1.0.0 since I want this one to be included with the paper.

v1.0.1

Document fixes and added MJB dataset.

Added examples from the hyper.fit paper.

v1.0.2

Minor change of k.vec usage based on referee comments:

X=t(t(X)-k.vec*scat.vec^2)-arrayvecmult(covarray,k.vec) ->

X=t(t(X)-k.vec*scat.vec^2)

DO has checked, and the latter should be correct.

v1.0.3 (on CRAN)

Minor fixes to accompany the paper release.

Added dots argument so that hyper.fit argumements are passed to optim, LA or LD.

v1.1.0

Added the option for a prior function that can take the parameters and should return a logged likelihood.

TO DO:

Need to check how to fix the sigma adjustment when including errors- current implementation in hyper.sigcor only appears to converge correctly when there is no error in the data. Dan and I are looking into this.

Perhaps add Nd generative Gaussian fitting option on top of the current (N-1)d plane + intrinsic scatter.

Longer term: add generic generative distributions.

Longer term: add correct treatment of truncations.