Does anybody actually use logNormalize?
If so, what is supposed to be the definition?
I don't think that our implementation matches the javadoc and I don't think
the javadoc makes sense (to me).
Can somebody explain the original intent?
There are a few definitions which I could see make sense, but I can't tell
which is the intended one. None seem to be what we have.
a) v.normalize(p) gives a vector such that v.normalize(p).norm(p) == 1.
v.logNormalize(p) might log(1+v.normalize(p)).
b) v.logNormalize might be log(1+v).normalize(p)
Option (b) makes a lot of sense to me although I don't see why it warrants a
special method.
What we seem to have is
log(1 + v).normalize(p) / log(p)
Does anybody know why we divide by log(p)???
