Very valuable library

[Sylvain Bougerel]

On Nov 29, 2007 1:52 AM, Luis M. de la Cruz Salas <luiggix at gmail.com> wrote:
> Hi all,
> I have a set of random points in 2D (and eventually in 3D). I need to
> find the nearest points to a definite centre (it could be part of my
> original set of points) inside a circle or a sphere of this centre.
> libkdtree++ is what I'm looking for, but I'm not an expert, could
> someone give suggestions in using this library for my purposes?. I'm
> developing a library for RBFs (Partial differential equations stuff).
>

Hi Luis,

This is the typical use case of the library. And it is illustrated in
the file test_kdtree.cpp:
~/libkdtree_0.6.2/exemples/test_kdtree.cpp
The function that find the nearest point to a definite point is
rightfully named find_nearest().

I recommend you register on the mailing list, and lookout for the next
update (probably during the week end). find_nearest() is undergoing a
massive overhaul resulting in good performance improvements, and
results of highest quality. (The current algorithm sometime fails to
find the real nearest points).

>
> PS. Months ago I was looking for a library of this kind in BOOST, but
> there is not such. I think many people would be happy if you propose
> libkdtree++ to BOOST.
>

I'll be happy too if the library was making it's way into boost. And,
eventually, we'll probably propose it. However there are still a lot
of things to improve in the library before I feel ready to propose it.
:)

If you have anymore question, or if you do not understand what
test_kdtree.cpp is doing, do not hesitate to ask any questions!

- Sylvain