[SciPy-dev] SciPy-Dev Digest, Vol 74, Issue 5

Jake VanderPlas jakevdp@gmail....
Thu Dec 3 13:08:46 CST 2009


>> >Hi Jake and SciPy,
>> >
>> >I wonder what is the status of this endeavor?
>> >
>> >For our little project we need fast search for a spherical neighborhood
>> >in a cloud of points.  Only now I discovered that scipy has KDTree and
>> >cKDTree which could be  used, BUT both interfaces are somewhat
>> >misfit for this simple goal one way or another.
>> >
>> >Also I've ran into libkdtree++ library with Python bindings which seems
>> >to do close what I need (just does range search, not sphere but it could
>> >be more or less efficiently computed post-hoc) and do it very
>> >efficiently upon my simple tests: http://libkdtree.alioth.debian.org/
>> >but it is under Artistic License 2.0...  but may be it might be of use
>> >to inspire modification of scipy's ways to interface with the user
>> >;)
>> >
>> >And Jake, how are you going along with your project since there were no
>> >follow ups on this thread I wonder if there was any progress?
>>
>> Yaroslav,
>> I have not done much more since the previous emails.  I got the
>> impression from people's responses that the community is not
>> interested in this code until it can be both more general and more
>> complete (allowing for flexible distance metrics, multiple data types,
>> approximate searches, alternate tree construction schemes, etc.)
>>
>
> This is a common problem in open source, propose something simple and you
> get asked to undertake world domination ;) I think simple Euclidean
> distances and double would be a good start as long as the design doesn't
> require a complete refactoring to add more general metrics. The code itself
> needs a lot of style fixes. If you don't have the time maybe Yaroslav can
> pick it up.
>
> Chuck

Chuck,
Can you give me some pointers on what style-fixes are needed?  Are you
referring to the C++ code, the python wrapper, or both?  I'm
relatively new to the python open-source world, and would love to gain
some experience in this sort of thing.  As far as future flexibility
goes, the C++ code is templated and able to handle pointers to custom
distance functions.  The current python wrapper is more rigid.
   -Jake


More information about the SciPy-Dev mailing list