TSTOOL home page | TSTOOL documentation page | TSTOOL link page

next up previous contents
Next: 6.14 nn_search Up: 6. Mex-Function Reference Previous: 6.12 largelyap - Compute

6.13 nn_prepare - Do nearest neighbor preprocessing

The intention of this mex-file was to reduce the computational overhead of preprocessing for nearest neighbor or range searching. With nn_prepare it is possible to do the preprocessing for a given point set only once and save the created tree structure into a Matlab variable. This Matlab variable, usually called atria, can then be used for repeated neighbor searches on the same point set. Most mex-files that rely on nearest neighbor or range search offer the possibility to use this variable atria as optional input argument. However, if the underlying point set is altered in any way, the proprocessing has to be repeated for the new point set. If the preprocessing output does not belong to the given point set, wrong results or program termination may occur.

Syntax:

Input arguments:

Example:

 
pointset = rand(40000, 3);
atria = nn_prepare(pointset);
[c, d] = corrsum(atria, pointset, 1:17:40000, 0.05, 0);
plot(log(d), log(c))
D = takens_estimator(atria, pointset, 1:17:40000, 0.05, 0)

next up previous contents
Next: 6.14 nn_search Up: 6. Mex-Function Reference Previous: 6.12 largelyap - Compute   Contents
TSTOOL

Copyright © 1997-2009 DPI Göttingen