[mlpack-git] (blog) master: Lozhnikov. Week 2 (11a2760)

[gitdub at mlpack.org]

Repository : https://github.com/mlpack/blog
On branch  : master
Link       : https://github.com/mlpack/blog/compare/b93431ae70a8cfa2c64d137200366f5cb7774f41...11a2760bafba4ed5a6453700addfb6a640c34deb

>---------------------------------------------------------------

commit 11a2760bafba4ed5a6453700addfb6a640c34deb
Author: Mikhail Lozhnikov <lozhnikovma at gmail.com>
Date:   Mon Jun 6 01:05:17 2016 +0300

    Lozhnikov. Week 2


>---------------------------------------------------------------

11a2760bafba4ed5a6453700addfb6a640c34deb
 content/blog/LozhnikovWeekTwo.md | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/content/blog/LozhnikovWeekTwo.md b/content/blog/LozhnikovWeekTwo.md
new file mode 100644
index 0000000..83d7ca1
--- /dev/null
+++ b/content/blog/LozhnikovWeekTwo.md
@@ -0,0 +1,12 @@
+Title: Implementation of tree types : Week 2
+Date: 2016-06-06 01:00:00
+Tags: gsoc, space trees, Hilbert R tree, Hilbert value
+Author: Mikhail Lozhnikov
+
+As I had planned in the previous blog post I spent last week testing, cleaning and refactoring the code. I added a lot of comments and some tests for the Hilbert R tree. Also I corrected a huge number of errors.
+
+Since I decided to use an armadillo object instead of the std::list object in order to store the Hilbert values of points I had to rewrite the DiscreteHilbertValue entirely with a template parameter TreeElemType. The TreeElemType parameter was introduced in order to define the best datatype (uint64_t for double and uint32_t for float) for the Hilbert value.
+
+In addition I removed the dataset variable from the RectangleTree class. After that RectangleTree has to use the local dataset instead of the global dataset. In order to complete the removal I have to correct all tree traversers, base cases and pruning rules. I'll do that soon.
+
+