Add Travis files

Author: vmarkovtsev

.travis.yml

@@ -0,0 +1,24 @@
+language: python
+sudo: false
+addons:
+  apt:
+    sources:
+    - ubuntu-toolchain-r-test
+    packages:
+    - gcc-5
+    - g++-5
+cache:
+  directories:
+    - $HOME/.cache/pip
+python:
+  - "3.4"
+  - "3.5"
+  - "3.6"
+install:
+  - pip install --upgrade pip
+  - pip install numpy scipy
+  - CC=gcc-5 CXX=g++-5 pip install -e .
+script:
+  - python3 -m unittest discover .
+notifications:
+  email: false

README.md

@@ -3,5 +3,44 @@
 Linear Assignmment Problem solver using Jonker-Volgenant algorithm
 ==================================================================
 
-This project is the rewrite of [pyLAPJV](https://github.com/hrldcpr/pyLAPJV) for
-Python 3 and updated core code.
+This project is the rewrite of [pyLAPJV](https://github.com/hrldcpr/pyLAPJV) which
+supports Python 3 and updated core code. It is a native Python 3 module and does
+not work with Python 2.x, use pyLAPJV if you are stuck in the previous decade.
+
+[Linear assignment problem](https://en.wikipedia.org/wiki/Assignment_problem)
+is the bijection between two sets with equal cardinality which optimizes the sum
+of the individual mapping costs taken from the fixed cost matrix. It naturally
+arises e.g. when we want to fit [t-SNE](https://lvdmaaten.github.io/tsne/) results
+into a rectangular regular grid.
+See this awesome notebook for the details why LAP matters:
+[CloudToGrid](https://github.com/kylemcdonald/CloudToGrid/blob/master/CloudToGrid.ipynb).
+
+Jonker-Volgenant algorithm is described in the paper:
+ 
+R. Jonker and A. Volgenant, "A Shortest Augmenting Path Algorithm for Dense and Sparse Linear Assignment Problems," _Computing_, vol. 38, pp. 325-340, 1987.
+
+It is faster in practice than the [Hungarian algorithm](https://en.wikipedia.org/wiki/Hungarian_algorithm),
+though it is not clear how complexities differ. The mentioned article is not publicly
+available, and spending money just to figure this out seems to be an overkill.
+
+The C++ source of the algorithm comes from http://www.magiclogic.com/assignment.html
+It has been reworked and partially optimized with OpenMP 4.0 SIMD.
+
+Installing
+----------
+```
+pip3 install lapjv
+```
+
+Usage
+-----
+Refer to [test.py](test.py) for the complete code.
+
+```
+from lapjv import lapjv
+row_ind, col_ind, _ = lapjv(cost_matrix)
+```
+
+License
+-------
+MIT.

setup.py

@@ -1,4 +1,5 @@
 from setuptools import setup, Extension
+import numpy
 
 setup(
     name="lapjv",
@@ -13,6 +14,7 @@
     ext_modules=[Extension("lapjv", sources=["python.cc"], extra_compile_args=[
         "-fopenmp", "-std=c++11", "-march=native", "-ftree-vectorize"])],
     install_requires=["numpy"],
+    include_dirs = [numpy.get_include()],
     classifiers=[
         "Development Status :: 4 - Beta",
         "Intended Audience :: Developers",