Merge pull request #7 from ADicksonLab/min_dist

Efficient minimum distance calculations

requirements.in

@@ -2,4 +2,5 @@
 
 numpy
 scipy
-pint
+pint
+scikit-learn

setup.py

@@ -27,6 +27,7 @@
     'numpy',
     'scipy',
     'pint',
+    'scikit-learn'
 ]
 
 # extras requirements list

src/geomm/distance.py

@@ -1,5 +1,28 @@
 import numpy as np
 import scipy.spatial.distance as dist
+from sklearn.neighbors import KDTree
 
 def distance():
     return None
+
+def minimum_distance(coordsA, coordsB):
+    """Calculate the minimum distance between members of coordsA and coordsB.
+    Uses a fast binary search algorithm of order N*log(N).
+
+    Parameters
+    ----------
+
+    coordsA : arraylike, shape (Natoms_A, 3)
+        First set of coordinates.
+
+    coordsB : arraylike, shape (Natoms_B, 3)
+        Second set of coordinates.
+
+    """
+
+    # make sure the number of dimensions is 3
+    assert (coordsA.shape[1] == 3) and (coordsB.shape[1] == 3), \
+        "Minimum distance expecting arrays of shape (N, 3)"
+
+    tree = KDTree(coordsA)
+    return(tree.query(coordsB, dualtree=False, k=1)[0].min())