Add braycurtis

Implements a Dask array function for computing the Bray-Curtis distance
between two 1-D arrays. Also tests it by comparing to the SciPy
implementation of the same name.

dask_distance/__init__.py

@@ -15,6 +15,38 @@
 from . import _utils
 
 
+#######################################
+#                                     #
+#  Numeric vector distance functions  #
+#                                     #
+#######################################
+
+
+@_utils._broadcast_uv_wrapper
+def braycurtis(u, v):
+    """
+    Finds the Bray-Curtis distance between two 1-D arrays.
+
+    .. math::
+
+       \\frac{ \sum_{i} \lvert u_{i} - v_{i} \\rvert }{ \sum_{i} \lvert u_{i} + v_{i} \\rvert }
+
+    Args:
+        u:           1-D array or collection of 1-D arrays
+        v:           1-D array or collection of 1-D arrays
+
+    Returns:
+        float:       Bray-Curtis distance
+    """
+
+    u = u.astype(float)
+    v = v.astype(float)
+
+    result = abs(u - v).sum(axis=-1) / abs(u + v).sum(axis=-1)
+
+    return result
+
+
 #####################################################
 #                                                   #
 #  Boolean vector distance/dissimilarity functions  #

tests/test_dask_distance.py

@@ -13,6 +13,92 @@
 import dask_distance
 
 
+@pytest.mark.parametrize(
+    "funcname", [
+        "braycurtis",
+    ]
+)
+@pytest.mark.parametrize("et, u, v", [
+    (ValueError, np.zeros((2,), dtype=bool), np.zeros((3,), dtype=bool)),
+    (ValueError, np.zeros((1, 2, 1,), dtype=bool), np.zeros((3,), dtype=bool)),
+    (ValueError, np.zeros((2,), dtype=bool), np.zeros((1, 3, 1,), dtype=bool)),
+])
+def test_1d_dist_err(funcname, et, u, v):
+    da_func = getattr(dask_distance, funcname)
+
+    with pytest.raises(et):
+        da_func(u, v)
+
+
+@pytest.mark.parametrize(
+    "funcname", [
+        "braycurtis",
+    ]
+)
+@pytest.mark.parametrize(
+    "seed", [
+        0,
+        137,
+        34,
+    ]
+)
+@pytest.mark.parametrize(
+    "size, chunks", [
+        (10, 5),
+    ]
+)
+def test_1d_dist(funcname, seed, size, chunks):
+    np.random.seed(seed)
+
+    a_u = 2 * np.random.random((size,)) - 1
+    a_v = 2 * np.random.random((size,)) - 1
+
+    d_u = da.from_array(a_u, chunks=chunks)
+    d_v = da.from_array(a_v, chunks=chunks)
+
+    sp_func = getattr(spdist, funcname)
+    da_func = getattr(dask_distance, funcname)
+
+    a_r = sp_func(a_u, a_v)
+    d_r = da_func(d_u, d_v)
+
+    assert np.allclose(np.array(d_r)[()], a_r, equal_nan=True)
+
+
+@pytest.mark.parametrize(
+    "funcname", [
+        "braycurtis",
+    ]
+)
+@pytest.mark.parametrize(
+    "seed", [
+        0,
+        137,
+        34,
+    ]
+)
+@pytest.mark.parametrize(
+    "u_shape, u_chunks, v_shape, v_chunks", [
+        ((2, 10), (1, 5), (3, 10), (1, 5)),
+    ]
+)
+def test_2d_dist(funcname, seed, u_shape, u_chunks, v_shape, v_chunks):
+    np.random.seed(seed)
+
+    a_u = 2 * np.random.random(u_shape) - 1
+    a_v = 2 * np.random.random(v_shape) - 1
+
+    d_u = da.from_array(a_u, chunks=u_chunks)
+    d_v = da.from_array(a_v, chunks=v_chunks)
+
+    da_func = getattr(dask_distance, funcname)
+
+    a_r = spdist.cdist(a_u, a_v, funcname)
+    d_r = da_func(d_u, d_v)
+
+    assert np.allclose(np.array(d_r)[()], a_r, equal_nan=True)
+
+
 @pytest.mark.parametrize(
     "funcname", [
         "dice",