Added n-ary vector summation

appengine/findit/libs/math/test/vectors_test.py

Index: appengine/findit/libs/math/test/vectors_test.py
diff --git a/appengine/findit/libs/math/test/vectors_test.py b/appengine/findit/libs/math/test/vectors_test.py
new file mode 100644
index 0000000000000000000000000000000000000000..e6ccd3e758025bbff7b564acc09743fbaba10c1b
--- /dev/null
+++ b/appengine/findit/libs/math/test/vectors_test.py
@@ -0,0 +1,45 @@
+# Copyright 2016 The Chromium Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+import unittest
+import math
+import numpy as np
+
+from libs.math.vectors import vsum
+
+BIG = 1e100
+LITTLE = 1
+
+class VectorsTest(unittest.TestCase):

[Martin Barbella, 2016/12/02 19:52:37]

2 blank lines before this one. Does gpylint complain about anything else here?

[wrengr, 2016/12/02 20:54:46]

It really wants the numpy import to be grouped with the libs.* one... which
seems wrong. Also complains about the copyright header not saying Google Inc

[Martin Barbella, 2016/12/02 22:12:11]

Yeah, those are fine to ignore.

stgao, katesonia: Do we have any presubmit rules related to this? Might be
useful if not, and it would also allow you to ignore the rules we don't care
about.

[Sharu Jiang, 2016/12/02 22:26:12]

No, somehow the presubmit won't check the 2 empty lines before class/function. I
guess we need to modify some git tools in infra repo, though I am not quite sure
how to add those rules.

+
+  def setUp(self):
+    self._xs = [BIG, LITTLE, -BIG]
+
+  def testFsumKeepsPrecision(self):
+    """This test is to make sure that ``testVsum`` should actually work.
+
+    N.B., because ``math.fsum`` uses Shewchuk's algorithm whereas ``vsum``
+    uses Kahan's algorithm, the former is more precise than the latter. So
+    we need to exercise some caution when designing our tests.
+    """
+    # N.B., with precision loss, we'll get 0 rather than LITTLE.
+    self.assertEqual(LITTLE, math.fsum(self._xs))
+
+  def testVsumKeepsPrecision(self):

[Martin Barbella, 2016/12/02 19:52:37]

Please add docstrings for all tests. They're pretty handy in the event of
failures (unittest recognizes them).

[wrengr, 2016/12/02 20:54:46]

None of the other tests in the repo have them, but whatevs. Done.

[Martin Barbella, 2016/12/02 22:12:11]

Thanks. They're really helpful. We probably should have had them on the other
ones, but let's at least try to add them for new ones.

+    vs = [np.array([x, -x, 2*x]) for x in self._xs]
+    total = vsum(vs)
+    self.assertIsNotNone(total)
+
+    self.assertListEqual([LITTLE, -LITTLE, 2*LITTLE], total.tolist())
+
+  def testVsumEmptyWithoutShape(self):
+    self.assertIsNone(vsum([]))
+
+  def testVSumEmptyWithShape(self):
+    expected_shape = (3,)
+    total = vsum([], shape=expected_shape)
+    self.assertIsNotNone(total)
+    self.assertTupleEqual(expected_shape, total.shape)
+    self.assertListEqual(np.zeros(expected_shape).tolist(), total.tolist())
+