More tests

nistats/tests/test_utils.py

@@ -6,6 +6,10 @@
 from ..utils import multiple_mahalanobis, z_score, multiple_fast_inv
 from nose.tools import assert_true
 from numpy.testing import assert_almost_equal, assert_array_almost_equal
+from ..utils import (matrix_rank, full_rank, pos_recipr)
+from nose.tools import (assert_true, assert_equal, assert_false,
+                        assert_raises)
+
 
 def test_z_score():
     p = np.random.rand(10)
@@ -41,6 +45,43 @@ def test_multiple_fast_inv():
     X_inv = multiple_fast_inv(X)
     assert_almost_equal(X_inv_ref, X_inv)
 
+def test_full_rank():
+    rng = np.random.RandomState(20110831)
+    X = rng.standard_normal((40,5))
+    # A quick rank check
+    assert_equal(matrix_rank(X), 5)
+    X[:,0] = X[:,1] + X[:,2]
+    assert_equal(matrix_rank(X), 4)
+    Y1 = full_rank(X)
+    assert_equal(Y1.shape, (40,4))
+    Y2 = full_rank(X, r=3)
+    assert_equal(Y2.shape, (40,3))
+    Y3 = full_rank(X, r=4)
+    assert_equal(Y3.shape, (40,4))
+    # Windows - there seems to be some randomness in the SVD result; standardize
+    # column signs before comparison
+    flipper = np.sign(Y1[0]) * np.sign(Y3[0])
+    assert_almost_equal(Y1, Y3 * flipper)
+
+
+def test_pos_recipr():
+    X = np.array([2,1,-1,0], dtype=np.int8)
+    eX = np.array([0.5,1,0,0])
+    Y = pos_recipr(X)
+    yield assert_array_almost_equal, Y, eX
+    yield assert_equal, Y.dtype.type, np.float64
+    X2 = X.reshape((2,2))
+    Y2 = pos_recipr(X2)
+    yield assert_array_almost_equal, Y2, eX.reshape((2,2))
+    # check that lists have arrived
+    XL = [0, 1, -1]
+    yield assert_array_almost_equal, pos_recipr(XL), [0, 1, 0]
+    # scalars
+    yield assert_equal, pos_recipr(-1), 0
+    yield assert_equal, pos_recipr(0), 0
+    yield assert_equal, pos_recipr(2), 0.5
+
+
 
 if __name__ == "__main__":
     import nose