Create test_utils.py

tests/test_utils.py

@@ -0,0 +1,85 @@
+import pytest
+import numpy as np
+
+from eo_tides.utils import idw
+
+
+# Test Inverse Distance Weighted function
+def test_idw():
+    # Basic psuedo-1D example
+    input_z = [1, 2, 3, 4, 5]
+    input_x = [0, 1, 2, 3, 4]
+    input_y = [0, 0, 0, 0, 0]
+    output_x = [0.5, 1.5, 2.5, 3.5]
+    output_y = [0.0, 0.0, 0.0, 0.0]
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=2)
+    assert np.allclose(out, [1.5, 2.5, 3.5, 4.5])
+
+    # Verify that k > input points gives error
+    with pytest.raises(ValueError):
+        idw(input_z, input_x, input_y, output_x, output_y, k=6)
+
+    # 2D nearest neighbour case
+    input_z = [1, 2, 3, 4]
+    input_x = [0, 4, 0, 4]
+    input_y = [0, 0, 4, 4]
+    output_x = [1, 4, 0, 3]
+    output_y = [0, 1, 3, 4]
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=1)
+    assert np.allclose(out, [1, 2, 3, 4])
+
+    # Two neighbours
+    input_z = [1, 2, 3, 4]
+    input_x = [0, 4, 0, 4]
+    input_y = [0, 0, 4, 4]
+    output_x = [2, 0, 4, 2]
+    output_y = [0, 2, 2, 4]
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=2)
+    assert np.allclose(out, [1.5, 2, 3, 3.5])
+
+    # Four neighbours
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=4)
+    assert np.allclose(out, [2.11, 2.30, 2.69, 2.88], rtol=0.01)
+
+    # Four neighbours; max distance of 2
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=4, max_dist=2)
+    assert np.allclose(out, [1.5, 2, 3, 3.5])
+
+    # Four neighbours; max distance of 2, k_min of 4 (should return NaN)
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=4, max_dist=2, k_min=4)
+    assert np.isnan(out).all()
+
+    # Four neighbours; power function p=0
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=4, p=0)
+    assert np.allclose(out, [2.5, 2.5, 2.5, 2.5])
+
+    # Four neighbours; power function p=2
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=4, p=2)
+    assert np.allclose(out, [1.83, 2.17, 2.83, 3.17], rtol=0.01)
+
+    # Different units, nearest neighbour case
+    input_z = [10, 20, 30, 40]
+    input_x = [1125296, 1155530, 1125296, 1155530]
+    input_y = [-4169722, -4169722, -4214782, -4214782]
+    output_x = [1124952, 1159593, 1120439, 1155284]
+    output_y = [-4169749, -4172892, -4211108, -4214332]
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=1)
+    assert np.allclose(out, [10, 20, 30, 40])
+
+    # Verify distance works on different units
+    output_x = [1142134, 1138930]
+    output_y = [-4171232, -4213451]
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=4, max_dist=20000)
+    assert np.allclose(out, [15, 35], rtol=0.1)
+
+    # Test multidimensional input
+    input_z = np.column_stack(([1, 2, 3, 4], [10, 20, 30, 40]))
+    input_x = [0, 4, 0, 4]
+    input_y = [0, 0, 4, 4]
+    output_x = [1, 4, 0, 3]
+    output_y = [0, 1, 3, 4]
+    out = idw(input_z, input_x, input_y, output_x, output_y, k=1)
+    assert input_z.shape == out.shape
+    assert np.allclose(out[:, 0], [1, 2, 3, 4])
+    assert np.allclose(out[:, 1], [10, 20, 30, 40])
+