[BUG] Fixed incorrect bounding enforcement (#1871)

* fixed incorrect warping windows

* fixed distance tests that use window

aeon/classification/distance_based/tests/test_time_series_neighbors.py

@@ -33,7 +33,7 @@
     "euclidean": 19,
     "dtw": 21,
     "wdtw": 21,
-    "msm": 10,
+    "msm": 20,
     "erp": 19,
     "edr": 20,
     "lcss": 12,

aeon/clustering/tests/test_k_medoids.py

@@ -227,6 +227,6 @@ def test_custom_distance_params():
     # Test passing distance param
     default_dist = _get_model_centres(data, distance="msm")
     custom_params_dist = _get_model_centres(
-        data, distance="msm", distance_params={"window": 0.2}
+        data, distance="msm", distance_params={"window": 0.01}
     )
     assert not np.array_equal(default_dist, custom_params_dist)

aeon/distances/_edr.py

@@ -8,10 +8,7 @@
 from numba import njit
 from numba.typed import List as NumbaList
 
-from aeon.distances._alignment_paths import (
-    _add_inf_to_out_of_bounds_cost_matrix,
-    compute_min_return_path,
-)
+from aeon.distances._alignment_paths import compute_min_return_path
 from aeon.distances._bounding_matrix import create_bounding_matrix
 from aeon.distances._euclidean import _univariate_euclidean_distance
 from aeon.distances._utils import _convert_to_list, _is_multivariate
@@ -206,7 +203,15 @@ def _edr_cost_matrix(
     if epsilon is None:
         epsilon = float(max(np.std(x), np.std(y))) / 4
 
-    cost_matrix = np.zeros((x_size + 1, y_size + 1))
+    cost_matrix = np.full((x_size + 1, y_size + 1), np.inf)
+
+    for i in range(1, x_size + 1):
+        if bounding_matrix[i - 1, 0]:
+            cost_matrix[i, 0] = 0
+    for j in range(y_size):
+        if bounding_matrix[0, j - 1]:
+            cost_matrix[0, j] = 0
+    cost_matrix[0, 0] = 0
 
     for i in range(1, x_size + 1):
         for j in range(1, y_size + 1):
@@ -419,12 +424,8 @@ def edr_alignment_path(
     >>> edr_alignment_path(x, y)
     ([(0, 0), (1, 1), (2, 2), (3, 3)], 0.25)
     """
-    x_size = x.shape[-1]
-    y_size = y.shape[-1]
-    bounding_matrix = create_bounding_matrix(x_size, y_size, window, itakura_max_slope)
     cost_matrix = edr_cost_matrix(x, y, window, epsilon, itakura_max_slope)
-    # Need to do this because the cost matrix contains 0s and not inf in out of bounds
-    cost_matrix = _add_inf_to_out_of_bounds_cost_matrix(cost_matrix, bounding_matrix)
-    return compute_min_return_path(cost_matrix), float(
-        cost_matrix[x_size - 1, y_size - 1] / max(x_size, y_size)
+    return (
+        compute_min_return_path(cost_matrix),
+        cost_matrix[x.shape[-1] - 1, y.shape[-1] - 1] / max(x.shape[-1], y.shape[-1]),
     )

aeon/distances/_erp.py

@@ -8,10 +8,7 @@
 from numba import njit
 from numba.typed import List as NumbaList
 
-from aeon.distances._alignment_paths import (
-    _add_inf_to_out_of_bounds_cost_matrix,
-    compute_min_return_path,
-)
+from aeon.distances._alignment_paths import compute_min_return_path
 from aeon.distances._bounding_matrix import create_bounding_matrix
 from aeon.distances._euclidean import _univariate_euclidean_distance
 from aeon.distances._utils import _convert_to_list, _is_multivariate
@@ -209,13 +206,13 @@ def _erp_cost_matrix(
     x_size = x.shape[1]
     y_size = y.shape[1]
 
-    cost_matrix = np.zeros((x_size + 1, y_size + 1))
-
+    cost_matrix = np.full((x_size + 1, y_size + 1), np.inf)
     gx_distance, x_sum = _precompute_g(x, g, g_arr)
     gy_distance, y_sum = _precompute_g(y, g, g_arr)
 
     cost_matrix[1:, 0] = x_sum
     cost_matrix[0, 1:] = y_sum
+    cost_matrix[0, 0] = 0.0
 
     for i in range(1, x_size + 1):
         for j in range(1, y_size + 1):
@@ -458,12 +455,7 @@ def erp_alignment_path(
     >>> erp_alignment_path(x, y)
     ([(0, 0), (1, 1), (2, 2), (3, 3)], 2.0)
     """
-    bounding_matrix = create_bounding_matrix(
-        x.shape[-1], y.shape[-1], window, itakura_max_slope
-    )
-    cost_matrix = _add_inf_to_out_of_bounds_cost_matrix(
-        erp_cost_matrix(x, y, window, g, g_arr), bounding_matrix
-    )
+    cost_matrix = erp_cost_matrix(x, y, window, g, g_arr)
     return (
         compute_min_return_path(cost_matrix),
         cost_matrix[x.shape[-1] - 1, y.shape[-1] - 1],

aeon/distances/_msm.py

@@ -8,10 +8,7 @@
 from numba import njit
 from numba.typed import List as NumbaList
 
-from aeon.distances._alignment_paths import (
-    _add_inf_to_out_of_bounds_cost_matrix,
-    compute_min_return_path,
-)
+from aeon.distances._alignment_paths import compute_min_return_path
 from aeon.distances._bounding_matrix import create_bounding_matrix
 from aeon.distances._squared import _univariate_squared_distance
 from aeon.distances._utils import _convert_to_list, _is_multivariate
@@ -262,7 +259,7 @@ def _independent_cost_matrix(
 ) -> np.ndarray:
     x_size = x.shape[0]
     y_size = y.shape[0]
-    cost_matrix = np.zeros((x_size, y_size))
+    cost_matrix = np.full((x_size, y_size), np.inf)
     cost_matrix[0, 0] = np.abs(x[0] - y[0])
 
     for i in range(1, x_size):
@@ -293,7 +290,7 @@ def _msm_dependent_cost_matrix(
 ) -> np.ndarray:
     x_size = x.shape[1]
     y_size = y.shape[1]
-    cost_matrix = np.zeros((x_size, y_size))
+    cost_matrix = np.full((x_size, y_size), np.inf)
     cost_matrix[0, 0] = np.sum(np.abs(x[:, 0] - y[:, 0]))
 
     for i in range(1, x_size):
@@ -546,11 +543,8 @@ def msm_alignment_path(
     >>> msm_alignment_path(x, y)
     ([(0, 0), (1, 1), (2, 2), (3, 3)], 2.0)
     """
-    x_size = x.shape[-1]
-    y_size = y.shape[-1]
-    bounding_matrix = create_bounding_matrix(x_size, y_size, window, itakura_max_slope)
     cost_matrix = msm_cost_matrix(x, y, window, independent, c, itakura_max_slope)
-
-    # Need to do this because the cost matrix contains 0s and not inf in out of bounds
-    cost_matrix = _add_inf_to_out_of_bounds_cost_matrix(cost_matrix, bounding_matrix)
-    return compute_min_return_path(cost_matrix), cost_matrix[x_size - 1, y_size - 1]
+    return (
+        compute_min_return_path(cost_matrix),
+        cost_matrix[x.shape[-1] - 1, y.shape[-1] - 1],
+    )

aeon/distances/_twe.py

@@ -8,10 +8,7 @@
 from numba import njit
 from numba.typed import List as NumbaList
 
-from aeon.distances._alignment_paths import (
-    _add_inf_to_out_of_bounds_cost_matrix,
-    compute_min_return_path,
-)
+from aeon.distances._alignment_paths import compute_min_return_path
 from aeon.distances._bounding_matrix import create_bounding_matrix
 from aeon.distances._euclidean import _univariate_euclidean_distance
 from aeon.distances._utils import _convert_to_list, _is_multivariate
@@ -200,9 +197,8 @@ def _twe_cost_matrix(
 ) -> np.ndarray:
     x_size = x.shape[1]
     y_size = y.shape[1]
-    cost_matrix = np.zeros((x_size, y_size))
-    cost_matrix[0, 1:] = np.inf
-    cost_matrix[1:, 0] = np.inf
+    cost_matrix = np.full((x_size, y_size), np.inf)
+    cost_matrix[0, 0] = 0.0
 
     del_add = nu + lmbda
 
@@ -461,12 +457,7 @@ def twe_alignment_path(
     >>> twe_alignment_path(x, y)
     ([(0, 0), (1, 1), (2, 2), (3, 3)], 2.0)
     """
-    bounding_matrix = create_bounding_matrix(
-        x.shape[-1], y.shape[-1], window, itakura_max_slope
-    )
     cost_matrix = twe_cost_matrix(x, y, window, nu, lmbda, itakura_max_slope)
-    # Need to do this because the cost matrix contains 0s and not inf in out of bounds
-    cost_matrix = _add_inf_to_out_of_bounds_cost_matrix(cost_matrix, bounding_matrix)
     return (
         compute_min_return_path(cost_matrix),
         cost_matrix[x.shape[-1] - 1, y.shape[-1] - 1],

aeon/distances/tests/test_distance_correctness.py

@@ -50,12 +50,12 @@
     "squared": 384147.0,
     "dtw": [384147.0, 315012.0, 275854.0],
     "wdtw": [137927.0, 68406.15849, 2.2296],
-    "erp": [168.0, 1107.0, 2275.0],
+    "erp": [2275.0, 2275.0, 2275.0],
     "lcss": [1.0, 0.45833, 0.08333],
     "edr": [1.0, 0.58333, 0.125],
     "ddtw": [80806.0, 76289.0625, 76289.0625],
     "wddtw": [38144.53125, 19121.4927, 1.34957],
-    "twe": [137.001, 567.0029999999999, 3030.036000000001],
+    "twe": [4536.0, 3192.0220, 3030.036000000001],
     "msm_ind": [1515.0, 1517.8000000000004, 1557.0],  # msm with independent distance
     "msm_dep": [1897.0, 1898.6000000000001, 1921.0],  # msm with dependent distance
 }
@@ -65,16 +65,16 @@
     "dtw": [757.259719, 330.834497, 330.834497],
     "wdtw": [165.41724, 3.308425, 0],
     "msm": [70.014828, 89.814828, 268.014828],
-    "erp": [0.2086269, 2.9942540, 102.097904],
+    "erp": [169.3715, 102.0979, 102.097904],
     "edr": [1.0, 0.26, 0.07],
     "lcss": [1.0, 0.26, 0.05],
     "ddtw": [297.18771, 160.51311645984856, 160.29823],
     "wddtw": [80.149117, 1.458858, 0.0],
-    "twe": [1.001, 12.620531031063596, 173.3596688781867],
+    "twe": [338.4842162018424, 173.35966887818674, 173.3596688781867],
     # msm with independent distance
     "msm_ind": [84.36021099999999, 140.13788899999997, 262.6939920000001],
     # msm with dependent distance
-    "msm_dep": [33.068257441993, 71.14080843368329, 190.73978686253804],
+    "msm_dep": [33.06825, 71.1408, 190.7397],
 }
 
 

aeon/testing/expected_results/expected_distance_results.py

@@ -130,16 +130,16 @@
         [0.3409987716261595, 2.7187979513671015],
     ],
     "lcss": [[0.30000000000000004, 1.0], [0.4, 1.0], [0.4, 1.0]],
-    "edr": [[0.3, 0.3], [0.1, 0.1], [0.5, 1.0]],
+    "edr": [[0.7, 1.0], [1.0, 1.0], [0.5, 1.0]],
     "twe": [
-        [5.087449975445656, 15.161815735222117],
-        [1.1499446039354893, 5.995665808293953],
+        [20.507374214378885, 78.81976840746147],
+        [21.48930550350685, 82.55907793607852],
         [15.243281318807819, 77.81976840746147],
         [27.97089924329228, 83.97624505343292],
     ],
     "msm": [
-        [4.080245996952201, 43.583053575960584],
-        [1.0, 15.829914369482566],
+        [12.099213975730216, 92.75733240032741],
+        [12.153142672084059, 102.90914530531768],
         [12.023580258367444, 88.80013932627139],
         [7.115130579734542, 61.80633627614831],
     ],
@@ -150,8 +150,8 @@
     ],
     "sbd": [[0.13378563362841267, 0.12052110294129567]],
     "erp": [
-        [6.1963403666089425, 23.958805888780923],
-        [2.2271884807416047, 9.205416143392629],
+        [13.782010409379064, 44.3099600330504],
+        [13.782010409379064, 44.3099600330504],
         [12.782010409379064, 44.3099600330504],
         [15.460501609188993, 44.3099600330504],
     ],