Fixed numba incompatibility. Force covariance matrix input parameter …

…to be of floating data type.

skchange/costs/multivariate_t_cost.py

@@ -1060,10 +1060,14 @@ def _check_fixed_param(
     ) -> np.ndarray:
         """Check if the fixed mean parameter is valid.
 
+        The covariance matrix is checked for positive definiteness,
+        and forced to a floating point representation for numba compatibility.
+
         Parameters
         ----------
         param : 2-tuple of float or np.ndarray
             Fixed mean and covariance matrix for the cost calculation.
+            Both are converted to float values or float arrays.
         X : np.ndarray
             Input data.
 
@@ -1074,7 +1078,11 @@ def _check_fixed_param(
         """
         mean, cov = param
         mean = check_mean(mean, X)
-        cov = check_cov(cov, X)
+
+        # Require floating point representation of
+        # the covariance matrix for numba compatibility:
+        cov = check_cov(cov, X, force_float=True)
+
         return mean, cov
 
     @property

skchange/costs/tests/test_multivariate_t_cost.py

@@ -417,6 +417,7 @@ def test_scale_matrix_numba_benchmark(
                 t_dof,
                 abs_tol=1.0e-9,
                 rel_tol=0.0,
+                max_iter=100,
             )
 
         # Time numba version
@@ -426,6 +427,7 @@ def test_scale_matrix_numba_benchmark(
             t_dof,
             abs_tol=1.0e-9,
             rel_tol=0.0,
+            max_iter=100,
         )
         end = perf_counter()
         times_njit.append(end - start)
@@ -820,7 +822,7 @@ def test_iterative_mv_t_dof_estimate_returns_inf_for_high_initial_dof():
 
 def test_multivariate_t_log_likelihood_returns_nan_for_non_pos_def_scale_matrix():
     """Test that log likelihood returns np.nan for non-pos. def. scale matrix."""
-    non_positive_definite_matrix = np.array([[1, 2], [2, 1]])
+    non_positive_definite_matrix = np.array([[1, 2], [2, 1]], dtype=np.float64)
     centered_samples = np.random.randn(100, 2)
     dof = 5.0
 

skchange/costs/utils.py

@@ -56,7 +56,7 @@ def check_var(var: VarType, X: np.ndarray) -> np.ndarray:
     return var
 
 
-def check_cov(cov: CovType, X: np.ndarray) -> np.ndarray:
+def check_cov(cov: CovType, X: np.ndarray, force_float: bool = False) -> np.ndarray:
     """Check if the fixed covariance matrix parameter is valid.
 
     Parameters
@@ -65,6 +65,9 @@ def check_cov(cov: CovType, X: np.ndarray) -> np.ndarray:
         Fixed covariance matrix for the cost calculation.
     X : np.ndarray
         2d input data.
+    force_float : bool, default=False
+        If True, force the covariance matrix to be of
+        floating point data type.
 
     Returns
     -------
@@ -83,4 +86,8 @@ def check_cov(cov: CovType, X: np.ndarray) -> np.ndarray:
         )
     if not np.all(np.linalg.eigvals(cov) > 0):
         raise ValueError("covariance matrix must be positive definite.")
+
+    if force_float:
+        cov = cov.astype(float)
+
     return cov