MAINT: stats.Mixture: make return type consistent

Author: mdhaber

scipy/stats/_distribution_infrastructure.py

@@ -4506,13 +4506,13 @@ def _sum(self, fun, *args):
         out = self._full(0, *args)
         for var, weight in zip(self._components, self._weights):
             out += getattr(var, fun)(*args) * weight
-        return out
+        return out[()]
 
     def _logsum(self, fun, *args):
         out = self._full(-np.inf, *args)
         for var, log_weight in zip(self._components, np.log(self._weights)):
             np.logaddexp(out, getattr(var, fun)(*args) + log_weight, out=out)
-        return out
+        return out[()]
 
     def support(self):
         a = self._full(np.inf)
@@ -4588,7 +4588,7 @@ def _moment_raw(self, order):
         out = self._full(0)
         for var, weight in zip(self._components, self._weights):
             out += var.moment(order, kind='raw') * weight
-        return out
+        return out[()]
 
     def _moment_central(self, order):
         order = int(order)
@@ -4599,7 +4599,7 @@ def _moment_central(self, order):
             a, b = var.mean(), self.mean()
             moment = var._moment_transform_center(order, moment_as, a, b)
             out += moment * weight
-        return out
+        return out[()]
 
     def _moment_standardized(self, order):
         return self._moment_central(order) / self.standard_deviation()**order

scipy/stats/tests/test_continuous.py

@@ -1794,12 +1794,18 @@ def test_input_validation(self):
         with pytest.raises(ValueError, match=message):
             Mixture([Normal(), Normal()], weights=[1.5, -0.5])
 
-    def test_basic(self):
+    @pytest.mark.parametrize('shape', [(), (10,)])
+    def test_basic(self, shape):
         rng = np.random.default_rng(582348972387243524)
         X = Mixture((Normal(mu=-0.25, sigma=1.1), Normal(mu=0.5, sigma=0.9)),
                     weights=(0.4, 0.6))
         Y = MixedDist()
-        x = rng.random(10)
+        x = rng.random(shape)
+
+        def assert_allclose(res, ref, **kwargs):
+            if shape == ():
+                assert np.isscalar(res)
+            np.testing.assert_allclose(res, ref, **kwargs)
 
         assert_allclose(X.logentropy(), Y.logentropy())
         assert_allclose(X.entropy(), Y.entropy())