MAINT: stats.ContinuousDistribution: lower-precision RNG generation, …

…fix docs?

scipy/stats/_distribution_infrastructure.py

@@ -28,6 +28,8 @@ def _isnull(x):
 _NO_CACHE = "no_cache"
 
 # TODO:
+#  Test sample dtypes
+#  Add dtype kwarg (especially for distributions with no parameters)
 #  When drawing endpoint/out-of-bounds values of a parameter, draw them from
 #   the endpoints/out-of-bounds region of the full `domain`, not `typical`.
 #  Distributions without shape parameters probably need to accept a `dtype` parameter;
@@ -4206,15 +4208,16 @@ def sample(self, shape=(), *, method=None, rng=None, qmc_engine=None):
         rng = self._validate_rng(rng) or self.rng or np.random.default_rng()
 
         if qmc_engine is None:
-            return self._sample_dispatch(sample_shape, full_shape, method=method,
-                                         rng=rng, **self._parameters)
+            res = self._sample_dispatch(sample_shape, full_shape, method=method,
+                                        rng=rng, **self._parameters)
         else:
             # needs input validation for qrng
             d = int(np.prod(full_shape[1:]))
             length = full_shape[0] if full_shape else 1
             qrng = qmc_engine(d=d, seed=rng)
-            return self._qmc_sample_dispatch(length, full_shape, method=method,
-                                             qrng=qrng, **self._parameters)
+            res = self._qmc_sample_dispatch(length, full_shape, method=method,
+                                            qrng=qrng, **self._parameters)
+        return res.astype(self._dtype, copy=False)
 
     @_dispatch
     def _sample_dispatch(self, sample_shape, full_shape, *, method, rng, **params):
@@ -4229,7 +4232,7 @@ def _sample_formula(self, sample_shape, full_shape, *, rng, **params):
         raise NotImplementedError(self._not_implemented)
 
     def _sample_inverse_transform(self, sample_shape, full_shape, *, rng, **params):
-        uniform = rng.uniform(size=full_shape)
+        uniform = rng.random(size=full_shape, dtype=self._dtype)
         return self._icdf_dispatch(uniform, **params)
 
     @_dispatch
@@ -4246,7 +4249,7 @@ def _qmc_sample_formula(self, length, full_shape, *, qrng, **params):
 
     def _qmc_sample_inverse_transform(self, length, full_shape, *, qrng, **params):
         uniform = qrng.random(length)
-        uniform = np.reshape(uniform, full_shape)
+        uniform = np.reshape(uniform, full_shape).astype(self._dtype)
         return self._icdf_dispatch(uniform, **params)
 
     ### Moments