Split test_shared into test_model & test_sampling

.github/workflows/tests.yml

@@ -62,7 +62,6 @@ jobs:
           - |
             pymc/tests/tuning/test_scaling.py
             pymc/tests/tuning/test_starting.py
-            pymc/tests/test_shared.py
             pymc/tests/test_sampling.py
             pymc/tests/distributions/test_dist_math.py
             pymc/tests/distributions/test_transform.py

pymc/tests/test_model.py

@@ -1373,3 +1373,14 @@ def test_missing_symmetric():
     logp_inputs = list(graph_inputs([logp]))
     assert x_obs_vv in logp_inputs
     assert x_unobs_vv in logp_inputs
+
+
+class TestShared(SeededTest):
+    def test_deterministic(self):
+        with pm.Model() as model:
+            data_values = np.array([0.5, 0.4, 5, 2])
+            X = aesara.shared(np.asarray(data_values, dtype=aesara.config.floatX), borrow=True)
+            pm.Normal("y", 0, 1, observed=X)
+            assert np.all(
+                np.isclose(model.compile_logp(sum=False)({}), st.norm().logpdf(data_values))
+            )

pymc/tests/test_sampling.py

@@ -2302,3 +2302,37 @@ def test_float32(self):
                 with warnings.catch_warnings():
                     warnings.filterwarnings("ignore", ".*number of samples.*", UserWarning)
                     pm.sample(draws=10, tune=10, chains=1, step=sampler())
+
+
+class TestShared(SeededTest):
+    def test_sample(self):
+        x = np.random.normal(size=100)
+        y = x + np.random.normal(scale=1e-2, size=100)
+
+        x_pred = np.linspace(-3, 3, 200)
+
+        x_shared = aesara.shared(x)
+
+        with pm.Model() as model:
+            b = pm.Normal("b", 0.0, 10.0)
+            pm.Normal("obs", b * x_shared, np.sqrt(1e-2), observed=y, shape=x_shared.shape)
+            prior_trace0 = pm.sample_prior_predictive(1000)
+
+            idata = pm.sample(1000, tune=1000, chains=1)
+            pp_trace0 = pm.sample_posterior_predictive(idata)
+
+            x_shared.set_value(x_pred)
+            prior_trace1 = pm.sample_prior_predictive(1000)
+            pp_trace1 = pm.sample_posterior_predictive(idata)
+
+        assert prior_trace0.prior["b"].shape == (1, 1000)
+        assert prior_trace0.prior_predictive["obs"].shape == (1, 1000, 100)
+        np.testing.assert_allclose(
+            x, pp_trace0.posterior_predictive["obs"].mean(("chain", "draw")), atol=1e-1
+        )
+
+        assert prior_trace1.prior["b"].shape == (1, 1000)
+        assert prior_trace1.prior_predictive["obs"].shape == (1, 1000, 200)
+        np.testing.assert_allclose(
+            x_pred, pp_trace1.posterior_predictive["obs"].mean(("chain", "draw")), atol=1e-1
+        )