Update tests

pymc/sampling/forward.py

@@ -820,7 +820,7 @@ def sample_posterior_predictive(
     else:
         vars_ = model.observed_RVs + observed_dependent_deterministics(model)
         if observed_data is not None:
-            vars_ += [model[x] for x in observed_data if x in model]
+            vars_ += [model[x] for x in observed_data if x in model and x not in vars_]
 
     vars_to_sample = list(get_default_varnames(vars_, include_transformed=False))
 

tests/sampling/test_forward.py

@@ -481,16 +481,6 @@ def test_normal_scalar(self):
                 chains=nchains,
             )
 
-        # test that trace is used in ppc
-        with pm.Model() as model_ppc:
-            mu = pm.Normal("mu", 0.0, 1.0)
-            a = pm.Normal("a", mu=mu, sigma=1)
-
-        ppc = pm.sample_posterior_predictive(
-            trace=trace, model=model_ppc, return_inferencedata=False
-        )
-        assert "a" in ppc
-
         with model:
             # test list input
             ppc0 = pm.sample_posterior_predictive(
@@ -550,6 +540,51 @@ def test_normal_scalar_idata(self):
             ppc = pm.sample_posterior_predictive(idata, return_inferencedata=False)
             assert ppc["a"].shape == (nchains, ndraws)
 
+    def test_external_trace(self):
+        nchains = 2
+        ndraws = 500
+        with pm.Model() as model:
+            mu = pm.Normal("mu", 0.0, 1.0)
+            a = pm.Normal("a", mu=mu, sigma=1, observed=0.0)
+            trace = pm.sample(
+                draws=ndraws,
+                chains=nchains,
+            )
+
+        # test that trace is used in ppc
+        with pm.Model() as model_ppc:
+            mu = pm.Normal("mu", 0.0, 1.0)
+            a = pm.Normal("a", mu=mu, sigma=1)
+
+        ppc = pm.sample_posterior_predictive(
+            trace=trace, model=model_ppc, return_inferencedata=False
+        )
+        assert list(ppc.keys()) == ["a"]
+
+    @pytest.mark.xfail(reason="Auto-imputation of variables not supported in this setting")
+    def test_external_trace_det(self):
+        nchains = 2
+        ndraws = 500
+        with pm.Model() as model:
+            mu = pm.Normal("mu", 0.0, 1.0)
+            a = pm.Normal("a", mu=mu, sigma=1, observed=0.0)
+            b = pm.Deterministic("b", a + 1)
+            trace = pm.sample(
+                draws=ndraws,
+                chains=nchains,
+            )
+
+        # test that trace is used in ppc
+        with pm.Model() as model_ppc:
+            mu = pm.Normal("mu", 0.0, 1.0)
+            a = pm.Normal("a", mu=mu, sigma=1)
+            b = pm.Deterministic("b", a + 1)
+
+        ppc = pm.sample_posterior_predictive(
+            trace=trace, model=model_ppc, return_inferencedata=False
+        )
+        assert list(ppc.keys()) == ["a", "b"]
+
     def test_normal_vector(self):
         with pm.Model() as model:
             mu = pm.Normal("mu", 0.0, 1.0)