Replace RVs by respective value variables in the graph of untransform…

…ed variables

This fixes incorrect behavior, where deterministic projection from transformed (sampling) to untransformed space would be nonsensical when transforms depend on other graph variables

pymc/model.py

@@ -900,10 +900,11 @@ def value_vars(self):
     @property
     def unobserved_value_vars(self):
         """List of all random variables (including untransformed projections),
-        as well as deterministics used as inputs and outputs of the the model's
+        as well as deterministics used as inputs and outputs of the model's
         log-likelihood graph
         """
         vars = []
+        untransformed_vars = []
         for rv in self.free_RVs:
             value_var = self.rvs_to_values[rv]
             transform = getattr(value_var.tag, "transform", None)
@@ -912,13 +913,16 @@ def unobserved_value_vars(self):
                 # each transformed variable
                 untrans_value_var = transform.backward(value_var, *rv.owner.inputs)
                 untrans_value_var.name = rv.name
-                vars.append(untrans_value_var)
+                untransformed_vars.append(untrans_value_var)
             vars.append(value_var)
 
+        # Remove rvs from untransformed values graph
+        untransformed_vars, _ = rvs_to_value_vars(untransformed_vars, apply_transforms=True)
+
         # Remove rvs from deterministics graph
         deterministics, _ = rvs_to_value_vars(self.deterministics, apply_transforms=True)
 
-        return vars + deterministics
+        return vars + untransformed_vars + deterministics
 
     @property
     def basic_RVs(self):

pymc/tests/test_sampling.py

@@ -23,6 +23,7 @@
 import numpy as np
 import numpy.testing as npt
 import pytest
+import scipy.special
 
 from aesara import shared
 from arviz import InferenceData
@@ -326,6 +327,16 @@ def test_deterministic_of_unobserved(self):
 
         np.testing.assert_allclose(idata.posterior["y"], idata.posterior["x"] + 100)
 
+    def test_transform_with_rv_depenency(self):
+        # Test that untransformed variables that depend on upstream variables are properly handled
+        with pm.Model() as m:
+            x = pm.HalfNormal("x", observed=1)
+            transform = pm.transforms.IntervalTransform(lambda *inputs: (inputs[-2], inputs[-1]))
+            y = pm.Uniform("y", lower=0, upper=x, transform=transform)
+            trace = pm.sample(tune=10, draws=50, return_inferencedata=False, random_seed=336)
+
+        assert np.allclose(scipy.special.expit(trace["y_interval__"]), trace["y"])
+
 
 def test_sample_find_MAP_does_not_modify_start():
     # see https://github.com/pymc-devs/pymc/pull/4458