Merge pull request #161 from ran-wei-verses/v1.0.0_alpha

flatten pB action dims for parameter learning with multi actions

pymdp/agent.py

@@ -180,7 +180,7 @@ def __init__(
                 policy_len,
                 control_fac_idx,
             )
-            B, self.action_maps = self._flatten_B_action_dims(B, self.B_action_dependencies)
+            B, pB, self.action_maps = self._flatten_B_action_dims(B, pB, self.B_action_dependencies)
             policies = self._construct_flattend_policies(policies_multi, self.action_maps)
             self.sampling_mode = "full"
 
@@ -575,26 +575,26 @@ def _construct_dependencies(self, A_dependencies, B_dependencies, B_action_depen
             B_action_dependencies = [[f] for f in range(self.num_factors)]
         return A_dependencies, B_dependencies, B_action_dependencies
 
-    def _flatten_B_action_dims(self, B, B_action_dependencies):
+    def _flatten_B_action_dims(self, B, pB, B_action_dependencies):
         assert hasattr(B[0], "shape"), "Elements of B must be tensors and have attribute shape"
         action_maps = []  # mapping from multi action dependencies to flat action dependencies for each B
         B_flat = []
+        pB_flat = []
         for i, (B_f, action_dependency) in enumerate(zip(B, B_action_dependencies)):
             if action_dependency == []:
                 B_flat.append(jnp.expand_dims(B_f, axis=-1))
+                if pB is not None:
+                    pB_flat.append(jnp.expand_dims(pB[i], axis=-1))
                 action_maps.append(
-                    {
-                        "multi_dependency": [],
-                        "multi_dims": [],
-                        "flat_dependency": [i],
-                        "flat_dims": [1],
-                    }
+                    {"multi_dependency": [], "multi_dims": [], "flat_dependency": [i], "flat_dims": [1]}
                 )
                 continue
 
             dims = [self.num_controls_multi[d] for d in action_dependency]
             target_shape = list(B_f.shape)[: -len(action_dependency)] + [pymath.prod(dims)]
             B_flat.append(B_f.reshape(target_shape))
+            if pB is not None:
+                pB_flat.append(pB[i].reshape(target_shape))
             action_maps.append(
                 {
                     "multi_dependency": action_dependency,
@@ -603,7 +603,9 @@ def _flatten_B_action_dims(self, B, B_action_dependencies):
                     "flat_dims": [pymath.prod(dims)],
                 }
             )
-        return B_flat, action_maps
+        if pB is None:
+            pB_flat = None
+        return B_flat, pB_flat, action_maps
 
     def _construct_flattend_policies(self, policies, action_maps):
         policies_flat = []