internal

meridian/constants.py

@@ -825,3 +825,9 @@
     'posterior_selected_draw_count_per_chain'
 )
 POSTERIOR_THINNING_SEED = 'posterior_thinning_seed'
+
+
+# Calibration constants.
+IS_CALIBRATED = 'is_calibrated'
+CALIBRATION_OUTPUTS = 'calibration_outputs'
+

meridian/model/prior_distribution.py

@@ -29,7 +29,6 @@
 from meridian import constants
 import numpy as np
 
-
 __all__ = [
     'IndependentMultivariateDistribution',
     'PriorDistribution',
@@ -92,7 +91,7 @@ class PriorDistribution:
   | `contribution_m`      | `n_media_channels`         |
   | `contribution_rf`     | `n_rf_channels`            |
   | `contribution_om`     | `n_organic_media_channels` |
-  | `contribution_orf`    | `n_organic_f_channels`     |
+  | `contribution_orf`    | `n_organic_rf_channels`     |
   | `contribution_n`      | `n_non_media_channels`     |
 
   (σ) `n_geos` if `unique_sigma_for_each_geo`, otherwise this is `1`
@@ -1206,9 +1205,23 @@ def distributions_are_equal(
     del a_params[constants.DISTRIBUTION]
     del b_params[constants.DISTRIBUTION]
 
+  if 'distributions' in a_params and 'distributions' in b_params:
+    a_dists = a_params['distributions']
+    b_dists = b_params['distributions']
+    if len(a_dists) != len(b_dists):
+      return False
+    for a_d, b_d in zip(a_dists, b_dists):
+      if not distributions_are_equal(a_d, b_d):
+        return False
+    del a_params['distributions']
+    del b_params['distributions']
+
   if constants.DISTRIBUTION in a_params or constants.DISTRIBUTION in b_params:
     return False
 
+  if 'distributions' in a_params or 'distributions' in b_params:
+    return False
+
   if a_params.keys() != b_params.keys():
     return False
 
@@ -1286,13 +1299,14 @@ def lognormal_dist_from_range(
   mass_percent = np.asarray(mass_percent)
 
   if not ((0.0 < low).all() and (low < high).all()):  # pytype: disable=attribute-error
-    raise ValueError("'low' and 'high' values must be non-negative and satisfy "
-                     "high > low.")
+    raise ValueError(
+        "'low' and 'high' values must be non-negative and satisfy high > low."
+    )
 
   if not ((0.0 < mass_percent).all() and (mass_percent < 1.0).all()):  # pytype: disable=attribute-error
     raise ValueError(
         "'mass_percent' values must be between 0 and 1, exclusive."
-        )
+    )
 
   normal = backend.tfd.Normal(0, 1)
   mass_lower = 0.5 - (mass_percent / 2)

meridian/model/prior_distribution_test.py

@@ -24,7 +24,6 @@
 from meridian.model import prior_distribution
 import numpy as np
 
-
 _N_GEOS = 10
 _N_GEOS_NATIONAL = 1
 _N_MEDIA_CHANNELS = 6
@@ -1529,6 +1528,30 @@ def test_get_total_media_contribution_prior(self):
           ),
           expected_result=True,
       ),
+      dict(
+          testcase_name='same_independent_multivariate_distributions',
+          get_a=lambda: prior_distribution.IndependentMultivariateDistribution([
+              backend.tfd.LogNormal(0.7, 0.4),
+              backend.tfd.Normal(0.0, 1.0),
+          ]),
+          get_b=lambda: prior_distribution.IndependentMultivariateDistribution([
+              backend.tfd.LogNormal(0.7, 0.4),
+              backend.tfd.Normal(0.0, 1.0),
+          ]),
+          expected_result=True,
+      ),
+      dict(
+          testcase_name='different_independent_multivariate_distributions',
+          get_a=lambda: prior_distribution.IndependentMultivariateDistribution([
+              backend.tfd.LogNormal(0.7, 0.4),
+              backend.tfd.Normal(0.0, 1.0),
+          ]),
+          get_b=lambda: prior_distribution.IndependentMultivariateDistribution([
+              backend.tfd.LogNormal(0.7, 0.4),
+              backend.tfd.Normal(0.0, 2.0),
+          ]),
+          expected_result=False,
+      ),
       dict(
           testcase_name='different_outer_complex_distributions',
           get_a=lambda: backend.tfd.BatchBroadcast(