Merge pull request #269 from lnccbrown/fix-posterior-predictive

Fixed a bug in sample_posterior_predictive()

src/hssm/distribution_utils/dist.py

@@ -191,6 +191,9 @@ def rng_fn(
                 size = args[-1]
                 args = args[:-1]
 
+            if size is None:
+                size = 1
+
             # Although we got around the ndims_supp issue, the size parameter passed
             # here is still an array with one element. We need to take it out.
             if not np.isscalar(size):
@@ -225,7 +228,7 @@ def rng_fn(
                 # All parameters are scalars
 
                 theta = np.stack(arg_arrays)
-                n_samples = 1 if not size else size
+                n_samples = size
             else:
                 # Preprocess all parameters, reshape them into a matrix of dimension
                 # (size, n_params) where size is the number of elements in the largest
@@ -240,7 +243,7 @@ def rng_fn(
                     [np.broadcast_to(arg, max_shape).reshape(-1) for arg in arg_arrays]
                 )
 
-                if size is None:
+                if size is None or size == 1:
                     n_samples = 1
                 elif size % new_data_size != 0:
                     raise ValueError(

tests/conftest.py

@@ -33,9 +33,9 @@ def data_angle():
 @pytest.fixture(scope="module")
 def data_ddm_reg():
     # Generate some fake simulation data
-    intercept = 0.3
-    x = np.random.uniform(0.5, 0.7, size=1000)
-    y = np.random.uniform(0.4, 0.1, size=1000)
+    intercept = 1.5
+    x = np.random.uniform(-5.0, 5.0, size=1000)
+    y = np.random.uniform(-5.0, 5.0, size=1000)
 
     v = intercept + 0.8 * x + 0.3 * y
     true_values = np.column_stack(

tests/test_mcmc.py

@@ -7,6 +7,8 @@
 
 def test_non_reg_models(data_ddm):
     model1 = hssm.HSSM(data_ddm)
+    model1.sample_prior_predictive(draws=10)
+
     model1.sample(cores=1, chains=1, tune=10, draws=10)
     model1.sample(sampler="nuts_numpyro", cores=1, chains=1, tune=10, draws=10)
 
@@ -18,6 +20,8 @@ def test_non_reg_models(data_ddm):
     model3.sample(cores=1, chains=1, tune=10, draws=10)
     model3.sample(cores=1, chains=1, tune=10, draws=10)
 
+    model1.sample_posterior_predictive(data=data_ddm.iloc[:10, :])
+
 
 def test_reg_models(data_ddm_reg):
     param_reg = dict(
@@ -30,6 +34,8 @@ def test_reg_models(data_ddm_reg):
     )
 
     model1 = hssm.HSSM(data_ddm_reg, v=param_reg)
+    model1.sample_prior_predictive(draws=10)
+
     model1.sample(cores=1, chains=1, tune=10, draws=10)
     model1.sample(sampler="nuts_numpyro", cores=1, chains=1, tune=10, draws=10)
 
@@ -43,6 +49,8 @@ def test_reg_models(data_ddm_reg):
     with pytest.raises(ValueError):
         model3.sample(sampler="nuts_numpyro", cores=1, chains=1, tune=10, draws=10)
 
+    model1.sample_posterior_predictive(data=data_ddm_reg.iloc[:10, :])
+
 
 def test_reg_models_a(data_ddm_reg):
     param_reg = dict(