# Can't get sample\_prior\_predictive to work with missing values

**URL:** <https://discourse.pymc.io/t/cant-get-sample-prior-predictive-to-work-with-missing-values/7559>\
**Category:** Questions\
**Created:** [June 8, 2021, 10:46pm UTC](https://discourse.pymc.io/t/cant-get-sample-prior-predictive-to-work-with-missing-values/7559 "2021-06-08T22:46:54Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![brendan](https://avatars.discourse-cdn.com/v4/letter/b/50afbb/32.png) [@brendan](https://discourse.pymc.io/u/brendan)\
**Post date:** [June 8, 2021, 10:46pm UTC](https://discourse.pymc.io/t/cant-get-sample-prior-predictive-to-work-with-missing-values/7559/1 "2021-06-08T22:46:54Z")

</div>

Hi all,

I have a regression model with a number of missing values, and I am trying to conduct prior predictive checks using pm.sample\_prior\_predictive However, every time I am getting “ValueError: array is not broadcastable to correct shape” (full error is below). I have tried specifying just a few variables using the var\_names parameter, but I get the same error. I can sample from the posterior just fine. Can anyone help me with this or give a workaround? This is my first time using pymc3 or working with any Bayesian model. Thank you in advance!

Seems related: [Can't seem to `sample_prior_predictive` on model with missing value imputation](https://discourse.pymc.io/t/cant-seem-to-sample-prior-predictive-on-model-with-missing-value-imputation/7130)

```auto
ValueError: array is not broadcastable to correct shape
Apply node that caused the error: AdvancedIncSubtensor1{no_inplace,set}(TensorConstant{[1.00e+20 .. 9.00e+01]}, OOB Config_missing, TensorConstant{[0 1 3 .. 61 62 64]})
Toposort index: 10
Inputs types: [TensorType(float64, vector), TensorType(float64, vector), TensorType(int64, vector)]
Inputs shapes: [(72,), (72,), (14,)]
Inputs strides: [(8,), (8,), (8,)]
Inputs values: ['not shown', 'not shown', 'not shown']
Outputs clients: [[Elemwise{Composite{(i0 + (i1 * i2) + (i3 * i4) + (i5 * i6) + (i7 * i8) + (i9 * i10) + (i11 * i12) + (i13 * i14) + (i15 * i16) + (i17 * i18) + (i19 * i20) + (i21 * i22) + (maximum(maximum(maximum(maximum(maximum(maximum(maximum(maximum(maximum(maximum(i23, i24), i25), i26), i27), i28), i29), i30), i31), i32), i33) * i34))}}[(0, 24)](InplaceDimShuffle{x}.0, InplaceDimShuffle{x}.0, DockAfterRNI, Site_DC, InplaceDimShuffle{x}.0, Site_Edge, InplaceDimShuffle{x}.0, Site_RNG, InplaceDimShuffle{x}.0, Site_SuperNode, InplaceDimShuffle{x}.0, Sub Region_AMERWEST, InplaceDimShuffle{x}.0, Sub Region_APAC, InplaceDimShuffle{x}.0, Sub Region_EMEA, InplaceDimShuffle{x}.0, ClassType_Class B, InplaceDimShuffle{x}.0, ClassType_Class C, InplaceDimShuffle{x}.0, ClassType_Class E, InplaceDimShuffle{x}.0, TensorConstant{(1,) of 0.0}, OOB Config, MGFX Config, Azn Optical Config, Port Config for Optical, WAN Final Config, SWAN Final Config, WAN TER Final Config, WAN IER Final Config, Fabric Config, Optical Acceptance, InplaceDimShuffle{x}.0)]]

Backtrace when the node is created(use Theano flag traceback.limit=N to make it longer):
File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3145, in run_cell_async
has_raised = await self.run_ast_nodes(code_ast.body, cell_name,
File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3337, in run_ast_nodes
if (await self.run_code(code, result, async_=asy)):
File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3417, in run_code
exec(code_obj, self.user_global_ns, self.user_ns)
File "<ipython-input-20-572dafd8db9a>", line 23, in <module>
WOVars[feature] = pm.Normal(feature, mu=starting_mu, sigma=starting_sigma, observed=df[feature])
File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py", line 83, in __new__
return model.Var(name, dist, data, total_size, dims=dims)
File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/model.py", line 1112, in Var
var = ObservedRV(
File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/model.py", line 1737, in __init__
data = as_tensor(data, name, model, distribution)
File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/model.py", line 1683, in as_tensor
dataTensor = tt.set_subtensor(constant[data.mask.nonzero()], missing_values)

HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/theano/compile/function_module.py in __call__ (self, *args, **kwargs)
    902 outputs =\
--> 903 self.fn() if output_subset is None else\
    904 self.fn(output_subset=output_subset)
ValueError: array is not broadcastable to correct shape

During handling of the above exception, another exception occurred:
ValueError Traceback (most recent call last)
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py in _draw_value(param, point, givens, size)
    834 try:
--> 835 return dist_tmp.random(point=point, size=size)
    836 except (ValueError, TypeError):
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/continuous.py in random(self, point, size)
    512 """
--> 513 mu, tau, _ = draw_values([self.mu, self.tau, self.sigma],
    514 point=point, size=size)
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py in draw_values(params, point, size)
    694 )
--> 695 value = _draw_value(param,
    696 point=point,
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py in _draw_value(param, point, givens, size)
    875 func = _compile_theano_function(param, input_vars)
--> 876 output = func(*input_vals)
    877 return output
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/numpy/lib/function_base.py in __call__ (self, *args, **kwargs)
   2107 
-> 2108 return self._vectorize_call(func=func, args=vargs)
   2109 
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/numpy/lib/function_base.py in _vectorize_call(self, func, args)
   2181 if self.signature is not None:
-> 2182 res = self._vectorize_call_with_signature(func, args)
   2183 elif not args:
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/numpy/lib/function_base.py in _vectorize_call_with_signature(self, func, args)
   2222 for index in np.ndindex(*broadcast_shape):
-> 2223 results = func(*(arg[index] for arg in args))
   2224 
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/theano/compile/function_module.py in __call__ (self, *args, **kwargs)
    913 thunk = self.fn.thunks[self.fn.position_of_error]
--> 914 gof.link.raise_with_op(
    915 node=self.fn.nodes[self.fn.position_of_error],
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/theano/gof/link.py in raise_with_op(node, thunk, exc_info, storage_map)
    324 pass
--> 325 reraise(exc_type, exc_value, exc_trace)
    326 
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/six.py in reraise(tp, value, tb)
    717 if value. __traceback__ is not tb:
--> 718 raise value.with_traceback(tb)
    719 raise value
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/theano/compile/function_module.py in __call__ (self, *args, **kwargs)
    902 outputs =\
--> 903 self.fn() if output_subset is None else\
    904 self.fn(output_subset=output_subset)
ValueError: array is not broadcastable to correct shape
Apply node that caused the error: AdvancedIncSubtensor1{no_inplace,set}(TensorConstant{[1.00e+20 .. 9.00e+01]}, OOB Config_missing, TensorConstant{[0 1 3 .. 61 62 64]})
Toposort index: 10
Inputs types: [TensorType(float64, vector), TensorType(float64, vector), TensorType(int64, vector)]
Inputs shapes: [(72,), (72,), (14,)]
Inputs strides: [(8,), (8,), (8,)]
Inputs values: ['not shown', 'not shown', 'not shown']
Outputs clients: [[Elemwise{Composite{(i0 + (i1 * i2) + (i3 * i4) + (i5 * i6) + (i7 * i8) + (i9 * i10) + (i11 * i12) + (i13 * i14) + (i15 * i16) + (i17 * i18) + (i19 * i20) + (i21 * i22) + (maximum(maximum(maximum(maximum(maximum(maximum(maximum(maximum(maximum(maximum(i23, i24), i25), i26), i27), i28), i29), i30), i31), i32), i33) * i34))}}[(0, 24)](InplaceDimShuffle{x}.0, InplaceDimShuffle{x}.0, DockAfterRNI, Site_DC, InplaceDimShuffle{x}.0, Site_Edge, InplaceDimShuffle{x}.0, Site_RNG, InplaceDimShuffle{x}.0, Site_SuperNode, InplaceDimShuffle{x}.0, Sub Region_AMERWEST, InplaceDimShuffle{x}.0, Sub Region_APAC, InplaceDimShuffle{x}.0, Sub Region_EMEA, InplaceDimShuffle{x}.0, ClassType_Class B, InplaceDimShuffle{x}.0, ClassType_Class C, InplaceDimShuffle{x}.0, ClassType_Class E, InplaceDimShuffle{x}.0, TensorConstant{(1,) of 0.0}, OOB Config, MGFX Config, Azn Optical Config, Port Config for Optical, WAN Final Config, SWAN Final Config, WAN TER Final Config, WAN IER Final Config, Fabric Config, Optical Acceptance, InplaceDimShuffle{x}.0)]]

Backtrace when the node is created(use Theano flag traceback.limit=N to make it longer):
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3145, in run_cell_async
    has_raised = await self.run_ast_nodes(code_ast.body, cell_name,
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3337, in run_ast_nodes
    if (await self.run_code(code, result, async_=asy)):
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3417, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-20-572dafd8db9a>", line 23, in <module>
    WOVars[feature] = pm.Normal(feature, mu=starting_mu, sigma=starting_sigma, observed=df[feature])
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py", line 83, in __new__
    return model.Var(name, dist, data, total_size, dims=dims)
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/model.py", line 1112, in Var
    var = ObservedRV(
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/model.py", line 1737, in __init__
    data = as_tensor(data, name, model, distribution)
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/model.py", line 1683, in as_tensor
    dataTensor = tt.set_subtensor(constant[data.mask.nonzero()], missing_values)

HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.

During handling of the above exception, another exception occurred:
ValueError Traceback (most recent call last)
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/theano/compile/function_module.py in __call__ (self, *args, **kwargs)
    902 outputs =\
--> 903 self.fn() if output_subset is None else\
    904 self.fn(output_subset=output_subset)
ValueError: array is not broadcastable to correct shape

During handling of the above exception, another exception occurred:
ValueError Traceback (most recent call last)
<ipython-input-46-e140f9e8d68c> in <module>
      3 # # az.concat(trace, az.from_pymc3(posterior_predictive=post_pred), inplace=True)
      4 with model:
----> 5 prior = pm.sample_prior_predictive()
      6 az.plot_ppc(az.from_pymc3(prior=prior, model=model))
      7 
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/sampling.py in sample_prior_predictive(samples, model, vars, var_names, random_seed)
   1955 names = get_default_varnames(vars_, include_transformed=False)
   1956 # draw_values fails with auto-transformed variables. transform them later!
-> 1957 values = draw_values([model[name] for name in names], size=samples)
   1958 
   1959 data = {k: v for k, v in zip(names, values)}
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py in draw_values(params, point, size)
    649 # This may fail for autotransformed RVs, which don't
    650 # have the random method
--> 651 value = _draw_value(next_,
    652 point=point,
    653 givens=temp_givens,
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py in _draw_value(param, point, givens, size)
    841 # we don't want to store these drawn values to the context
    842 with _DrawValuesContextBlocker():
--> 843 val = np.atleast_1d(dist_tmp.random(point=point,
    844 size=None))
    845 # Sometimes point may change the size of val but not the
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/continuous.py in random(self, point, size)
    511 array
    512 """
--> 513 mu, tau, _ = draw_values([self.mu, self.tau, self.sigma],
    514 point=point, size=size)
    515 return generate_samples(stats.norm.rvs, loc=mu, scale=tau**-0.5,
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py in draw_values(params, point, size)
    693 drawn[(node, size)]
    694 )
--> 695 value = _draw_value(param,
    696 point=point,
    697 givens=givens.values(),
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py in _draw_value(param, point, givens, size)
    874 input_vals = []
    875 func = _compile_theano_function(param, input_vars)
--> 876 output = func(*input_vals)
    877 return output
    878 raise ValueError('Unexpected type in draw_value: %s' % type(param))
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/numpy/lib/function_base.py in __call__ (self, *args, **kwargs)
   2106 vargs.extend([kwargs[_n] for _n in names])
   2107 
-> 2108 return self._vectorize_call(func=func, args=vargs)
   2109 
   2110 def _get_ufunc_and_otypes(self, func, args):
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/numpy/lib/function_base.py in _vectorize_call(self, func, args)
   2180 """Vectorized call to `func` over positional `args`."""
   2181 if self.signature is not None:
-> 2182 res = self._vectorize_call_with_signature(func, args)
   2183 elif not args:
   2184 res = func()
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/numpy/lib/function_base.py in _vectorize_call_with_signature(self, func, args)
   2221 
   2222 for index in np.ndindex(*broadcast_shape):
-> 2223 results = func(*(arg[index] for arg in args))
   2224 
   2225 n_results = len(results) if isinstance(results, tuple) else 1
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/theano/compile/function_module.py in __call__ (self, *args, **kwargs)
    912 if hasattr(self.fn, 'thunks'):
    913 thunk = self.fn.thunks[self.fn.position_of_error]
--> 914 gof.link.raise_with_op(
    915 node=self.fn.nodes[self.fn.position_of_error],
    916 thunk=thunk,
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/theano/gof/link.py in raise_with_op(node, thunk, exc_info, storage_map)
    323 # extra long error message in that case.
    324 pass
--> 325 reraise(exc_type, exc_value, exc_trace)
    326 
    327 
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/six.py in reraise(tp, value, tb)
    716 value = tp()
    717 if value. __traceback__ is not tb:
--> 718 raise value.with_traceback(tb)
    719 raise value
    720 finally:
/anaconda/envs/azureml_py38/lib/python3.8/site-packages/theano/compile/function_module.py in __call__ (self, *args, **kwargs)
    901 try:
    902 outputs =\
--> 903 self.fn() if output_subset is None else\
    904 self.fn(output_subset=output_subset)
    905 except Exception:
ValueError: array is not broadcastable to correct shape
Apply node that caused the error: AdvancedIncSubtensor1{no_inplace,set}(TensorConstant{[1.00e+20 .. 9.00e+01]}, OOB Config_missing, TensorConstant{[0 1 3 .. 61 62 64]})
Toposort index: 10
Inputs types: [TensorType(float64, vector), TensorType(float64, vector), TensorType(int64, vector)]
Inputs shapes: [(72,), (72,), (14,)]
Inputs strides: [(8,), (8,), (8,)]
Inputs values: ['not shown', 'not shown', 'not shown']
Outputs clients: [[Elemwise{Composite{(i0 + (i1 * i2) + (i3 * i4) + (i5 * i6) + (i7 * i8) + (i9 * i10) + (i11 * i12) + (i13 * i14) + (i15 * i16) + (i17 * i18) + (i19 * i20) + (i21 * i22) + (maximum(maximum(maximum(maximum(maximum(maximum(maximum(maximum(maximum(maximum(i23, i24), i25), i26), i27), i28), i29), i30), i31), i32), i33) * i34))}}[(0, 24)](InplaceDimShuffle{x}.0, InplaceDimShuffle{x}.0, DockAfterRNI, Site_DC, InplaceDimShuffle{x}.0, Site_Edge, InplaceDimShuffle{x}.0, Site_RNG, InplaceDimShuffle{x}.0, Site_SuperNode, InplaceDimShuffle{x}.0, Sub Region_AMERWEST, InplaceDimShuffle{x}.0, Sub Region_APAC, InplaceDimShuffle{x}.0, Sub Region_EMEA, InplaceDimShuffle{x}.0, ClassType_Class B, InplaceDimShuffle{x}.0, ClassType_Class C, InplaceDimShuffle{x}.0, ClassType_Class E, InplaceDimShuffle{x}.0, TensorConstant{(1,) of 0.0}, OOB Config, MGFX Config, Azn Optical Config, Port Config for Optical, WAN Final Config, SWAN Final Config, WAN TER Final Config, WAN IER Final Config, Fabric Config, Optical Acceptance, InplaceDimShuffle{x}.0)]]

Backtrace when the node is created(use Theano flag traceback.limit=N to make it longer):
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3145, in run_cell_async
    has_raised = await self.run_ast_nodes(code_ast.body, cell_name,
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3337, in run_ast_nodes
    if (await self.run_code(code, result, async_=asy)):
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3417, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-20-572dafd8db9a>", line 23, in <module>
    WOVars[feature] = pm.Normal(feature, mu=starting_mu, sigma=starting_sigma, observed=df[feature])
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/distributions/distribution.py", line 83, in __new__
    return model.Var(name, dist, data, total_size, dims=dims)
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/model.py", line 1112, in Var
    var = ObservedRV(
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/model.py", line 1737, in __init__
    data = as_tensor(data, name, model, distribution)
  File "/anaconda/envs/azureml_py38/lib/python3.8/site-packages/pymc3/model.py", line 1683, in as_tensor
    dataTensor = tt.set_subtensor(constant[data.mask.nonzero()], missing_values)

HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.

```

---

<div class="post-metadata">

**Author:** ![jonsedar](https://yyz2.discourse-cdn.com/flex036/user_avatar/discourse.pymc.io/jonsedar/32/2590_2.png) [@jonsedar](https://discourse.pymc.io/u/jonsedar)\
**Post date:** [June 9, 2021, 4:54am UTC](https://discourse.pymc.io/t/cant-get-sample-prior-predictive-to-work-with-missing-values/7559/2 "2021-06-09T04:54:02Z")

</div>

Yep, that sounds the same as what I found a little while ago:

- Discourse post: [Can't seem to `sample_prior_predictive` on model with missing value imputation](https://discourse.pymc.io/t/cant-seem-to-sample-prior-predictive-on-model-with-missing-value-imputation/7130)
- My gist with MRE: [MRE\_missing\_values · GitHub](https://gist.github.com/jonsedar/070319334bcf033773cc3e9495c79ea0)
- The currently open issue and dicussion on Github: [Give NoDistribution a random method so that `sample_prior_predictive` can work · Issue #4598 · pymc-devs/pymc3 · GitHub](https://github.com/pymc-devs/pymc3/issues/4598)

As per @ricardoV94 's note, the issue seems avoidable by using `shape` rather than `dims`

---

<div class="post-metadata">

**Author:** ![brendan](https://avatars.discourse-cdn.com/v4/letter/b/50afbb/32.png) [@brendan](https://discourse.pymc.io/u/brendan)\
**Post date:** [June 9, 2021, 3:23pm UTC](https://discourse.pymc.io/t/cant-get-sample-prior-predictive-to-work-with-missing-values/7559/3 "2021-06-09T15:23:47Z")

</div>

Hi Jon, thanks for the reply. However, in my model I use neither shape nor dims. Do you know of any workaround? For now I am just checking the parameters of each prior manually, then sampling them with numpy and comparing to my data, but this could get very tedious if I make the model more complex.

---

<div class="post-metadata">

**Author:** ![jonsedar](https://yyz2.discourse-cdn.com/flex036/user_avatar/discourse.pymc.io/jonsedar/32/2590_2.png) [@jonsedar](https://discourse.pymc.io/u/jonsedar)\
**Post date:** [June 9, 2021, 8:47pm UTC](https://discourse.pymc.io/t/cant-get-sample-prior-predictive-to-work-with-missing-values/7559/4 "2021-06-09T20:47:43Z")

</div>

Hmmm, are you able to share the model code? Might make it easier to grok and suggest a solution.

In the meantime, have you tried explicitly stating the shape?

Cheers, Jon

---

<div class="post-metadata">

**Author:** ![brendan](https://avatars.discourse-cdn.com/v4/letter/b/50afbb/32.png) [@brendan](https://discourse.pymc.io/u/brendan)\
**Post date:** [June 10, 2021, 11:32pm UTC](https://discourse.pymc.io/t/cant-get-sample-prior-predictive-to-work-with-missing-values/7559/5 "2021-06-10T23:32:22Z")

</div>

Thanks for the reply @jonsedar. Here is a stripped down version of my code. It works fine with the pm.sample\_prior\_predictive line commented out. df[“OOB Config”] is my variable with missing values. I have tried specifying “shape = 1” but it just gives a slightly different error.

```auto
test = pm.Model()
with test:
    OOB = pm.Normal("OOB", mu=df["OOB Config"].mean(), sigma=10.0, observed=df["OOB Config"])
    noise = pm.Normal("eps", mu=0, sigma=1.0)
    y = pm.Deterministic("y", OOB + noise)
    trace = pm.sample(10)
    prior = pm.sample_prior_predictive()

```

Best,

-Brendan