Coerce AMR inputs to float (#604)

* add failing test case

* lint

* coerce datatypes

* move LacOperon model to main

* remove important parameter for LacOperon

* make distributional parameters optional

* lint

pyciemss/mira_integration/compiled_dynamics.py

@@ -102,7 +102,7 @@ def _compile_param_values_mira(
         elif isinstance(param_value, pyro.distributions.Distribution):
             values[param_name] = pyro.nn.PyroSample(param_value)
         elif isinstance(param_value, (numbers.Number, numpy.ndarray, torch.Tensor)):
-            values[param_name] = torch.as_tensor(param_value)
+            values[param_name] = torch.as_tensor(param_value, dtype=torch.float32)
         else:
             raise TypeError(f"Unknown parameter type: {type(param_value)}")
 
@@ -126,7 +126,7 @@ def _eval_deriv_mira(
     dX: State[torch.Tensor] = dict()
     for i, var in enumerate(src.variables.values()):
         k = get_name(var)
-        dX[k] = numeric_deriv[..., i]
+        dX[k] = numeric_deriv[..., i].float()
     return dX
 
 
@@ -146,7 +146,7 @@ def _eval_initial_state_mira(
     X: State[torch.Tensor] = dict()
     for i, var in enumerate(src.variables.values()):
         k = get_name(var)
-        X[k] = numeric_initial_state[..., i]
+        X[k] = numeric_initial_state[..., i].float()
     return X
 
 
@@ -179,7 +179,7 @@ def _eval_observables_mira(
     observables: State[torch.Tensor] = dict()
     for i, obs in enumerate(src.observables.values()):
         k = get_name(obs)
-        observables[k] = numeric_observables[..., i]
+        observables[k] = numeric_observables[..., i].float()
 
     return observables
 

tests/fixtures.py

@@ -30,13 +30,15 @@ def __init__(
         data_mapping: Dict[str, str] = {},
         data_mapped_to_observable: bool = False,
         optimize_kwargs: Dict[str, Any] = None,
+        has_distributional_parameters: bool = True,
     ):
         self.url = url
         self.important_parameter = important_parameter
         self.data_path = data_path
         self.data_mapping = data_mapping
         self.data_mapped_to_observable = data_mapped_to_observable
         self.optimize_kwargs = optimize_kwargs
+        self.has_distributional_parameters = has_distributional_parameters
 
 
 # See https://github.com/DARPA-ASKEM/Model-Representations/issues/62 for discussion of valid models.
@@ -85,7 +87,11 @@ def __init__(
     ModelFixture(
         os.path.join(MODELS_PATH, "LV_rabbits_wolves_model03_regnet.json"), "beta"
     ),
-    # ModelFixture(os.path.join(MODELS_PATH, "LV_goat_chupacabra_regnet.json"), "beta"),
+    ModelFixture(
+        os.path.join(MODELS_PATH, "LacOperon.json"),
+        "k_1",
+        has_distributional_parameters=False,
+    ),
 ]
 
 STOCKFLOW_MODELS = [

tests/test_interfaces.py

@@ -89,14 +89,16 @@ def setup_calibrate(model_fixture, start_time, end_time, logging_step_size):
 
 
 @pytest.mark.parametrize("sample_method", SAMPLE_METHODS)
-@pytest.mark.parametrize("model_url", MODEL_URLS)
+@pytest.mark.parametrize("model", MODELS)
 @pytest.mark.parametrize("start_time", START_TIMES)
 @pytest.mark.parametrize("end_time", END_TIMES)
 @pytest.mark.parametrize("logging_step_size", LOGGING_STEP_SIZES)
 @pytest.mark.parametrize("num_samples", NUM_SAMPLES)
 def test_sample_no_interventions(
-    sample_method, model_url, start_time, end_time, logging_step_size, num_samples
+    sample_method, model, start_time, end_time, logging_step_size, num_samples
 ):
+    model_url = model.url
+
     with pyro.poutine.seed(rng_seed=0):
         result1 = sample_method(
             model_url, end_time, logging_step_size, num_samples, start_time=start_time
@@ -115,7 +117,8 @@ def test_sample_no_interventions(
         check_result_sizes(result, start_time, end_time, logging_step_size, num_samples)
 
     check_states_match(result1, result2)
-    check_states_match_in_all_but_values(result1, result3)
+    if model.has_distributional_parameters:
+        check_states_match_in_all_but_values(result1, result3)
 
     if sample_method.__name__ == "dummy_ensemble_sample":
         assert "total_state" in result1.keys()