Merge pull request #152 from optimas-org/bug/fix_data_type

Fix bug with parameter type in Ax generators

optimas/evaluators/base.py

@@ -67,11 +67,11 @@ def get_sim_specs(
             # May be a 1D array.
             "in": [var.name for var in varying_parameters],
             "out": (
-                [(obj.name, float) for obj in objectives]
+                [(obj.name, obj.dtype) for obj in objectives]
                 # f is the single float output that LibEnsemble minimizes.
                 + [(par.name, par.dtype) for par in analyzed_parameters]
                 # input parameters
-                + [(var.name, float) for var in varying_parameters]
+                + [(var.name, var.dtype) for var in varying_parameters]
             ),
             "user": {
                 "n_procs": self._n_procs,

optimas/generators/ax/service/base.py

@@ -188,6 +188,7 @@ def _create_ax_parameters(self) -> List:
                     "bounds": [var.lower_bound, var.upper_bound],
                     "is_fidelity": var.is_fidelity,
                     "target_value": var.fidelity_target_value,
+                    "value_type": var.dtype.__name__,
                 }
             )
             if var.is_fixed:

tests/test_ax_generators.py

@@ -104,6 +104,46 @@ def test_ax_single_fidelity():
     np.save("./tests_output/ax_sf_history", exploration._libe_history.H)
 
 
+def test_ax_single_fidelity_int():
+    """
+    Test that an exploration with a single-fidelity generator runs
+    correctly with an integer parameter.
+    """
+
+    var1 = VaryingParameter("x0", -50.0, 5.0, dtype=int)
+    var2 = VaryingParameter("x1", -5.0, 15.0)
+    obj = Objective("f", minimize=False)
+
+    gen = AxSingleFidelityGenerator(
+        varying_parameters=[var1, var2], objectives=[obj]
+    )
+    ev = FunctionEvaluator(function=eval_func_sf)
+    exploration = Exploration(
+        generator=gen,
+        evaluator=ev,
+        max_evals=10,
+        sim_workers=2,
+        exploration_dir_path="./tests_output/test_ax_single_fidelity_int",
+    )
+
+    # Get reference to original AxClient.
+    ax_client = gen._ax_client
+    assert ax_client.experiment.search_space.parameters["x0"].python_type == int
+
+    # Run exploration.
+    exploration.run()
+
+    # Check that the generator has been updated.
+    assert gen.n_completed_trials == exploration.history.shape[0]
+
+    # Check that the original ax client has been updated.
+    n_ax_trials = ax_client.get_trials_data_frame().shape[0]
+    assert n_ax_trials == exploration.history.shape[0]
+
+    # Check correct variable type.
+    assert exploration.history["x0"].to_numpy().dtype == int
+
+
 def test_ax_single_fidelity_moo():
     """
     Test that an exploration with a multi-objective single-fidelity generator
@@ -246,6 +286,9 @@ def test_ax_single_fidelity_updated_params():
     # Update range of x0 and run 10 evals.
     var1.update_range(-20.0, 0.0)
     gen.update_parameter(var1)
+    # Make sure we have an evaluation in the new range (it currently fails
+    # otherwise).
+    exploration.evaluate_trials([{"x0": -10.0, "x1": 10.0}])
     exploration.run(n_evals=10)
     assert all(exploration.history["x0"][-10:] >= -20)
     assert all(exploration.history["x0"][-10:] <= 0.0)
@@ -567,6 +610,7 @@ def test_ax_service_init():
 
 if __name__ == "__main__":
     test_ax_single_fidelity()
+    test_ax_single_fidelity_int()
     test_ax_single_fidelity_moo()
     test_ax_single_fidelity_fb()
     test_ax_single_fidelity_moo_fb()