Handle case where the PF set is empty

xopt/generators/bayesian/bayesian_generator.py

@@ -817,13 +817,21 @@ def get_pareto_front(self):
             (self.torch_reference_point.unsqueeze(0), objective_data)
         )
         var_data = torch.vstack(
-            (torch.zeros_like(variable_data[0]).unsqueeze(0), variable_data)
+            (
+                torch.full_like(variable_data[0], float("Nan")).unsqueeze(0),
+                variable_data,
+            )
         )
         non_dominated = is_non_dominated(obj_data)
+
         weights = set_botorch_weights(self.vocs).to(**self._tkwargs)
 
         # note need to undo weights for real number output
-        return var_data[non_dominated], obj_data[non_dominated] / weights
+        # only return values if non nan values exist
+        if torch.all(torch.isnan(var_data[non_dominated])):
+            return None, None
+        else:
+            return var_data[non_dominated], obj_data[non_dominated] / weights
 
 
 def formatted_base_docstring():

xopt/generators/bayesian/mobo.py

@@ -93,6 +93,12 @@ def _get_initial_conditions(self, n_candidates=1) -> Union[Tensor, None]:
                 num_restarts = self.numerical_optimizer.n_restarts
 
                 pf_locations, _ = self.get_pareto_front()
+
+                # if there is no pareto front just return None to revert back to
+                # default behavior
+                if pf_locations is None:
+                    return None
+
                 initial_points = torch.clone(pf_locations)
 
                 # add the q dimension

xopt/tests/generators/bayesian/test_mobo.py

@@ -159,6 +159,14 @@ def test_initial_conditions(self):
         assert len(initial_points) == 1
         gen.generate(1)
 
+        # try with no points on the pareto front
+        gen.reference_point = {"y1": 0.0, "y2": 0.0}
+        gen.numerical_optimizer.n_restarts = 20
+
+        initial_points = gen._get_initial_conditions()
+        assert initial_points is None
+        gen.generate(1)
+
     def test_log_mobo(self):
         evaluator = Evaluator(function=evaluate_TNK)
         reference_point = tnk_reference_point
@@ -175,6 +183,7 @@ def test_log_mobo(self):
             dump = ele.model_dump()
             generator = MOBOGenerator(vocs=tnk_vocs, **dump)
             X = Xopt(generator=generator, evaluator=evaluator, vocs=tnk_vocs)
+            X.generator.numerical_optimizer.max_iter = 1
             X.random_evaluate(3)
             X.step()