fix TimeAsFeature transform for MOO (#3178)

Summary:
Pull Request resolved: #3178

see title. For MOO, fixed features can be empty in L931 in ax/modelbridge/torch.py. This led to an issue where `start_time` was not in the parameters dict, causing issues when untransforming. This fixes the issue

Reviewed By: Balandat

Differential Revision: D67216689

fbshipit-source-id: 4f7985d9cbd9e20c23c4c94a47185eb0bdc72c7b

ax/modelbridge/transforms/tests/test_time_as_feature_transform.py

@@ -105,6 +105,12 @@ def test_TransformObservationFeatures(self) -> None:
             obsf_trans[0],
             ObservationFeatures({"x": 2.5, "duration": 0.5, "start_time": 5.0}),
         )
+        # test untransforming observation features that do not have
+        # start/end time (important for fixed features in MOO when un-
+        # transforming objective thresholds)
+        obsf_trans = [ObservationFeatures({"x": 2.5})]
+        obsf_untrans = self.t.untransform_observation_features(obsf_trans)
+        self.assertEqual(obsf_untrans, obsf_trans)
 
     def test_TransformSearchSpace(self) -> None:
         ss2 = deepcopy(self.search_space)

ax/modelbridge/transforms/time_as_feature.py

@@ -22,8 +22,7 @@
 from ax.models.types import TConfig
 from ax.utils.common.logger import get_logger
 from ax.utils.common.timeutils import unixtime_to_pandas_ts
-from ax.utils.common.typeutils import checked_cast
-from pyre_extensions import none_throws
+from pyre_extensions import assert_is_instance, none_throws
 
 if TYPE_CHECKING:
     # import as module to make sphinx-autodoc-typehints happy
@@ -139,12 +138,14 @@ def untransform_observation_features(
         self, observation_features: list[ObservationFeatures]
     ) -> list[ObservationFeatures]:
         for obsf in observation_features:
-            start_time = checked_cast(float, obsf.parameters.pop("start_time"))
-            obsf.start_time = unixtime_to_pandas_ts(start_time)
-            obsf.end_time = unixtime_to_pandas_ts(
-                checked_cast(float, obsf.parameters.pop("duration"))
-                * self.duration_range
-                + self.min_duration
-                + start_time
-            )
+            start_time = obsf.parameters.pop("start_time", None)
+            duration = obsf.parameters.pop("duration", None)
+            if start_time is not None:
+                start_time = assert_is_instance(start_time, float)
+                obsf.start_time = unixtime_to_pandas_ts(start_time)
+                if duration is not None:
+                    duration = assert_is_instance(duration, float)
+                    obsf.end_time = unixtime_to_pandas_ts(
+                        duration * self.duration_range + self.min_duration + start_time
+                    )
         return observation_features