compute_smoothed_constraint_indicator -> compute_smoothed_feasibility_indicator (#1935)

Summary:
Pull Request resolved: #1935

D47365085 introduced the aptly named `compute_feasibility_indicator`. This commit brings the smoothed counterpart in alignment with the naming convention.

Reviewed By: Balandat

Differential Revision: D47436246

fbshipit-source-id: 00be7f9fad99f1ce7e317e7b385629010e066d09

Author: SebastianAment

botorch/acquisition/input_constructors.py

@@ -482,7 +482,7 @@ def construct_inputs_qEI(
             are considered satisfied if the output is less than zero.
         eta: Temperature parameter(s) governing the smoothness of the sigmoid
             approximation to the constraint indicators. For more details, on this
-            parameter, see the docs of `compute_smoothed_constraint_indicator`.
+            parameter, see the docs of `compute_smoothed_feasibility_indicator`.
         ignored: Not used.
 
     Returns:
@@ -548,7 +548,7 @@ def construct_inputs_qNEI(
             are considered satisfied if the output is less than zero.
         eta: Temperature parameter(s) governing the smoothness of the sigmoid
             approximation to the constraint indicators. For more details, on this
-            parameter, see the docs of `compute_smoothed_constraint_indicator`.
+            parameter, see the docs of `compute_smoothed_feasibility_indicator`.
         ignored: Not used.
 
     Returns:
@@ -620,7 +620,7 @@ def construct_inputs_qPI(
             are considered satisfied if the output is less than zero.
         eta: Temperature parameter(s) governing the smoothness of the sigmoid
             approximation to the constraint indicators. For more details, on this
-            parameter, see the docs of `compute_smoothed_constraint_indicator`.
+            parameter, see the docs of `compute_smoothed_feasibility_indicator`.
         ignored: Not used.
 
     Returns:

botorch/acquisition/monte_carlo.py

@@ -42,7 +42,7 @@
 from botorch.exceptions.errors import UnsupportedError
 from botorch.models.model import Model
 from botorch.sampling.base import MCSampler
-from botorch.utils.objective import compute_smoothed_constraint_indicator
+from botorch.utils.objective import compute_smoothed_feasibility_indicator
 from botorch.utils.transforms import (
     concatenate_pending_points,
     match_batch_shape,
@@ -215,7 +215,7 @@ def __init__(
                 acquistion utilities, e.g. all improvement-based acquisition functions.
             eta: Temperature parameter(s) governing the smoothness of the sigmoid
                 approximation to the constraint indicators. For more details, on this
-                parameter, see the docs of `compute_smoothed_constraint_indicator`.
+                parameter, see the docs of `compute_smoothed_feasibility_indicator`.
         """
         if constraints is not None and isinstance(objective, ConstrainedMCObjective):
             raise ValueError(
@@ -305,7 +305,7 @@ def _apply_constraints(self, acqval: Tensor, samples: Tensor) -> Tensor:
                     "Constraint-weighting requires unconstrained "
                     "acquisition values to be non-negative."
                 )
-            acqval = acqval * compute_smoothed_constraint_indicator(
+            acqval = acqval * compute_smoothed_feasibility_indicator(
                 constraints=self._constraints, samples=samples, eta=self._eta
             )
         return acqval
@@ -366,7 +366,7 @@ def __init__(
                 are considered satisfied if the output is less than zero.
             eta: Temperature parameter(s) governing the smoothness of the sigmoid
                 approximation to the constraint indicators. For more details, on this
-                parameter, see the docs of `compute_smoothed_constraint_indicator`.
+                parameter, see the docs of `compute_smoothed_feasibility_indicator`.
         """
         super().__init__(
             model=model,
@@ -457,7 +457,7 @@ def __init__(
                 are considered satisfied if the output is less than zero.
             eta: Temperature parameter(s) governing the smoothness of the sigmoid
                 approximation to the constraint indicators. For more details, on this
-                parameter, see the docs of `compute_smoothed_constraint_indicator`.
+                parameter, see the docs of `compute_smoothed_feasibility_indicator`.
 
         TODO: similar to qNEHVI, when we are using sequential greedy candidate
         selection, we could incorporate pending points X_baseline and compute
@@ -671,7 +671,7 @@ def __init__(
                 scalar is less than zero.
             eta: Temperature parameter(s) governing the smoothness of the sigmoid
                 approximation to the constraint indicators. For more details, on this
-                parameter, see the docs of `compute_smoothed_constraint_indicator`.
+                parameter, see the docs of `compute_smoothed_feasibility_indicator`.
         """
         super().__init__(
             model=model,

botorch/acquisition/multi_objective/monte_carlo.py

@@ -57,7 +57,7 @@
 from botorch.utils.multi_objective.box_decompositions.utils import (
     _pad_batch_pareto_frontier,
 )
-from botorch.utils.objective import compute_smoothed_constraint_indicator
+from botorch.utils.objective import compute_smoothed_feasibility_indicator
 from botorch.utils.torch import BufferDict
 from botorch.utils.transforms import (
     concatenate_pending_points,
@@ -279,7 +279,7 @@ def _compute_qehvi(self, samples: Tensor, X: Optional[Tensor] = None) -> Tensor:
         obj = self.objective(samples, X=X)
         q = obj.shape[-2]
         if self.constraints is not None:
-            feas_weights = compute_smoothed_constraint_indicator(
+            feas_weights = compute_smoothed_feasibility_indicator(
                 constraints=self.constraints, samples=samples, eta=self.eta
             )  # `sample_shape x batch-shape x q`
         self._cache_q_subset_indices(q_out=q)
@@ -414,7 +414,7 @@ def __init__(
                 tensor the length of the tensor must match the number of provided
                 constraints. The i-th constraint is then estimated with the i-th
                 eta value. For more details, on this parameter, see the docs of
-                `compute_smoothed_constraint_indicator`.
+                `compute_smoothed_feasibility_indicator`.
             prune_baseline: If True, remove points in `X_baseline` that are
                 highly unlikely to be the pareto optimal and better than the
                 reference point. This can significantly improve computation time and

botorch/utils/objective.py

@@ -87,7 +87,7 @@ def apply_constraints_nonnegative_soft(
     Returns:
         A `n_samples x b x q (x m')`-dim tensor of feasibility-weighted objectives.
     """
-    w = compute_smoothed_constraint_indicator(
+    w = compute_smoothed_feasibility_indicator(
         constraints=constraints, samples=samples, eta=eta
     )
     if obj.dim() == samples.dim():
@@ -116,7 +116,7 @@ def compute_feasibility_indicator(
     return ind
 
 
-def compute_smoothed_constraint_indicator(
+def compute_smoothed_feasibility_indicator(
     constraints: List[Callable[[Tensor], Tensor]],
     samples: Tensor,
     eta: Union[Tensor, float],

test/utils/test_objective.py

@@ -9,7 +9,7 @@
 from botorch.utils import apply_constraints, get_objective_weights_transform
 from botorch.utils.objective import (
     compute_feasibility_indicator,
-    compute_smoothed_constraint_indicator,
+    compute_smoothed_feasibility_indicator,
 )
 from botorch.utils.testing import BotorchTestCase
 from torch import Tensor
@@ -196,14 +196,14 @@ def test_constraint_indicators(self):
         self.assertAllClose(ind, torch.zeros_like(ind))
         self.assertEqual(ind.dtype, torch.bool)
 
-        smoothed_ind = compute_smoothed_constraint_indicator(
+        smoothed_ind = compute_smoothed_feasibility_indicator(
             constraints=[zeros_f], samples=samples, eta=1e-3
         )
         self.assertAllClose(smoothed_ind, ones_f(samples) / 2)
 
         # two constraints
         samples = torch.randn(1)
-        smoothed_ind = compute_smoothed_constraint_indicator(
+        smoothed_ind = compute_smoothed_feasibility_indicator(
             constraints=[zeros_f, zeros_f],
             samples=samples,
             eta=1e-3,
@@ -218,13 +218,13 @@ def test_constraint_indicators(self):
         )
         self.assertAllClose(ind, torch.ones_like(ind))
 
-        smoothed_ind = compute_smoothed_constraint_indicator(
+        smoothed_ind = compute_smoothed_feasibility_indicator(
             constraints=[minus_one_f], samples=samples, eta=1e-3
         )
         self.assertTrue((smoothed_ind > 3 / 4).all())
 
         with self.assertRaisesRegex(ValueError, "Number of provided constraints"):
-            compute_smoothed_constraint_indicator(
+            compute_smoothed_feasibility_indicator(
                 constraints=[zeros_f, zeros_f],
                 samples=samples,
                 eta=torch.tensor([0.1], device=self.device),