Removing soft_eval_constraints

Summary: `soft_eval_constraints` has been deprecated for 7 months. This commit removed the function before the next BoTorch release.

Differential Revision: D54207128

botorch/utils/objective.py

@@ -10,8 +10,6 @@
 
 from __future__ import annotations
 
-import warnings
-
 from typing import Callable, List, Optional, Union
 
 import torch
@@ -182,32 +180,6 @@ def compute_smoothed_feasibility_indicator(
     return is_feasible if log else is_feasible.exp()
 
 
-# TODO: deprecate this function
-def soft_eval_constraint(lhs: Tensor, eta: float = 1e-3) -> Tensor:
-    r"""Element-wise evaluation of a constraint in a 'soft' fashion
-
-    `value(x) = 1 / (1 + exp(x / eta))`
-
-    Args:
-        lhs: The left hand side of the constraint `lhs <= 0`.
-        eta: The temperature parameter of the softmax function. As eta
-            decreases, this approximates the Heaviside step function.
-
-    Returns:
-        Element-wise 'soft' feasibility indicator of the same shape as `lhs`.
-        For each element `x`, `value(x) -> 0` as `x` becomes positive, and
-        `value(x) -> 1` as x becomes negative.
-    """
-    warnings.warn(
-        "`soft_eval_constraint` is deprecated. Please consider `torch.utils.sigmoid` "
-        + "with its `fat` and `log` options to compute feasibility indicators.",
-        DeprecationWarning,
-    )
-    if eta <= 0:
-        raise ValueError("eta must be positive.")
-    return torch.sigmoid(-lhs / eta)
-
-
 def apply_constraints(
     obj: Tensor,
     constraints: List[Callable[[Tensor], Tensor]],

test/utils/test_objective.py

@@ -10,7 +10,6 @@
 from botorch.utils.objective import (
     compute_feasibility_indicator,
     compute_smoothed_feasibility_indicator,
-    soft_eval_constraint,
 )
 from botorch.utils.testing import BotorchTestCase
 from torch import Tensor
@@ -75,15 +74,6 @@ def test_apply_constraints(self):
                 eta=0.0,
             )
 
-        # soft_eval_constraint is not in the path of apply_constraints, adding this test
-        # for coverage.
-        with self.assertRaisesRegex(ValueError, "eta must be positive."):
-            soft_eval_constraint(lhs=obj, eta=0.0)
-        ind = soft_eval_constraint(lhs=ones_f(samples), eta=1e-6)
-        self.assertAllClose(ind, torch.zeros_like(ind))
-        ind = soft_eval_constraint(lhs=-ones_f(samples), eta=1e-6)
-        self.assertAllClose(ind, torch.ones_like(ind))
-
     def test_apply_constraints_multi_output(self):
         # nonnegative objective, one constraint
         tkwargs = {"device": self.device}