add OptimizationGradientError

botorch/exceptions/errors.py

@@ -72,3 +72,17 @@ def __init__(
         super().__init__(*args, **kwargs)
         self.current_x = current_x
         self.runtime = runtime
+
+
+class OptimizationGradientError(BotorchError):
+    r"""Exception raised when gradient array `gradf` containts NaNs."""
+
+    def __init__(self, /, *args: Any, current_x: np.ndarray, **kwargs: Any) -> None:
+        r"""
+        Args:
+            *args: Standard args to `BoTorchError`.
+            current_x: A numpy array representing the current iterate.
+            **kwargs: Standard kwargs to `BoTorchError`.
+        """
+        super().__init__(*args, **kwargs)
+        self.current_x = current_x

botorch/generation/gen.py

@@ -18,6 +18,7 @@
 import numpy as np
 import torch
 from botorch.acquisition import AcquisitionFunction
+from botorch.exceptions.errors import OptimizationGradientError
 from botorch.exceptions.warnings import OptimizationWarning
 from botorch.generation.utils import (
     _convert_nonlinear_inequality_constraints,
@@ -215,7 +216,7 @@ def f_np_wrapper(x: np.ndarray, f: Callable):
                 )
                 if initial_conditions.dtype != torch.double:
                     msg += " Consider using `dtype=torch.double`."
-                raise RuntimeError(msg)
+                raise OptimizationGradientError(msg, current_x=x)
             fval = loss.item()
             return fval, gradf
 

test/exceptions/test_errors.py

@@ -12,6 +12,7 @@
     CandidateGenerationError,
     DeprecationError,
     InputDataError,
+    OptimizationGradientError,
     OptimizationTimeoutError,
     UnsupportedError,
 )
@@ -49,3 +50,9 @@ def test_OptimizationTimeoutError(self):
         self.assertTrue(np.array_equal(error.current_x, np.array([1.0])))
         with self.assertRaises(OptimizationTimeoutError):
             raise error
+
+    def test_OptimizationGradientError(self):
+        error = OptimizationGradientError("message", current_x=np.array([1.0]))
+        self.assertTrue(np.array_equal(error.current_x, np.array([1.0])))
+        with self.assertRaises(OptimizationGradientError):
+            raise error

test/generation/test_gen.py

@@ -10,6 +10,7 @@
 
 import torch
 from botorch.acquisition import qExpectedImprovement, qKnowledgeGradient
+from botorch.exceptions.errors import OptimizationGradientError
 from botorch.exceptions.warnings import OptimizationWarning
 from botorch.fit import fit_gpytorch_mll
 from botorch.generation.gen import (
@@ -318,7 +319,7 @@ def test_gen_candidates_scipy_nan_handling(self):
             test_grad = torch.tensor([0.5, 0.2, float("nan")], **ckwargs)
             # test NaN in grad
             with mock.patch("torch.autograd.grad", return_value=[test_grad]):
-                with self.assertRaisesRegex(RuntimeError, expected_regex):
+                with self.assertRaisesRegex(OptimizationGradientError, expected_regex):
                     gen_candidates_scipy(
                         initial_conditions=test_ics,
                         acquisition_function=mock.Mock(return_value=test_ics),