Fix sample_points_around_best when using 20 dimensional inputs or `…

…prob_perturb` (#1290)

Summary:
Pull Request resolved: #1290

When using 20 dimensional inputs, `sample_points_around_best` would return `n // 2` samples, which would lead to errors when reshaping it in `gen_batch_initial_conditions`. Similarly, if one were to use `prob_perturb` with <20 dimensions, they would get `n + n // 2` samples. This changes the two `if` statements that control this to use the same conditional, eliminating the bug.

Reviewed By: Balandat

Differential Revision: D37705879

fbshipit-source-id: b6716dc934c3760cc00bd9d0c74305d6bf05ea27

botorch/optim/initializers.py

@@ -683,16 +683,17 @@ def sample_points_around_best(
             # the view() is to ensure that best_idcs is not a scalar tensor
             best_idcs = torch.topk(f_pred, n_best).indices.view(-1)
             best_X = X[best_idcs]
+    use_perturbed_sampling = best_X.shape[-1] >= 20 or prob_perturb is not None
     n_trunc_normal_points = (
-        n_discrete_points // 2 if best_X.shape[-1] >= 20 else n_discrete_points
+        n_discrete_points // 2 if use_perturbed_sampling else n_discrete_points
     )
     perturbed_X = sample_truncated_normal_perturbations(
         X=best_X,
         n_discrete_points=n_trunc_normal_points,
         sigma=sigma,
         bounds=bounds,
     )
-    if best_X.shape[-1] > 20 or prob_perturb is not None:
+    if use_perturbed_sampling:
         perturbed_subset_dims_X = sample_perturbed_subset_dims(
             X=best_X,
             bounds=bounds,

test/optim/test_initializers.py

@@ -710,12 +710,12 @@ def test_sample_points_around_best(self):
                     self.assertEqual(eq_mask[idcs].sum(), 4)
                 self.assertTrue((X_rnd >= 1).all())
                 self.assertTrue((X_rnd <= 2).all())
-            # test that subset_dims is called if d>=21
-            X_train = 1 + torch.rand(20, 21, **tkwargs)
+            # test that subset_dims is called if d>=20
+            X_train = 1 + torch.rand(10, 20, **tkwargs)
             model = MockModel(
                 MockPosterior(mean=(2 * X_train + 1).sum(dim=-1, keepdim=True))
             )
-            bounds = torch.ones(2, 21, **tkwargs)
+            bounds = torch.ones(2, 20, **tkwargs)
             bounds[1] = 2
             # test NEI with X_baseline
             acqf = qNoisyExpectedImprovement(
@@ -728,7 +728,7 @@ def test_sample_points_around_best(self):
                 X_rnd = sample_points_around_best(
                     acq_function=acqf, n_discrete_points=5, sigma=1e-3, bounds=bounds
                 )
-            self.assertTrue(X_rnd.shape, torch.Size([5, 2]))
+            self.assertEqual(X_rnd.shape, torch.Size([5, 20]))
             self.assertTrue((X_rnd >= 1).all())
             self.assertTrue((X_rnd <= 2).all())
             mock_subset_dims.assert_called_once()