Switch code formatter to ruff (#2582)

Summary:
Pull Request resolved: #2582

Use the ruff code formatter. This diff

- Changes the Meta-internal config to use ruff.
- Updates `pyproject.toml` to use ruff as the code formatter.
- Applies the new formatting across the codebase.

This does NOT change the code sorting engine (that is still usort). We can update that in a next step.

Reviewed By: saitcakmak

Differential Revision: D64517061

fbshipit-source-id: bf9a4ed258b9489fc757c810fe8bbff0a5d018c4

Author: Balandat

botorch/acquisition/cached_cholesky.py

@@ -8,6 +8,7 @@
 Abstract class for acquisition functions leveraging a cached Cholesky
 decomposition of the posterior covariance over f(X_baseline).
 """
+
 from __future__ import annotations
 
 import warnings

botorch/acquisition/fixed_feature.py

@@ -93,7 +93,6 @@ def __init__(
         if isinstance(values, Tensor):
             new_values = values.detach().clone()
         else:
-
             dtype = get_dtype_of_sequence(values)
             device = get_device_of_sequence(values)
 

botorch/acquisition/input_constructors.py

@@ -1258,7 +1258,6 @@ def construct_inputs_qKG(
     }
 
     if with_current_value:
-
         X = _get_dataset_field(training_data, "X", first_only=True)
         _bounds = torch.as_tensor(bounds, dtype=X.dtype, device=X.device)
 
@@ -1868,7 +1867,6 @@ def _get_ref_point(
     objective_thresholds: Tensor,
     objective: MCMultiOutputObjective | None = None,
 ) -> Tensor:
-
     if objective is None:
         ref_point = objective_thresholds
     elif isinstance(objective, RiskMeasureMCObjective):

botorch/acquisition/multi_objective/analytic.py

@@ -16,7 +16,6 @@
 
 """
 
-
 from __future__ import annotations
 
 from itertools import product

botorch/acquisition/multi_objective/base.py

@@ -8,7 +8,6 @@
 Base classes for multi-objective acquisition functions.
 """
 
-
 from __future__ import annotations
 
 from abc import ABC, abstractmethod

botorch/acquisition/multi_objective/joint_entropy_search.py

@@ -15,6 +15,7 @@
     2022.
 
 """
+
 from __future__ import annotations
 
 from abc import abstractmethod

botorch/acquisition/multi_objective/logei.py

@@ -48,7 +48,6 @@
 class qLogExpectedHypervolumeImprovement(
     MultiObjectiveMCAcquisitionFunction, SubsetIndexCachingMixin
 ):
-
     _log: bool = True
 
     def __init__(
@@ -321,7 +320,6 @@ class qLogNoisyExpectedHypervolumeImprovement(
     NoisyExpectedHypervolumeMixin,
     qLogExpectedHypervolumeImprovement,
 ):
-
     _log: bool = True
 
     def __init__(

botorch/acquisition/prior_guided.py

@@ -15,6 +15,7 @@
     Augmenting Acquisition Functions with User Beliefs for Bayesian Optimization.
     ICLR 2022.
 """
+
 from __future__ import annotations
 
 from botorch.acquisition.acquisition import AcquisitionFunction

botorch/exceptions/warnings.py

@@ -7,6 +7,7 @@
 r"""
 Botorch Warnings.
 """
+
 import warnings
 
 

botorch/generation/utils.py

@@ -21,7 +21,7 @@
 
 
 def _convert_nonlinear_inequality_constraints(
-    nonlinear_inequality_constraints: list[Callable | tuple[Callable, bool]]
+    nonlinear_inequality_constraints: list[Callable | tuple[Callable, bool]],
 ) -> list[tuple[Callable, bool]]:
     """Convert legacy defintions of nonlinear inequality constraints into the new
     format. Assumes intra-point constraints.

botorch/models/fully_bayesian.py

@@ -30,7 +30,6 @@
     Seventh Conference on Uncertainty in Artificial Intelligence, 2021.
 """
 
-
 import math
 from abc import abstractmethod
 from collections.abc import Mapping

botorch/models/fully_bayesian_multitask.py

@@ -4,9 +4,7 @@
 # LICENSE file in the root directory of this source tree.
 
 
-r"""Multi-task Gaussian Process Regression models with fully Bayesian inference.
-"""
-
+r"""Multi-task Gaussian Process Regression models with fully Bayesian inference."""
 
 from collections.abc import Mapping
 from typing import Any, NoReturn

botorch/models/gp_regression_mixed.py

@@ -64,9 +64,8 @@ def __init__(
         train_Y: Tensor,
         cat_dims: list[int],
         train_Yvar: Tensor | None = None,
-        cont_kernel_factory: None | (
-            Callable[[torch.Size, int, list[int]], Kernel]
-        ) = None,
+        cont_kernel_factory: None
+        | (Callable[[torch.Size, int, list[int]], Kernel]) = None,
         likelihood: Likelihood | None = None,
         outcome_transform: OutcomeTransform | _DefaultType | None = DEFAULT,
         input_transform: InputTransform | None = None,  # TODO

botorch/models/kernels/contextual_lcea.py

@@ -291,7 +291,8 @@ def _eval_context_covar(self) -> Tensor:
         else:
             context_outputscales = self.outputscale_list * self.context_weight
         context_covar = (
-            (context_outputscales.unsqueeze(-2))  # (ns) x 1 x num_contexts
+            (context_outputscales.unsqueeze(-2))
+            # (ns) x 1 x num_contexts
             .mul(context_covar)
             .mul(context_outputscales.unsqueeze(-1))  # (ns) x num_contexts x 1
         )

botorch/models/transforms/input.py

@@ -13,6 +13,7 @@
 is typically part of a Model and applied within the model.forward()
 method.
 """
+
 from __future__ import annotations
 
 from abc import ABC, abstractmethod

botorch/optim/initializers.py

@@ -12,6 +12,7 @@
     surrogates and dynamic coordinate search in high-dimensional
     expensive black-box optimization, Engineering Optimization, 2013.
 """
+
 from __future__ import annotations
 
 import warnings

botorch/optim/parameter_constraints.py

@@ -396,7 +396,7 @@ def _generate_unfixed_nonlin_constraints(
     values = torch.tensor(list(fixed_features.values()), dtype=torch.double)
 
     def _wrap_nonlin_constraint(
-        constraint: Callable[[Tensor], Tensor]
+        constraint: Callable[[Tensor], Tensor],
     ) -> Callable[[Tensor], Tensor]:
         def new_nonlin_constraint(X: Tensor) -> Tensor:
             ivalues = values.to(X).expand(*X.shape[:-1], len(fixed_features))

botorch/optim/utils/timeout.py

@@ -42,7 +42,6 @@ def minimize_with_timeout(
     track of the runtime and the optimization variables at the current iteration.
     """
     if timeout_sec is not None:
-
         start_time = time.monotonic()
         callback_data = {"num_iterations": 0}  # update from withing callback below
 

botorch/posteriors/ensemble.py

@@ -68,7 +68,8 @@ def variance(self) -> Tensor:
         return self.values.var(dim=-3)
 
     def _extended_shape(
-        self, sample_shape: torch.Size = torch.Size()  # noqa: B008
+        self,
+        sample_shape: torch.Size = torch.Size(),  # noqa: B008
     ) -> torch.Size:
         r"""Returns the shape of the samples produced by the posterior with
         the given `sample_shape`.

botorch/posteriors/gpytorch.py

@@ -72,7 +72,8 @@ def batch_range(self) -> tuple[int, int]:
             return (0, -1)
 
     def _extended_shape(
-        self, sample_shape: torch.Size = torch.Size()  # noqa: B008
+        self,
+        sample_shape: torch.Size = torch.Size(),  # noqa: B008
     ) -> torch.Size:
         r"""Returns the shape of the samples produced by the posterior with
         the given `sample_shape`.

botorch/posteriors/higher_order.py

@@ -87,7 +87,8 @@ def batch_range(self) -> tuple[int, int]:
         return (0, -1)
 
     def _extended_shape(
-        self, sample_shape: torch.Size = torch.Size()  # noqa: B008
+        self,
+        sample_shape: torch.Size = torch.Size(),  # noqa: B008
     ) -> torch.Size:
         r"""Returns the shape of the samples produced by the posterior with
         the given `sample_shape`.

botorch/posteriors/posterior.py

@@ -108,7 +108,8 @@ def density(self, value: Tensor) -> Tensor:
         )  # pragma: no cover
 
     def _extended_shape(
-        self, sample_shape: torch.Size = torch.Size()  # noqa: B008
+        self,
+        sample_shape: torch.Size = torch.Size(),  # noqa: B008
     ) -> torch.Size:
         r"""Returns the shape of the samples produced by the posterior with
         the given `sample_shape`.

botorch/posteriors/posterior_list.py

@@ -108,7 +108,8 @@ def dtype(self) -> torch.dtype:
         return next(iter(dtypes))
 
     def _extended_shape(
-        self, sample_shape: torch.Size = torch.Size()  # noqa: B008
+        self,
+        sample_shape: torch.Size = torch.Size(),  # noqa: B008
     ) -> torch.Size:
         r"""Returns the shape of the samples produced by the posterior with
         the given `sample_shape`.

botorch/posteriors/torch.py

@@ -113,7 +113,8 @@ def density(self, value: Tensor) -> Tensor:
         return self.log_prob(value).exp()
 
     def _extended_shape(
-        self, sample_shape: torch.Size = torch.Size()  # noqa: B008
+        self,
+        sample_shape: torch.Size = torch.Size(),  # noqa: B008
     ) -> torch.Size:
         r"""Returns the shape of the samples produced by the distribution with
         the given `sample_shape`.

botorch/posteriors/transformed.py

@@ -70,7 +70,8 @@ def dtype(self) -> torch.dtype:
         return self._posterior.dtype
 
     def _extended_shape(
-        self, sample_shape: torch.Size = torch.Size()  # noqa: B008
+        self,
+        sample_shape: torch.Size = torch.Size(),  # noqa: B008
     ) -> torch.Size:
         r"""Returns the shape of the samples produced by the posterior with
         the given `sample_shape`.

botorch/sampling/pathwise/prior_samplers.py

@@ -62,7 +62,6 @@ def _draw_kernel_feature_paths_fallback(
     output_transform: TOutputTransform | None = None,
     weight_generator: Callable[[Size], Tensor] | None = None,
 ) -> GeneralizedLinearPath:
-
     # Generate a kernel feature map
     feature_map = map_generator(
         kernel=covar_module,

botorch/test_functions/multi_objective.py

@@ -1267,7 +1267,6 @@ def evaluate_slack_true(self, X: Tensor) -> Tensor:
 
 
 class C2DTLZ2(DTLZ2, ConstrainedBaseTestProblem):
-
     num_constraints = 1
     _r = 0.2
     # approximate from nsga-ii, TODO: replace with analytic

botorch/test_functions/synthetic.py

@@ -163,7 +163,6 @@ def evaluate_true(self, X: Tensor) -> Tensor:
 
 
 class Beale(SyntheticTestFunction):
-
     dim = 2
     _optimal_value = 0.0
     _bounds = [(-4.5, 4.5), (-4.5, 4.5)]
@@ -207,7 +206,6 @@ def evaluate_true(self, X: Tensor) -> Tensor:
 
 
 class Bukin(SyntheticTestFunction):
-
     dim = 2
     _bounds = [(-15.0, -5.0), (-3.0, 3.0)]
     _optimal_value = 0.0
@@ -241,7 +239,6 @@ def evaluate_true(self, X: Tensor) -> Tensor:
 
 
 class DropWave(SyntheticTestFunction):
-
     dim = 2
     _bounds = [(-5.12, 5.12), (-5.12, 5.12)]
     _optimal_value = -1.0
@@ -256,7 +253,6 @@ def evaluate_true(self, X: Tensor) -> Tensor:
 
 
 class DixonPrice(SyntheticTestFunction):
-
     _optimal_value = 0.0
 
     def __init__(
@@ -639,7 +635,6 @@ def evaluate_true(self, X: Tensor) -> Tensor:
 
 
 class Rastrigin(SyntheticTestFunction):
-
     _optimal_value = 0.0
 
     def __init__(
@@ -764,7 +759,6 @@ def evaluate_true(self, X: Tensor) -> Tensor:
 
 
 class SixHumpCamel(SyntheticTestFunction):
-
     dim = 2
     _bounds = [(-3.0, 3.0), (-2.0, 2.0)]
     _optimal_value = -1.0316
@@ -815,7 +809,6 @@ def evaluate_true(self, X: Tensor) -> Tensor:
 
 
 class ThreeHumpCamel(SyntheticTestFunction):
-
     dim = 2
     _bounds = [(-5.0, 5.0), (-5.0, 5.0)]
     _optimal_value = 0.0

botorch/test_utils/mock.py

@@ -8,6 +8,7 @@
 Utilities for speeding up optimization in tests.
 
 """
+
 from __future__ import annotations
 
 from collections.abc import Generator

botorch/utils/constraints.py

@@ -19,7 +19,7 @@
 
 
 def get_outcome_constraint_transforms(
-    outcome_constraints: tuple[Tensor, Tensor] | None
+    outcome_constraints: tuple[Tensor, Tensor] | None,
 ) -> list[Callable[[Tensor], Tensor]] | None:
     r"""Create outcome constraint callables from outcome constraint tensors.
 

botorch/utils/multi_objective/box_decompositions/utils.py

@@ -6,7 +6,6 @@
 
 r"""Utilities for box decomposition algorithms."""
 
-
 import torch
 from botorch.exceptions.errors import BotorchTensorDimensionError, UnsupportedError
 from botorch.utils.multi_objective.pareto import is_non_dominated

botorch/utils/multi_objective/scalarization.py

@@ -14,6 +14,7 @@
     for expensive multiobjective optimization problems," in IEEE Transactions
     on Evolutionary Computation, vol. 10, no. 1, pp. 50-66, Feb. 2006.
 """
+
 from __future__ import annotations
 
 from collections.abc import Callable

botorch/utils/test_helpers.py

@@ -40,7 +40,6 @@
 
 
 def _get_mcmc_samples(num_samples: int, dim: int, infer_noise: bool, **tkwargs):
-
     mcmc_samples = {
         "lengthscale": 1 + torch.rand(num_samples, 1, dim, **tkwargs),
         "outputscale": 1 + torch.rand(num_samples, **tkwargs),

botorch/utils/testing.py

@@ -277,7 +277,8 @@ def batch_shape(self) -> torch.Size:
         raise NotImplementedError  # pragma: no cover
 
     def _extended_shape(
-        self, sample_shape: torch.Size = torch.Size()  # noqa: B008
+        self,
+        sample_shape: torch.Size = torch.Size(),  # noqa: B008
     ) -> torch.Size:
         return sample_shape + self.base_sample_shape
 

botorch/utils/transforms.py

@@ -269,7 +269,6 @@ def decorator(
         def decorated(
             acqf: AcquisitionFunction, X: Any, *args: Any, **kwargs: Any
         ) -> Any:
-
             # Allow using acquisition functions for other inputs (e.g. lists of strings)
             if not isinstance(X, Tensor):
                 return method(acqf, X, *args, **kwargs)
@@ -311,7 +310,7 @@ def decorated(
 
 
 def concatenate_pending_points(
-    method: Callable[[Any, Tensor], Any]
+    method: Callable[[Any, Tensor], Any],
 ) -> Callable[[Any, Tensor], Any]:
     r"""Decorator concatenating X_pending into an acquisition function's argument.
 

pyproject.toml

@@ -8,3 +8,6 @@ write_to = "./botorch/version.py"
 
 [tool.usort]
 first_party_detection = false
+
+[tool.ufmt]
+formatter = "ruff-api"

test/acquisition/multi_objective/test_hypervolume_knowledge_gradient.py

@@ -370,9 +370,7 @@ def test_evaluate_q_hvkg(self):
                         ModelListGP, "fantasize", return_value=mfm
                     ) as patch_f, mock.patch(
                         NO, new_callable=mock.PropertyMock
-                    ) as mock_num_outputs, warnings.catch_warnings(
-                        record=True
-                    ) as ws:
+                    ) as mock_num_outputs, warnings.catch_warnings(record=True) as ws:
                         mock_num_outputs.return_value = 3
                         qHVKG = acqf_class(
                             model=model,