Require python 3.7 (#378)

Summary:
Pull Request resolved: #378

This allows us to use
- postponed evaluations of annotations (PEP 563)
- dataclasses (PEP 557)

Reviewed By: ldworkin

Differential Revision: D20113101

fbshipit-source-id: 34e5ae22e9f590894a4c0ab200139380a297a248

.circleci/config.yml

@@ -94,9 +94,9 @@ commands:
 
 jobs:
 
-  lint_test_py36_pip:
+  lint_test_py38_pip:
     docker:
-      - image: circleci/python:3.6.9
+      - image: circleci/python:3.8.1
     steps:
       - checkout
       - pip_install:
@@ -124,7 +124,7 @@ jobs:
 
   auto_deploy_site:
     docker:
-      - image: circleci/python:3.6.9-node
+      - image: circleci/python:3.8.1-node
     steps:
       - checkout
       - pip_install:
@@ -150,7 +150,7 @@ workflows:
 
   lint_and_test:
     jobs:
-      - lint_test_py36_pip:
+      - lint_test_py38_pip:
           filters: *exclude_ghpages_fbconfig
       - lint_test_py37_conda:
           filters: *exclude_ghpages_fbconfig

.conda/meta.yaml

@@ -13,7 +13,7 @@ build:
 
 requirements:
   host:
-    - python>=3.6
+    - python>=3.7
   run:
     - pytorch>=1.4
     - gpytorch>=1.0.0

CONTRIBUTING.md

@@ -34,7 +34,7 @@ Travis will fail on your PR if it does not adhere to the black formatting style.
 
 #### Type Hints
 
-BoTorch is fully typed using python 3.6+
+BoTorch is fully typed using python 3.7+
 [type hints](https://www.python.org/dev/peps/pep-0484/).
 We expect any contributions to also use proper type annotations. While we
 currently do not enforce full consistency of these in our continuous integration

README.md

@@ -49,7 +49,7 @@ Optimization simply use Ax.
 ## Installation
 
 **Installation Requirements**
-- Python >= 3.6
+- Python >= 3.7
 - PyTorch >= 1.4
 - gpytorch >= 1.0
 - scipy

botorch/acquisition/acquisition.py

@@ -8,6 +8,8 @@
 Abstract base module for all botorch acquisition functions.
 """
 
+from __future__ import annotations
+
 import warnings
 from abc import ABC, abstractmethod
 from typing import Optional

botorch/acquisition/active_learning.py

@@ -21,6 +21,8 @@
     ArXiv 2017.
 """
 
+from __future__ import annotations
+
 from typing import Optional
 
 from botorch import settings

botorch/acquisition/analytic.py

@@ -9,6 +9,8 @@
 Monte-Carlo sampling.
 """
 
+from __future__ import annotations
+
 from abc import ABC
 from copy import deepcopy
 from typing import Dict, Optional, Tuple, Union

botorch/acquisition/cost_aware.py

@@ -9,6 +9,8 @@
 To be used in a context where there is an objective/cost tradeoff.
 """
 
+from __future__ import annotations
+
 import warnings
 from abc import ABC, abstractmethod
 from typing import Any, Callable, Optional

botorch/acquisition/fixed_feature.py

@@ -9,6 +9,8 @@
 This is useful e.g. for performing contextual optimization.
 """
 
+from __future__ import annotations
+
 from typing import List, Union
 
 import torch

botorch/acquisition/knowledge_gradient.py

@@ -24,6 +24,8 @@
     bayesian optimization. NIPS 2016.
 """
 
+from __future__ import annotations
+
 from copy import deepcopy
 from typing import Callable, Optional, Tuple, Union
 

botorch/acquisition/max_value_entropy_search.py

@@ -21,6 +21,8 @@
     arXiv:1901.08275v1, 2019
 """
 
+from __future__ import annotations
+
 from copy import deepcopy
 from math import log
 from typing import Callable, Optional

botorch/acquisition/monte_carlo.py

@@ -18,6 +18,8 @@
     The reparameterization trick for acquisition functions. ArXiv 2017.
 """
 
+from __future__ import annotations
+
 import math
 from abc import ABC, abstractmethod
 from typing import Optional, Union

botorch/acquisition/objective.py

@@ -8,6 +8,8 @@
 Objective Modules to be used with acquisition functions.
 """
 
+from __future__ import annotations
+
 from abc import ABC, abstractmethod
 from typing import Callable, List
 

botorch/acquisition/utils.py

@@ -8,6 +8,8 @@
 Utilities for acquisition functions.
 """
 
+from __future__ import annotations
+
 import math
 from typing import Callable, Dict, List, Optional
 
@@ -34,7 +36,7 @@ def get_acquisition_function(
     qmc: bool = True,
     seed: Optional[int] = None,
     **kwargs,
-) -> "monte_carlo.MCAcquisitionFunction":
+) -> monte_carlo.MCAcquisitionFunction:
     r"""Convenience function for initializing botorch acquisition functions.
 
     Args:

botorch/cross_validation.py

@@ -8,6 +8,8 @@
 Cross-validation utilities using batch evaluation mode.
 """
 
+from __future__ import annotations
+
 from typing import Any, Dict, NamedTuple, Optional, Type
 
 import torch

botorch/fit.py

@@ -8,6 +8,8 @@
 Utilities for model fitting.
 """
 
+from __future__ import annotations
+
 import logging
 import warnings
 from copy import deepcopy
@@ -84,7 +86,8 @@ def fit_gpytorch_model(
             model_ = model_list_to_batched(mll_.model)
             mll.model.load_state_dict(model_.state_dict())
             return mll.eval()
-        except (NotImplementedError, UnsupportedError, RuntimeError, AttributeError):
+        # NotImplentedError is omitted since it derives from RuntimeError
+        except (UnsupportedError, RuntimeError, AttributeError):
             warnings.warn(FAILED_CONVERSION_MSG, BotorchWarning)
             return fit_gpytorch_model(
                 mll=mll, optimizer=optimizer, sequential=False, max_retries=max_retries

botorch/gen.py

@@ -8,6 +8,8 @@
 Candidate generation utilities.
 """
 
+from __future__ import annotations
+
 from typing import Any, Callable, Dict, List, Optional, Tuple, Type, Union
 
 import torch

botorch/models/converter.py

@@ -8,6 +8,8 @@
 Utilities for converting between different models.
 """
 
+from __future__ import annotations
+
 from copy import deepcopy
 
 import torch

botorch/models/cost.py

@@ -8,6 +8,8 @@
 Cost models to be used with multi-fidelity optimization.
 """
 
+from __future__ import annotations
+
 from typing import Dict, Optional
 
 import torch

botorch/models/deterministic.py

@@ -10,6 +10,8 @@
 known cost functions for cost-aware acquisition utilities.
 """
 
+from __future__ import annotations
+
 from abc import ABC, abstractmethod
 from typing import Any, Callable, List, Optional, Union
 
@@ -73,7 +75,7 @@ def __init__(self, f: Callable[[Tensor], Tensor], num_outputs: int = 1) -> None:
         self._f = f
         self._num_outputs = num_outputs
 
-    def subset_output(self, idcs: List[int]) -> "GenericDeterministicModel":
+    def subset_output(self, idcs: List[int]) -> GenericDeterministicModel:
         r"""Subset the model along the output dimension.
 
         Args:
@@ -128,7 +130,7 @@ def __init__(self, a: Tensor, b: Union[Tensor, float] = 0.01) -> None:
         self.register_buffer("b", b.expand(a.size(-1)))
         self._num_outputs = a.size(-1)
 
-    def subset_output(self, idcs: List[int]) -> "AffineDeterministicModel":
+    def subset_output(self, idcs: List[int]) -> AffineDeterministicModel:
         r"""Subset the model along the output dimension.
 
         Args:

botorch/models/gp_regression.py

@@ -8,6 +8,8 @@
 Gaussian Process Regression models based on GPyTorch models.
 """
 
+from __future__ import annotations
+
 from typing import Any, List, Optional, Union
 
 import torch
@@ -212,7 +214,7 @@ def fantasize(
         sampler: MCSampler,
         observation_noise: Union[bool, Tensor] = True,
         **kwargs: Any,
-    ) -> "FixedNoiseGP":
+    ) -> FixedNoiseGP:
         r"""Construct a fantasy model.
 
         Constructs a fantasy model in the following fashion:
@@ -254,7 +256,7 @@ def forward(self, x: Tensor) -> MultivariateNormal:
         covar_x = self.covar_module(x)
         return MultivariateNormal(mean_x, covar_x)
 
-    def subset_output(self, idcs: List[int]) -> "BatchedMultiOutputGPyTorchModel":
+    def subset_output(self, idcs: List[int]) -> BatchedMultiOutputGPyTorchModel:
         r"""Subset the model along the output dimension.
 
         Args:
@@ -336,8 +338,8 @@ def __init__(
 
     def condition_on_observations(
         self, X: Tensor, Y: Tensor, **kwargs: Any
-    ) -> "HeteroskedasticSingleTaskGP":
+    ) -> HeteroskedasticSingleTaskGP:
         raise NotImplementedError
 
-    def subset_output(self, idcs: List[int]) -> "HeteroskedasticSingleTaskGP":
+    def subset_output(self, idcs: List[int]) -> HeteroskedasticSingleTaskGP:
         raise NotImplementedError

botorch/models/gp_regression_fidelity.py

@@ -12,6 +12,8 @@
     multi-fidelity bayesian optimization for hyperparameter tuning. ArXiv 2019.
 """
 
+from __future__ import annotations
+
 from typing import Optional
 
 from gpytorch.kernels.kernel import ProductKernel

botorch/models/gpytorch.py

@@ -11,6 +11,8 @@
 GPyTorch Model class such as an ExactGP.
 """
 
+from __future__ import annotations
+
 import itertools
 import warnings
 from abc import ABC
@@ -136,7 +138,7 @@ def posterior(
             posterior = self.outcome_transform.untransform_posterior(posterior)
         return posterior
 
-    def condition_on_observations(self, X: Tensor, Y: Tensor, **kwargs: Any) -> "Model":
+    def condition_on_observations(self, X: Tensor, Y: Tensor, **kwargs: Any) -> Model:
         r"""Condition the model on new observations.
 
         Args:
@@ -330,7 +332,7 @@ def posterior(
 
     def condition_on_observations(
         self, X: Tensor, Y: Tensor, **kwargs: Any
-    ) -> "BatchedMultiOutputGPyTorchModel":
+    ) -> BatchedMultiOutputGPyTorchModel:
         r"""Condition the model on new observations.
 
         Args:
@@ -389,7 +391,7 @@ def condition_on_observations(
         fantasy_model._aug_batch_shape = fantasy_model.train_targets.shape[:-1]
         return fantasy_model
 
-    def subset_output(self, idcs: List[int]) -> "BatchedMultiOutputGPyTorchModel":
+    def subset_output(self, idcs: List[int]) -> BatchedMultiOutputGPyTorchModel:
         r"""Subset the model along the output dimension.
 
         Args:
@@ -524,7 +526,7 @@ def posterior(
 
     def condition_on_observations(
         self, X: Tensor, Y: Tensor, **kwargs: Any
-    ) -> "ModelListGPyTorchModel":
+    ) -> ModelListGPyTorchModel:
         class_name = self.__class__.__name__
         raise NotImplementedError(
             f"`condition_on_observations` not implemented in {class_name}"

botorch/models/kernels/downsampling.py

@@ -4,6 +4,8 @@
 # This source code is licensed under the MIT license found in the
 # LICENSE file in the root directory of this source tree.
 
+from __future__ import annotations
+
 from typing import Optional
 
 import torch

botorch/models/kernels/exponential_decay.py

@@ -4,6 +4,8 @@
 # This source code is licensed under the MIT license found in the
 # LICENSE file in the root directory of this source tree.
 
+from __future__ import annotations
+
 from typing import Optional
 
 import torch

botorch/models/kernels/linear_truncated_fidelity.py

@@ -4,6 +4,8 @@
 # This source code is licensed under the MIT license found in the
 # LICENSE file in the root directory of this source tree.
 
+from __future__ import annotations
+
 from copy import deepcopy
 from typing import Any, List, Optional
 
@@ -220,7 +222,7 @@ def forward(self, x1: Tensor, x2: Tensor, diag: bool = False, **params) -> Tenso
 
         return covar_unbiased + bias_factor * covar_biased
 
-    def __getitem__(self, index) -> "LinearTruncatedFidelityKernel":
+    def __getitem__(self, index) -> LinearTruncatedFidelityKernel:
         new_kernel = deepcopy(self)
         new_kernel.covar_module_unbiased = self.covar_module_unbiased[index]
         new_kernel.covar_module_biased = self.covar_module_biased[index]