Metadata-Version: 2.1
Name: nevergrad
Version: 0.7.0
Summary: A Python toolbox for performing gradient-free optimization
Home-page: https://github.com/facebookresearch/nevergrad
Author: Facebook AI Research
License: MIT
Classifier: License :: OSI Approved :: MIT License
Classifier: Intended Audience :: Science/Research
Classifier: Topic :: Scientific/Engineering
Classifier: Programming Language :: Python
Requires-Python: >=3.6
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: numpy (>=1.21.1)
Requires-Dist: cma (>=2.6.0)
Requires-Dist: bayesian-optimization (>=1.2.0)
Requires-Dist: typing-extensions (>=3.6.6)
Requires-Dist: pandas
Provides-Extra: all
Requires-Dist: black (==22.3.0) ; extra == 'all'
Requires-Dist: mypy (>=0.800) ; extra == 'all'
Requires-Dist: pytest (>=4.3.0) ; extra == 'all'
Requires-Dist: pytest-cov (>=2.6.1) ; extra == 'all'
Requires-Dist: pylint (>=2.4.4) ; extra == 'all'
Requires-Dist: wheel (>=0.33.6) ; extra == 'all'
Requires-Dist: setuptools (>=41.2.0) ; extra == 'all'
Requires-Dist: sphinx (>=3.3.0) ; extra == 'all'
Requires-Dist: sphinx-rtd-theme (>=0.4.3) ; extra == 'all'
Requires-Dist: recommonmark (>=0.5.0) ; extra == 'all'
Requires-Dist: twine (>=3.1.1) ; extra == 'all'
Requires-Dist: autodocsumm (>=0.1.11) ; extra == 'all'
Requires-Dist: pandas (>=0.23.4) ; extra == 'all'
Requires-Dist: pyparsing (==2.2.1) ; extra == 'all'
Requires-Dist: docutils (==0.17.1) ; extra == 'all'
Requires-Dist: requests (>=2.21.0) ; extra == 'all'
Requires-Dist: xlwt (>=1.3.0) ; extra == 'all'
Requires-Dist: xlrd (>=1.2.0) ; extra == 'all'
Requires-Dist: opencv-python (>=4.1.2.30) ; extra == 'all'
Requires-Dist: matplotlib (>=2.2.3) ; extra == 'all'
Requires-Dist: gym (==0.24.1) ; extra == 'all'
Requires-Dist: gym-anm (>=1.0.1) ; extra == 'all'
Requires-Dist: pygame (>=2.1.2) ; extra == 'all'
Requires-Dist: torch (>=1.7.0) ; extra == 'all'
Requires-Dist: hiplot ; extra == 'all'
Requires-Dist: fcmaes (>=1.2.7) ; extra == 'all'
Requires-Dist: openpyxl (>=3.0.0) ; extra == 'all'
Requires-Dist: pyproj (>=2.6.1) ; extra == 'all'
Requires-Dist: Pillow (>=8.4.0) ; extra == 'all'
Requires-Dist: tqdm ; extra == 'all'
Requires-Dist: torchvision (>=0.11.1) ; extra == 'all'
Requires-Dist: pyomo (==5.7.1) ; extra == 'all'
Requires-Dist: mixsimulator (>=0.3.3) ; extra == 'all'
Requires-Dist: hyperopt (>=0.2.5) ; extra == 'all'
Requires-Dist: IOHexperimenter (>=0.2.8.7) ; extra == 'all'
Requires-Dist: cdt (>=0.5.23) ; extra == 'all'
Requires-Dist: tensorflow-estimator (>=2.7.0) ; extra == 'all'
Requires-Dist: scikit-learn (>=1.0.1) ; extra == 'all'
Requires-Dist: scikit-image (==0.18.3) ; extra == 'all'
Requires-Dist: tensorflow (>=2.7.0) ; extra == 'all'
Requires-Dist: image-quality (>=1.2.7) ; extra == 'all'
Requires-Dist: keras (>=2.4.3) ; extra == 'all'
Requires-Dist: pymoo (==0.5.0) ; extra == 'all'
Requires-Dist: Keras-Preprocessing ; extra == 'all'
Requires-Dist: silence-tensorflow ; extra == 'all'
Requires-Dist: tensorflow-probability ; extra == 'all'
Requires-Dist: bayes-optim (==0.2.5.5) ; extra == 'all'
Requires-Dist: nlopt ; extra == 'all'
Requires-Dist: pybullet (>=3.2.2) ; extra == 'all'
Requires-Dist: box2d-py (>=2.3.5) ; extra == 'all'
Requires-Dist: glfw ; extra == 'all'
Requires-Dist: mujoco ; extra == 'all'
Requires-Dist: olymp (==0.0.1b0) ; (sys_platform == "linux") and extra == 'all'
Provides-Extra: benchmark
Requires-Dist: requests (>=2.21.0) ; extra == 'benchmark'
Requires-Dist: xlwt (>=1.3.0) ; extra == 'benchmark'
Requires-Dist: xlrd (>=1.2.0) ; extra == 'benchmark'
Requires-Dist: opencv-python (>=4.1.2.30) ; extra == 'benchmark'
Requires-Dist: matplotlib (>=2.2.3) ; extra == 'benchmark'
Requires-Dist: gym (==0.24.1) ; extra == 'benchmark'
Requires-Dist: gym-anm (>=1.0.1) ; extra == 'benchmark'
Requires-Dist: pygame (>=2.1.2) ; extra == 'benchmark'
Requires-Dist: torch (>=1.7.0) ; extra == 'benchmark'
Requires-Dist: hiplot ; extra == 'benchmark'
Requires-Dist: fcmaes (>=1.2.7) ; extra == 'benchmark'
Requires-Dist: pandas (>=0.23.4) ; extra == 'benchmark'
Requires-Dist: openpyxl (>=3.0.0) ; extra == 'benchmark'
Requires-Dist: pyproj (>=2.6.1) ; extra == 'benchmark'
Requires-Dist: Pillow (>=8.4.0) ; extra == 'benchmark'
Requires-Dist: tqdm ; extra == 'benchmark'
Requires-Dist: torchvision (>=0.11.1) ; extra == 'benchmark'
Requires-Dist: pyomo (==5.7.1) ; extra == 'benchmark'
Requires-Dist: mixsimulator (>=0.3.3) ; extra == 'benchmark'
Requires-Dist: hyperopt (>=0.2.5) ; extra == 'benchmark'
Requires-Dist: IOHexperimenter (>=0.2.8.7) ; extra == 'benchmark'
Requires-Dist: cdt (>=0.5.23) ; extra == 'benchmark'
Requires-Dist: tensorflow-estimator (>=2.7.0) ; extra == 'benchmark'
Requires-Dist: scikit-learn (>=1.0.1) ; extra == 'benchmark'
Requires-Dist: scikit-image (==0.18.3) ; extra == 'benchmark'
Requires-Dist: tensorflow (>=2.7.0) ; extra == 'benchmark'
Requires-Dist: image-quality (>=1.2.7) ; extra == 'benchmark'
Requires-Dist: keras (>=2.4.3) ; extra == 'benchmark'
Requires-Dist: pymoo (==0.5.0) ; extra == 'benchmark'
Requires-Dist: Keras-Preprocessing ; extra == 'benchmark'
Requires-Dist: silence-tensorflow ; extra == 'benchmark'
Requires-Dist: tensorflow-probability ; extra == 'benchmark'
Requires-Dist: bayes-optim (==0.2.5.5) ; extra == 'benchmark'
Requires-Dist: nlopt ; extra == 'benchmark'
Requires-Dist: pybullet (>=3.2.2) ; extra == 'benchmark'
Requires-Dist: box2d-py (>=2.3.5) ; extra == 'benchmark'
Requires-Dist: glfw ; extra == 'benchmark'
Requires-Dist: mujoco ; extra == 'benchmark'
Requires-Dist: olymp (==0.0.1b0) ; (sys_platform == "linux") and extra == 'benchmark'
Provides-Extra: dev
Requires-Dist: black (==22.3.0) ; extra == 'dev'
Requires-Dist: mypy (>=0.800) ; extra == 'dev'
Requires-Dist: pytest (>=4.3.0) ; extra == 'dev'
Requires-Dist: pytest-cov (>=2.6.1) ; extra == 'dev'
Requires-Dist: pylint (>=2.4.4) ; extra == 'dev'
Requires-Dist: wheel (>=0.33.6) ; extra == 'dev'
Requires-Dist: setuptools (>=41.2.0) ; extra == 'dev'
Requires-Dist: sphinx (>=3.3.0) ; extra == 'dev'
Requires-Dist: sphinx-rtd-theme (>=0.4.3) ; extra == 'dev'
Requires-Dist: recommonmark (>=0.5.0) ; extra == 'dev'
Requires-Dist: twine (>=3.1.1) ; extra == 'dev'
Requires-Dist: autodocsumm (>=0.1.11) ; extra == 'dev'
Requires-Dist: pandas (>=0.23.4) ; extra == 'dev'
Requires-Dist: pyparsing (==2.2.1) ; extra == 'dev'
Requires-Dist: docutils (==0.17.1) ; extra == 'dev'

[![Support Ukraine](https://img.shields.io/badge/Support-Ukraine-FFD500?style=flat&labelColor=005BBB)](https://opensource.fb.com/support-ukraine) [![CircleCI](https://circleci.com/gh/facebookresearch/nevergrad/tree/main.svg?style=svg)](https://circleci.com/gh/facebookresearch/nevergrad/tree/main)

# Nevergrad - A gradient-free optimization platform

![Nevergrad](https://raw.githubusercontent.com/facebookresearch/nevergrad/0.7.0/docs/resources/Nevergrad-LogoMark.png)


`nevergrad` is a Python 3.6+ library. It can be installed with:

```
pip install nevergrad
```

More installation options, including windows installation, and complete instructions are available in the "Getting started" section of the [**documentation**](https://facebookresearch.github.io/nevergrad/).

You can join Nevergrad users Facebook group [here](https://www.facebook.com/groups/nevergradusers/).

Minimizing a function using an optimizer (here `NGOpt`) is straightforward:

```python
import nevergrad as ng

def square(x):
    return sum((x - .5)**2)

optimizer = ng.optimizers.NGOpt(parametrization=2, budget=100)
recommendation = optimizer.minimize(square)
print(recommendation.value)  # recommended value
>>> [0.49971112 0.5002944]
```

`nevergrad` can also support bounded continuous variables as well as discrete variables, and mixture of those.
To do this, one can specify the input space:

```python
import nevergrad as ng

def fake_training(learning_rate: float, batch_size: int, architecture: str) -> float:
    # optimal for learning_rate=0.2, batch_size=4, architecture="conv"
    return (learning_rate - 0.2)**2 + (batch_size - 4)**2 + (0 if architecture == "conv" else 10)

# Instrumentation class is used for functions with multiple inputs
# (positional and/or keywords)
parametrization = ng.p.Instrumentation(
    # a log-distributed scalar between 0.001 and 1.0
    learning_rate=ng.p.Log(lower=0.001, upper=1.0),
    # an integer from 1 to 12
    batch_size=ng.p.Scalar(lower=1, upper=12).set_integer_casting(),
    # either "conv" or "fc"
    architecture=ng.p.Choice(["conv", "fc"])
)

optimizer = ng.optimizers.NGOpt(parametrization=parametrization, budget=100)
recommendation = optimizer.minimize(fake_training)

# show the recommended keyword arguments of the function
print(recommendation.kwargs)
>>> {'learning_rate': 0.1998, 'batch_size': 4, 'architecture': 'conv'}
```

Learn more on parametrization in the [**documentation**](https://facebookresearch.github.io/nevergrad/)!

![Example of optimization](https://raw.githubusercontent.com/facebookresearch/nevergrad/0.7.0/docs/resources/TwoPointsDE.gif)

*Convergence of a population of points to the minima with two-points DE.*


## Documentation

Check out our [**documentation**](https://facebookresearch.github.io/nevergrad/)! It's still a work in progress, don't hesitate to submit issues and/or PR to update it and make it clearer!


## Citing

```bibtex
@misc{nevergrad,
    author = {J. Rapin and O. Teytaud},
    title = {{Nevergrad - A gradient-free optimization platform}},
    year = {2018},
    publisher = {GitHub},
    journal = {GitHub repository},
    howpublished = {\url{https://GitHub.com/FacebookResearch/Nevergrad}},
}
```

## License

`nevergrad` is released under the MIT license. See [LICENSE](https://github.com/facebookresearch/nevergrad/blob/0.7.0/LICENSE) for additional details about it.
See also our [Terms of Use](https://opensource.facebook.com/legal/terms) and [Privacy Policy](https://opensource.facebook.com/legal/privacy).
