GPJax


NameGPJax JSON
Version 0.5.5 PyPI version JSON
download
home_page
SummaryDidactic Gaussian processes in Jax.
upload_time2022-12-18 19:20:41
maintainer
docs_urlNone
authorThomas Pinder
requires_python
licenseLICENSE
keywords gaussian-processes jax machine-learning bayesian
VCS
bugtrack_url
requirements No requirements were recorded.
Travis-CI No Travis.
coveralls test coverage No coveralls.
            <!-- <h1 align='center'>GPJax</h1>
<h2 align='center'>Gaussian processes in Jax.</h2> -->
<p align="center">
<img width="700" height="300" src="https://github.com/JaxGaussianProcesses/GPJax/raw/master/docs/_static/gpjax_logo.svg" alt="GPJax's logo">
</p>

[![codecov](https://codecov.io/gh/JaxGaussianProcesses/GPJax/branch/master/graph/badge.svg?token=DM1DRDASU2)](https://codecov.io/gh/JaxGaussianProcesses/GPJax)
[![CodeFactor](https://www.codefactor.io/repository/github/jaxgaussianprocesses/gpjax/badge)](https://www.codefactor.io/repository/github/jaxgaussianprocesses/gpjax)
[![Documentation Status](https://readthedocs.org/projects/gpjax/badge/?version=latest)](https://gpjax.readthedocs.io/en/latest/?badge=latest)
[![PyPI version](https://badge.fury.io/py/GPJax.svg)](https://badge.fury.io/py/GPJax)
[![DOI](https://joss.theoj.org/papers/10.21105/joss.04455/status.svg)](https://doi.org/10.21105/joss.04455)
[![Downloads](https://pepy.tech/badge/gpjax)](https://pepy.tech/project/gpjax)
[![Slack Invite](https://img.shields.io/badge/Slack_Invite--blue?style=social&logo=slack)](https://join.slack.com/t/gpjax/shared_invite/zt-1da57pmjn-rdBCVg9kApirEEn2E5Q2Zw)

[**Quickstart**](#simple-example)
| [**Install guide**](#installation)
| [**Documentation**](https://gpjax.readthedocs.io/en/latest/)
| [**Slack Community**](https://join.slack.com/t/gpjax/shared_invite/zt-1da57pmjn-rdBCVg9kApirEEn2E5Q2Zw)

GPJax aims to provide a low-level interface to Gaussian process (GP) models in [Jax](https://github.com/google/jax), structured to give researchers maximum flexibility in extending the code to suit their own needs. The idea is that the code should be as close as possible to the maths we write on paper when working with GP models.

# Package support

GPJax was founded by [Thomas Pinder](https://github.com/thomaspinder). Today, the maintenance of GPJax is undertaken by [Thomas Pinder](https://github.com/thomaspinder) and [Daniel Dodd](https://github.com/Daniel-Dodd).

We would be delighted to receive contributions from interested individuals and groups. To learn how you can get involved, please read our [guide for contributing](https://github.com/JaxGaussianProcesses/GPJax/blob/master/CONTRIBUTING.md). If you have any questions, we encourage you to [open an issue](https://github.com/JaxGaussianProcesses/GPJax/issues/new/choose). For broader conversations, such as best GP fitting practices or questions about the mathematics of GPs, we invite you to [open a discussion](https://github.com/JaxGaussianProcesses/GPJax/discussions).

Feel free to join our [Slack Channel](https://join.slack.com/t/gpjax/shared_invite/zt-1da57pmjn-rdBCVg9kApirEEn2E5Q2Zw), where we can discuss the development of GPJax and broader support for Gaussian process modelling.

# Supported methods and interfaces

## Notebook examples

> - [**Conjugate Inference**](https://gpjax.readthedocs.io/en/latest/examples/regression.html)
> - [**Classification with MCMC**](https://gpjax.readthedocs.io/en/latest/examples/classification.html)
> - [**Sparse Variational Inference**](https://gpjax.readthedocs.io/en/latest/examples/uncollapsed_vi.html)
> - [**BlackJax Integration**](https://gpjax.readthedocs.io/en/latest/examples/classification.html)
> - [**Laplace Approximation**](https://gpjax.readthedocs.io/en/latest/examples/classification.html#Laplace-approximation)
> - [**TensorFlow Probability Integration**](https://gpjax.readthedocs.io/en/latest/examples/tfp_integration.html)
> - [**Inference on Non-Euclidean Spaces**](https://gpjax.readthedocs.io/en/latest/examples/kernels.html#Custom-Kernel)
> - [**Inference on Graphs**](https://gpjax.readthedocs.io/en/latest/examples/graph_kernels.html)
> - [**Learning Gaussian Process Barycentres**](https://gpjax.readthedocs.io/en/latest/examples/barycentres.html)
> - [**Deep Kernel Regression**](https://gpjax.readthedocs.io/en/latest/examples/haiku.html)
> - [**Natural Gradients**](https://gpjax.readthedocs.io/en/latest/examples/natgrads.html)

## Guides for customisation
> 
> - [**Custom kernels**](https://gpjax.readthedocs.io/en/latest/examples/kernels.html#Custom-Kernel)
> - [**UCI regression**](https://gpjax.readthedocs.io/en/latest/examples/yacht.html)


# Simple example

Let us import some dependencies and simulate a toy dataset $\mathcal{D}$.

```python
import gpjax as gpx
from jax import grad, jit
import jax.numpy as jnp
import jax.random as jr
import optax as ox

key = jr.PRNGKey(123)

f = lambda x: 10 * jnp.sin(x)

n = 50
x = jr.uniform(key=key, minval=-3.0, maxval=3.0, shape=(n,1)).sort()
y = f(x) + jr.normal(key, shape=(n,1))
D = gpx.Dataset(X=x, y=y)
```

The function of interest here, $f(\cdot)$, is sinusoidal, but our observations of it have been perturbed by Gaussian noise. We aim to utilise a Gaussian process to try and recover this latent function.

## 1. Constructing the prior and posterior

We begin by defining a zero-mean Gaussian process prior with a radial basis function kernel and assume the likelihood to be Gaussian.

```python
prior = gpx.Prior(kernel = gpx.RBF())
likelihood = gpx.Gaussian(num_datapoints = n)
```

Similar to how we would write on paper, the posterior is constructed by the product of our prior with our likelihood.

```python
posterior = prior * likelihood
```

## 2. Learning hyperparameters

Equipped with the posterior, we seek to learn the model's hyperparameters through gradient-optimisation of the marginal log-likelihood. We this below, adding Jax's [just-in-time (JIT)](https://jax.readthedocs.io/en/latest/jax-101/02-jitting.html) compilation to accelerate training. 

```python
mll = jit(posterior.marginal_log_likelihood(D, negative=True))
```

For purposes of optimisation, we'll use optax's Adam.
```
opt = ox.adam(learning_rate=1e-3)
```

We define an initial parameter state through the `initialise` callable.

```python
parameter_state = gpx.initialise(posterior, key=key)
```

Finally, we run an optimisation loop using the Adam optimiser via the `fit` callable.

```python
inference_state = gpx.fit(mll, parameter_state, opt, n_iters=500)
```

## 3. Making predictions

Using our learned hyperparameters, we can obtain the posterior distribution of the latent function at novel test points.

```python
learned_params, _ = inference_state.unpack()
xtest = jnp.linspace(-3., 3., 100).reshape(-1, 1)

latent_distribution = posterior(learned_params, D)(xtest)
predictive_distribution = likelihood(learned_params, latent_distribution)

predictive_mean = predictive_distribution.mean()
predictive_cov = predictive_distribution.covariance()
```

# Installation

## Stable version

The latest stable version of GPJax can be installed via [`pip`](https://pip.pypa.io/en/stable/):

```bash
pip install gpjax
```

> **Note**
>
> We recommend you check your installation version:
> ```python
> python -c 'import gpjax; print(gpjax.__version__)'
> ```



## Development version
> **Warning**
>
> This version is possibly unstable and may contain bugs. 

Clone a copy of the repository to your local machine and run the setup configuration in development mode.
```bash
git clone https://github.com/JaxGaussianProcesses/GPJax.git
cd GPJax
python setup.py develop
```

> **Note**
>
> We advise you create virtual environment before installing:
> ```
> conda create -n gpjax_experimental python=3.10.0
> conda activate gpjax_experimental
>  ```
>
> and recommend you check your installation passes the supplied unit tests:
>
> ```python
> python -m pytest tests/
> ```

# Citing GPJax

If you use GPJax in your research, please cite our [JOSS paper](https://joss.theoj.org/papers/10.21105/joss.04455#).

```
@article{Pinder2022,
  doi = {10.21105/joss.04455},
  url = {https://doi.org/10.21105/joss.04455},
  year = {2022},
  publisher = {The Open Journal},
  volume = {7},
  number = {75},
  pages = {4455},
  author = {Thomas Pinder and Daniel Dodd},
  title = {GPJax: A Gaussian Process Framework in JAX},
  journal = {Journal of Open Source Software}
}
```
            

Raw data

            {
    "_id": null,
    "home_page": "",
    "name": "GPJax",
    "maintainer": "",
    "docs_url": null,
    "requires_python": "",
    "maintainer_email": "",
    "keywords": "gaussian-processes jax machine-learning bayesian",
    "author": "Thomas Pinder",
    "author_email": "t.pinder2@lancaster.ac.uk",
    "download_url": "https://files.pythonhosted.org/packages/76/a5/2f1ad92789cef687678e87209499f04298da890efeda0d22fac8d57c9009/GPJax-0.5.5.tar.gz",
    "platform": null,
    "description": "<!-- <h1 align='center'>GPJax</h1>\n<h2 align='center'>Gaussian processes in Jax.</h2> -->\n<p align=\"center\">\n<img width=\"700\" height=\"300\" src=\"https://github.com/JaxGaussianProcesses/GPJax/raw/master/docs/_static/gpjax_logo.svg\" alt=\"GPJax's logo\">\n</p>\n\n[![codecov](https://codecov.io/gh/JaxGaussianProcesses/GPJax/branch/master/graph/badge.svg?token=DM1DRDASU2)](https://codecov.io/gh/JaxGaussianProcesses/GPJax)\n[![CodeFactor](https://www.codefactor.io/repository/github/jaxgaussianprocesses/gpjax/badge)](https://www.codefactor.io/repository/github/jaxgaussianprocesses/gpjax)\n[![Documentation Status](https://readthedocs.org/projects/gpjax/badge/?version=latest)](https://gpjax.readthedocs.io/en/latest/?badge=latest)\n[![PyPI version](https://badge.fury.io/py/GPJax.svg)](https://badge.fury.io/py/GPJax)\n[![DOI](https://joss.theoj.org/papers/10.21105/joss.04455/status.svg)](https://doi.org/10.21105/joss.04455)\n[![Downloads](https://pepy.tech/badge/gpjax)](https://pepy.tech/project/gpjax)\n[![Slack Invite](https://img.shields.io/badge/Slack_Invite--blue?style=social&logo=slack)](https://join.slack.com/t/gpjax/shared_invite/zt-1da57pmjn-rdBCVg9kApirEEn2E5Q2Zw)\n\n[**Quickstart**](#simple-example)\n| [**Install guide**](#installation)\n| [**Documentation**](https://gpjax.readthedocs.io/en/latest/)\n| [**Slack Community**](https://join.slack.com/t/gpjax/shared_invite/zt-1da57pmjn-rdBCVg9kApirEEn2E5Q2Zw)\n\nGPJax aims to provide a low-level interface to Gaussian process (GP) models in [Jax](https://github.com/google/jax), structured to give researchers maximum flexibility in extending the code to suit their own needs. The idea is that the code should be as close as possible to the maths we write on paper when working with GP models.\n\n# Package support\n\nGPJax was founded by [Thomas Pinder](https://github.com/thomaspinder). Today, the maintenance of GPJax is undertaken by [Thomas Pinder](https://github.com/thomaspinder) and [Daniel Dodd](https://github.com/Daniel-Dodd).\n\nWe would be delighted to receive contributions from interested individuals and groups. To learn how you can get involved, please read our [guide for contributing](https://github.com/JaxGaussianProcesses/GPJax/blob/master/CONTRIBUTING.md). If you have any questions, we encourage you to [open an issue](https://github.com/JaxGaussianProcesses/GPJax/issues/new/choose). For broader conversations, such as best GP fitting practices or questions about the mathematics of GPs, we invite you to [open a discussion](https://github.com/JaxGaussianProcesses/GPJax/discussions).\n\nFeel free to join our [Slack Channel](https://join.slack.com/t/gpjax/shared_invite/zt-1da57pmjn-rdBCVg9kApirEEn2E5Q2Zw), where we can discuss the development of GPJax and broader support for Gaussian process modelling.\n\n# Supported methods and interfaces\n\n## Notebook examples\n\n> - [**Conjugate Inference**](https://gpjax.readthedocs.io/en/latest/examples/regression.html)\n> - [**Classification with MCMC**](https://gpjax.readthedocs.io/en/latest/examples/classification.html)\n> - [**Sparse Variational Inference**](https://gpjax.readthedocs.io/en/latest/examples/uncollapsed_vi.html)\n> - [**BlackJax Integration**](https://gpjax.readthedocs.io/en/latest/examples/classification.html)\n> - [**Laplace Approximation**](https://gpjax.readthedocs.io/en/latest/examples/classification.html#Laplace-approximation)\n> - [**TensorFlow Probability Integration**](https://gpjax.readthedocs.io/en/latest/examples/tfp_integration.html)\n> - [**Inference on Non-Euclidean Spaces**](https://gpjax.readthedocs.io/en/latest/examples/kernels.html#Custom-Kernel)\n> - [**Inference on Graphs**](https://gpjax.readthedocs.io/en/latest/examples/graph_kernels.html)\n> - [**Learning Gaussian Process Barycentres**](https://gpjax.readthedocs.io/en/latest/examples/barycentres.html)\n> - [**Deep Kernel Regression**](https://gpjax.readthedocs.io/en/latest/examples/haiku.html)\n> - [**Natural Gradients**](https://gpjax.readthedocs.io/en/latest/examples/natgrads.html)\n\n## Guides for customisation\n> \n> - [**Custom kernels**](https://gpjax.readthedocs.io/en/latest/examples/kernels.html#Custom-Kernel)\n> - [**UCI regression**](https://gpjax.readthedocs.io/en/latest/examples/yacht.html)\n\n\n# Simple example\n\nLet us import some dependencies and simulate a toy dataset $\\mathcal{D}$.\n\n```python\nimport gpjax as gpx\nfrom jax import grad, jit\nimport jax.numpy as jnp\nimport jax.random as jr\nimport optax as ox\n\nkey = jr.PRNGKey(123)\n\nf = lambda x: 10 * jnp.sin(x)\n\nn = 50\nx = jr.uniform(key=key, minval=-3.0, maxval=3.0, shape=(n,1)).sort()\ny = f(x) + jr.normal(key, shape=(n,1))\nD = gpx.Dataset(X=x, y=y)\n```\n\nThe function of interest here, $f(\\cdot)$, is sinusoidal, but our observations of it have been perturbed by Gaussian noise. We aim to utilise a Gaussian process to try and recover this latent function.\n\n## 1. Constructing the prior and posterior\n\nWe begin by defining a zero-mean Gaussian process prior with a radial basis function kernel and assume the likelihood to be Gaussian.\n\n```python\nprior = gpx.Prior(kernel = gpx.RBF())\nlikelihood = gpx.Gaussian(num_datapoints = n)\n```\n\nSimilar to how we would write on paper, the posterior is constructed by the product of our prior with our likelihood.\n\n```python\nposterior = prior * likelihood\n```\n\n## 2. Learning hyperparameters\n\nEquipped with the posterior, we seek to learn the model's hyperparameters through gradient-optimisation of the marginal log-likelihood. We this below, adding Jax's [just-in-time (JIT)](https://jax.readthedocs.io/en/latest/jax-101/02-jitting.html) compilation to accelerate training. \n\n```python\nmll = jit(posterior.marginal_log_likelihood(D, negative=True))\n```\n\nFor purposes of optimisation, we'll use optax's Adam.\n```\nopt = ox.adam(learning_rate=1e-3)\n```\n\nWe define an initial parameter state through the `initialise` callable.\n\n```python\nparameter_state = gpx.initialise(posterior, key=key)\n```\n\nFinally, we run an optimisation loop using the Adam optimiser via the `fit` callable.\n\n```python\ninference_state = gpx.fit(mll, parameter_state, opt, n_iters=500)\n```\n\n## 3. Making predictions\n\nUsing our learned hyperparameters, we can obtain the posterior distribution of the latent function at novel test points.\n\n```python\nlearned_params, _ = inference_state.unpack()\nxtest = jnp.linspace(-3., 3., 100).reshape(-1, 1)\n\nlatent_distribution = posterior(learned_params, D)(xtest)\npredictive_distribution = likelihood(learned_params, latent_distribution)\n\npredictive_mean = predictive_distribution.mean()\npredictive_cov = predictive_distribution.covariance()\n```\n\n# Installation\n\n## Stable version\n\nThe latest stable version of GPJax can be installed via [`pip`](https://pip.pypa.io/en/stable/):\n\n```bash\npip install gpjax\n```\n\n> **Note**\n>\n> We recommend you check your installation version:\n> ```python\n> python -c 'import gpjax; print(gpjax.__version__)'\n> ```\n\n\n\n## Development version\n> **Warning**\n>\n> This version is possibly unstable and may contain bugs. \n\nClone a copy of the repository to your local machine and run the setup configuration in development mode.\n```bash\ngit clone https://github.com/JaxGaussianProcesses/GPJax.git\ncd GPJax\npython setup.py develop\n```\n\n> **Note**\n>\n> We advise you create virtual environment before installing:\n> ```\n> conda create -n gpjax_experimental python=3.10.0\n> conda activate gpjax_experimental\n>  ```\n>\n> and recommend you check your installation passes the supplied unit tests:\n>\n> ```python\n> python -m pytest tests/\n> ```\n\n# Citing GPJax\n\nIf you use GPJax in your research, please cite our [JOSS paper](https://joss.theoj.org/papers/10.21105/joss.04455#).\n\n```\n@article{Pinder2022,\n  doi = {10.21105/joss.04455},\n  url = {https://doi.org/10.21105/joss.04455},\n  year = {2022},\n  publisher = {The Open Journal},\n  volume = {7},\n  number = {75},\n  pages = {4455},\n  author = {Thomas Pinder and Daniel Dodd},\n  title = {GPJax: A Gaussian Process Framework in JAX},\n  journal = {Journal of Open Source Software}\n}\n```",
    "bugtrack_url": null,
    "license": "LICENSE",
    "summary": "Didactic Gaussian processes in Jax.",
    "version": "0.5.5",
    "split_keywords": [
        "gaussian-processes",
        "jax",
        "machine-learning",
        "bayesian"
    ],
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "47cddd18bc0d9a598ff38b3a9e6928bb",
                "sha256": "c85673fa4e72a79d271b15d1341cd558eeb1dd628600dd4d0b83a20d0890e554"
            },
            "downloads": -1,
            "filename": "GPJax-0.5.5.tar.gz",
            "has_sig": false,
            "md5_digest": "47cddd18bc0d9a598ff38b3a9e6928bb",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 38692,
            "upload_time": "2022-12-18T19:20:41",
            "upload_time_iso_8601": "2022-12-18T19:20:41.587797Z",
            "url": "https://files.pythonhosted.org/packages/76/a5/2f1ad92789cef687678e87209499f04298da890efeda0d22fac8d57c9009/GPJax-0.5.5.tar.gz",
            "yanked": false,
            "yanked_reason": null
        }
    ],
    "upload_time": "2022-12-18 19:20:41",
    "github": false,
    "gitlab": false,
    "bitbucket": false,
    "lcname": "gpjax"
}
        
Elapsed time: 0.02349s