Gaussian Processes Tool

docs/changes/739.feature.rst

@@ -0,0 +1,19 @@
+This is a JAX implementation of the GP tool by `Hubener et al <https://arxiv.org/pdf/2205.12716.pdf#:~:text=ABSTRACT%20Analyses%20of%20quasi%2Dperiodic,flares%20and%20fast%20radio%20bursts.>`_ 
+for QPO detection and parameter analysis.
+
+This feature makes use of tinygp library for Gaussian Processes implementation, and jaxns for nested sampling,
+and is kept in the stingray.modeling.gpmodeling module.
+
+Main features of the module are:
+
+- get_prior: This function makes the prior function for a specified prior dictionary.
+- get_likelihood: This function makes the log_likelihood function for the given Kernel, Mean model and lightcurve.
+- GPResult class: The class which takes a Lightcurve, and performs Nested Sampling for a given prior and likelihood.
+
+The additional Dependencies for the code
+- jax
+- tinygp
+- jaxns
+- etils
+- tensorflow_probability
+- typing_extensions

docs/index.rst

@@ -113,6 +113,20 @@ To install all required and recommended dependencies in a recent installation, y
 
     $ pip install astropy scipy matplotlib numpy h5py tqdm numba pint-pulsar emcee corner statsmodels pyfftw tbb
 
+For the Gaussian Process modeling in `stingray.modeling.gpmodeling`, you'll need the following extra packages
+
++ jax
++ jaxns
++ tensorflow
++ tensorflow-probability
++ tinygp
++ etils
++ typing_extensions
+
+Most of these are installed via ``pip``, but if you have an Nvidia GPU available, you'll want to take special care 
+following the installation instructions for jax and tensorflow(-probability) in order to enable GPU support and 
+take advantage of those speed-ups. 
+
 For development work, you will need the following extra libraries:
 
 + pytest

setup.cfg

@@ -44,6 +44,12 @@ all =
     xarray
     pandas
     ultranest
+    jax
+    tinygp
+    jaxns
+    etils
+    tensorflow_probability
+    typing_extensions
 test =
     pytest
     pytest-astropy