ABC Routine

CHANGELOG.md

@@ -8,6 +8,7 @@ All notable changes to this project will be documented in this file.
 - [#1432](https://github.com/pints-team/pints/pull/1432) Added 2 new stochastic models: production and degradation model, Schlogl's system of chemical reactions. Moved the stochastic logistic model into `pints.stochastic` to take advantage of the `MarkovJumpModel`.
 - [#1420](https://github.com/pints-team/pints/pull/1420) The `Optimiser` class now distinguishes between a best-visited point (`x_best`, with score `f_best`) and a best-guessed point (`x_guessed`, with approximate score `f_guessed`). For most optimisers, the two values are equivalent. The `OptimisationController` still tracks `x_best` and `f_best` by default, but this can be modified using the methods `set_f_guessed_tracking` and `f_guessed_tracking`.
 - [#1417](https://github.com/pints-team/pints/pull/1417) Added a module `toy.stochastic` for stochastic models. In particular, `toy.stochastic.MarkovJumpModel` implements Gillespie's algorithm for easier future implementation of stochastic models.
+- [#1413](https://github.com/pints-team/pints/pull/1413) Added classes `pints.ABCController` and `pints.ABCSampler` for Approximate Bayesian computation (ABC) samplers. Added `pints.RejectionABC` which implements a simple rejection ABC sampling algorithm.
 
 ### Changed
 - [#1439](https://github.com/pints-team/pints/pull/1439), [#1433](https://github.com/pints-team/pints/pull/1433) PINTS is no longer tested on Python 3.5. Testing for Python 3.10 has been added.

docs/source/abc_samplers/base_classes.rst

@@ -0,0 +1,8 @@
+**********************
+ABC sampler base class
+**********************
+
+.. currentmodule:: pints
+
+.. autoclass:: ABCSampler
+.. autoclass:: ABCController

docs/source/abc_samplers/index.rst

@@ -0,0 +1,16 @@
+************
+ABC samplers
+************
+
+.. currentmodule:: pints
+
+Pints provides a number of samplers for Approximate Bayesian
+Computation (ABC), all implementing the :class:`ABCSampler`
+interface, that can be used to sample from a stochastic model
+given a :class:`LogPrior` and a :class:`ErrorMeasure`.
+
+
+.. toctree::
+
+    base_classes
+    rejection_abc

docs/source/abc_samplers/rejection_abc.rst

@@ -0,0 +1,7 @@
+*********************
+Rejection ABC sampler
+*********************
+
+.. currentmodule:: pints
+
+.. autoclass:: RejectionABC

docs/source/index.rst

@@ -23,6 +23,7 @@ Contents
 
 .. toctree::
 
+    abc_samplers/index
     boundaries
     core_classes_and_methods
     diagnostics
@@ -78,10 +79,10 @@ Sampling
 
    - SMC
 
-#. Likelihood free sampling (Need distance between data and states, e.g. least squares?)
+#. :class:`ABC sampling<ABCSampler>`
 
-   - ABC-MCMC
-   - ABC-SMC
+   - :class:`RejectionABC`, requires a :class:`LogPrior` that can be sampled
+     from and an error measure.
 
 #. 1st order sensitivity MCMC samplers (Need derivatives of :class:`LogPDF`)
 

examples/README.md

@@ -77,6 +77,9 @@ relevant code.
 - [Ellipsoidal nested sampling](./sampling/nested-ellipsoidal-sampling.ipynb)
 - [Rejection nested sampling](./sampling/nested-rejection-sampling.ipynb)
 
+### ABC
+- [Rejection ABC sampling](./sampling/rejection-abc.ipynb)
+
 ### Analysing sampling results
 - [Autocorrelation](./plotting/mcmc-autocorrelation.ipynb)
 - [Customise analysis plots](./plotting/customise-pints-plots.ipynb)

pints/__init__.py

@@ -236,11 +236,20 @@ def version(formatted=False):
 from ._nested._ellipsoid import NestedEllipsoidSampler
 
 
+#
+# ABC
+#
+from ._abc import ABCSampler
+from ._abc import ABCController
+from ._abc._abc_rejection import RejectionABC
+
+
 #
 # Sampling initialising
 #
 from ._sample_initial_points import sample_initial_points
 
+
 #
 # Transformations
 #