Merge pull request #1724 from pints-team/1705-transformed-likelihoods

Reintroducing LogLikelihood class

Author: MichaelClerx

CHANGELOG.md

@@ -6,13 +6,17 @@ All notable changes to this project will be documented in this file.
 ## Unreleased
 
 ### Added
+- [#1724](https://github.com/pints-team/pints/pull/1724) The `LogLikelihood` class has been reintroduced, to differentiate between probabilities of parameters and probabilities of data, given fixed parameters.
+- [#1724](https://github.com/pints-team/pints/pull/1724) Added `PooledLogLikelihood` and `SumOfIndependentLogLikelihoods`.
 - [#1716](https://github.com/pints-team/pints/pull/1716) PINTS is now tested on Python 3.14.
 - [#1715](https://github.com/pints-team/pints/pull/1715) Added methods `ProblemErrorMeasure.problem()`, `ProblemLogLikelihood.problem()`, `SingleOutputProblem.model()` and `MultiOutputProblem.model()`.
 - [#1508](https://github.com/pints-team/pints/pull/1508) Added a method `OptimisationController.set_parameter_tolerance` that allows methods to stop after a fixed number of iterations with no significant movement in parameter space.
 ### Changed
+- [#1724](https://github.com/pints-team/pints/pull/1724) Some methods that accepted `LogPDF`s now specifically require `LogLikelihood`s (e.g. `LogPosterior`, `NestedController`).
 - [#1713](https://github.com/pints-team/pints/pull/1713) PINTS now requires matplotlib 2.2 or newer.
 ### Deprecated
 - [#1508](https://github.com/pints-team/pints/pull/1508) The methods `OptimisationController.max_unchanged_iterations` and `set_max_unchanged_iterations` are deprecated, in favour of `function_tolerance` and `set_function_tolerance` respectively.
+- [#1724](https://github.com/pints-team/pints/pull/1724) The classes `PooledLogPDF` and `SumOfIndependentLogPDFs` are deprecated, in favour of `PooledLogLikelihood` and `SumOfIndependentLogLikelihoods` respectively.
 ### Removed
 ### Fixed
 - [#1713](https://github.com/pints-team/pints/pull/1713) Fixed Numpy 2.4.1 compatibility issues.

docs/source/index.rst

@@ -95,8 +95,8 @@ Sampling
    - :class:`No U-Turn Sampler with dual averaging (NUTS) <NoUTurnMCMC>`
    - :class:`RelativisticMCMC`
 
-#. :class:`Nested sampling<NestedSampler>`, require a :class:`LogPDF` and a
-   :class:`LogPrior` that can be sampled from.
+#. :class:`Nested sampling<NestedSampler>`, require a :class:`LogLikelihood`
+   and a :class:`LogPrior` that can be sampled from.
 
    - :class:`NestedEllipsoidSampler`
    - :class:`NestedRejectionSampler`

docs/source/log_likelihoods.rst

@@ -4,15 +4,17 @@ Log-likelihoods
 
 .. currentmodule:: pints
 
-The classes below all implement the :class:`ProblemLogLikelihood` interface,
-and can calculate a log-likelihood based on some time-series :class:`Problem`
-and an assumed noise model.
+The classes below all implement the :class:`LogLikelihood` interface.
+Most are :class:`ProblemLogLikelihood` implementations, which calculate a
+log-likelihood based on some time-series :class:`Problem` and an assumed noise
+model.
+Some are methods combining other likelihoods.
 
 Example::
 
-    logpdf = pints.GaussianLogLikelihood(problem)
+    log_likelihood = pints.GaussianLogLikelihood(problem)
     x = [1, 2, 3]
-    fx = logpdf(x)
+    fx = log_likelihood(x)
 
 
 .. autoclass:: AR1LogLikelihood
@@ -39,8 +41,17 @@ Example::
 
 .. autoclass:: MultiplicativeGaussianLogLikelihood
 
+.. autoclass:: PooledLogLikelihood
+
+.. autoclass:: PooledLogPDF
+
 .. autoclass:: ScaledLogLikelihood
 
 .. autoclass:: StudentTLogLikelihood
 
+.. autoclass:: SumOfIndependentLogLikelihoods
+
+.. autoclass:: SumOfIndependentLogPDFs
+
 .. autoclass:: UnknownNoiseLogLikelihood
+

docs/source/log_pdfs.rst

@@ -20,11 +20,9 @@ Example::
 
 .. autoclass:: LogPrior
 
-.. autoclass:: LogPosterior
-
-.. autoclass:: PooledLogPDF
+.. autoclass:: LogLikelihood
 
 .. autoclass:: ProblemLogLikelihood
 
-.. autoclass:: SumOfIndependentLogPDFs
+.. autoclass:: LogPosterior
 

examples/miscellaneous/model-rt-estimation-renewal-equation.ipynb

@@ -79,12 +79,11 @@ def version(formatted=False):
 # Logs of probability density functions (not necessarily normalised)
 #
 from ._log_pdfs import (
+    LogLikelihood,
     LogPDF,
     LogPrior,
     LogPosterior,
-    PooledLogPDF,
     ProblemLogLikelihood,
-    SumOfIndependentLogPDFs,
 )
 
 #
@@ -124,8 +123,12 @@ def version(formatted=False):
     KnownNoiseLogLikelihood,
     LogNormalLogLikelihood,
     MultiplicativeGaussianLogLikelihood,
+    PooledLogLikelihood,
+    PooledLogPDF,
     ScaledLogLikelihood,
     StudentTLogLikelihood,
+    SumOfIndependentLogLikelihoods,
+    SumOfIndependentLogPDFs,
     UnknownNoiseLogLikelihood,
 )
 

examples/toy/distribution-rosenbrock.ipynb

@@ -189,7 +189,7 @@ def upper(self):
 class LogPDFBoundaries(Boundaries):
     """
     Uses a :class:`pints.LogPDF` (e.g. a :class:`LogPrior`) as boundaries),
-    accepting log-likelihoods above a given threshold as within bounds.
+    accepting log-pdfs above a given threshold as within bounds.
 
     For a :class:`pints.LogPrior` based on :class:`pints.Boundaries`, see
     :class:`pints.UniformLogPrior`.