.gitlab-ci.yml

@@ -84,7 +84,7 @@ basic-3.11:
   script:
     - python -m pip install .
     - *list-env
-    - python test/test_samplers_import.py
+    - pytest test/test_samplers_import.py -v
 
 import-samplers-3.9:
   <<: *test-samplers-import

AUTHORS.md

@@ -11,6 +11,7 @@ Andrew Miller
 Antoni Ramos-Buades
 Apratim Ganguly
 Avi Vajpeyi
+Ben Patterson
 Bruce Edelman
 Carl-Johan Haster
 Cecilio Garcia-Quiros
@@ -73,6 +74,7 @@ Roberto Cotesta
 Rory Smith
 S. H. Oh
 Sacha Husa
+Sama Al-Shammari
 Samson Leong
 Scott Coughlin
 Serguei Ossokine
@@ -98,3 +100,4 @@ Andrew Fowlie
 Martin White
 Peter Tsun-Ho Pang
 Alexandre Sebastien Goettel
+Ann-Kristin Malz

bilby/core/result.py

@@ -69,7 +69,11 @@ def result_file_name(outdir, label, extension='json', gzip=False):
 def _determine_file_name(filename, outdir, label, extension, gzip):
     """ Helper method to determine the filename """
     if filename is not None:
-        return filename
+        if isinstance(filename, os.PathLike):
+            # convert PathLike object to string
+            return str(filename)
+        else:
+            return filename
     else:
         if (outdir is None) and (label is None):
             raise ValueError("No information given to load file")
@@ -1795,7 +1799,7 @@ class ResultList(list):
 
         if isinstance(result, Result):
             super(ResultList, self).append(result)
-        elif isinstance(result, str):
+        elif isinstance(result, (str, os.PathLike)):
             super(ResultList, self).append(read_in_result(result))
         else:
             raise TypeError("Could not append a non-Result type")
@@ -2057,7 +2061,7 @@ def plot_multiple(results, filename=None, labels=None, colours=None,
         hist_kwargs['color'] = c
         hist_kwargs["linestyle"] = linestyle
         kwargs["hist_kwargs"] = hist_kwargs
-        fig = result.plot_corner(fig=fig, save=False, color=c, contour_kwargs={"linestyle": linestyle}, **kwargs)
+        fig = result.plot_corner(fig=fig, save=False, color=c, contour_kwargs={"linestyles": linestyle}, **kwargs)
         default_filename += '_{}'.format(result.label)
         lines.append(mpllines.Line2D([0], [0], color=c, linestyle=linestyle))
         default_labels.append(result.label)

bilby/core/sampler/__init__.py

@@ -2,54 +2,69 @@ import datetime
 import inspect
 import sys
 
-import bilby
-from bilby.bilby_mcmc import Bilby_MCMC
-
 from ..prior import DeltaFunction, PriorDict
-from ..utils import command_line_args, env_package_list, loaded_modules_dict, logger
+from ..utils import (
+    command_line_args,
+    env_package_list,
+    get_entry_points,
+    loaded_modules_dict,
+    logger,
+)
 from . import proposal
 from .base_sampler import Sampler, SamplingMarginalisedParameterError
-from .cpnest import Cpnest
-from .dnest4 import DNest4
-from .dynamic_dynesty import DynamicDynesty
-from .dynesty import Dynesty
-from .emcee import Emcee
-from .fake_sampler import FakeSampler
-from .kombine import Kombine
-from .nessai import Nessai
-from .nestle import Nestle
-from .polychord import PyPolyChord
-from .ptemcee import Ptemcee
-from .ptmcmc import PTMCMCSampler
-from .pymc import Pymc
-from .pymultinest import Pymultinest
-from .ultranest import Ultranest
-from .zeus import Zeus
-
-IMPLEMENTED_SAMPLERS = {
-    "bilby_mcmc": Bilby_MCMC,
-    "cpnest": Cpnest,
-    "dnest4": DNest4,
-    "dynamic_dynesty": DynamicDynesty,
-    "dynesty": Dynesty,
-    "emcee": Emcee,
-    "kombine": Kombine,
-    "nessai": Nessai,
-    "nestle": Nestle,
-    "ptemcee": Ptemcee,
-    "ptmcmcsampler": PTMCMCSampler,
-    "pymc": Pymc,
-    "pymultinest": Pymultinest,
-    "pypolychord": PyPolyChord,
-    "ultranest": Ultranest,
-    "zeus": Zeus,
-    "fake_sampler": FakeSampler,
-}
+
+IMPLEMENTED_SAMPLERS = get_entry_points("bilby.samplers")
+
+
+def get_implemented_samplers():
+    """Get a list of the names of the implemented samplers.
+
+    This includes natively supported samplers (e.g. dynesty) and any additional
+    samplers that are supported through the sampler plugins.
+
+    Returns
+    -------
+    list
+        The list of implemented samplers.
+    """
+    return list(IMPLEMENTED_SAMPLERS.keys())
+
+
+def get_sampler_class(sampler):
+    """Get the class for a sampler from its name.
+
+    This includes natively supported samplers (e.g. dynesty) and any additional
+    samplers that are supported through the sampler plugins.
+
+    Parameters
+    ----------
+    sampler : str
+        The name of the sampler.
+
+    Returns
+    -------
+    Sampler
+        The sampler class.
+
+    Raises
+    ------
+    ValueError
+        Raised if the sampler is not implemented.
+    """
+    sampler = sampler.lower()
+    if sampler in IMPLEMENTED_SAMPLERS:
+        return IMPLEMENTED_SAMPLERS[sampler].load()
+    else:
+        raise ValueError(
+            f"Sampler {sampler} not yet implemented. "
+            f"The available samplers are: {get_implemented_samplers()}"
+        )
+
 
 if command_line_args.sampler_help:
     sampler = command_line_args.sampler_help
     if sampler in IMPLEMENTED_SAMPLERS:
-        sampler_class = IMPLEMENTED_SAMPLERS[sampler]
+        sampler_class = IMPLEMENTED_SAMPLERS[sampler].load()
         print(f'Help for sampler "{sampler}":')
         print(sampler_class.__doc__)
     else:
@@ -60,7 +75,7 @@ if command_line_args.sampler_help:
             )
         else:
             print(f"Requested sampler {sampler} not implemented")
-        print(f"Available samplers = {IMPLEMENTED_SAMPLERS}")
+        print(f"Available samplers = {get_implemented_samplers()}")
 
     sys.exit()
 
@@ -185,24 +200,20 @@ def run_sampler(
     if isinstance(sampler, Sampler):
         pass
     elif isinstance(sampler, str):
-        if sampler.lower() in IMPLEMENTED_SAMPLERS:
-            sampler_class = IMPLEMENTED_SAMPLERS[sampler.lower()]
-            sampler = sampler_class(
-                likelihood,
-                priors=priors,
-                outdir=outdir,
-                label=label,
-                injection_parameters=injection_parameters,
-                meta_data=meta_data,
-                use_ratio=use_ratio,
-                plot=plot,
-                result_class=result_class,
-                npool=npool,
-                **kwargs,
-            )
-        else:
-            print(IMPLEMENTED_SAMPLERS)
-            raise ValueError(f"Sampler {sampler} not yet implemented")
+        sampler_class = get_sampler_class(sampler)
+        sampler = sampler_class(
+            likelihood,
+            priors=priors,
+            outdir=outdir,
+            label=label,
+            injection_parameters=injection_parameters,
+            meta_data=meta_data,
+            use_ratio=use_ratio,
+            plot=plot,
+            result_class=result_class,
+            npool=npool,
+            **kwargs,
+        )
     elif inspect.isclass(sampler):
         sampler = sampler.__init__(
             likelihood,
@@ -219,7 +230,7 @@ def run_sampler(
     else:
         raise ValueError(
             "Provided sampler should be a Sampler object or name of a known "
-            f"sampler: {', '.join(IMPLEMENTED_SAMPLERS.keys())}."
+            f"sampler: {get_implemented_samplers()}."
         )
 
     if sampler.cached_result:

bilby/core/sampler/base_sampler.py

@@ -248,9 +248,10 @@ class Sampler(object):
 
         self.exit_code = exit_code
 
+        self._log_likelihood_eval_time = np.nan
         if not soft_init:
             self._verify_parameters()
-            self._time_likelihood()
+            self._log_likelihood_eval_time = self._time_likelihood()
             self._verify_use_ratio()
 
         self.kwargs = kwargs
@@ -433,6 +434,10 @@ class Sampler(object):
         n_evaluations: int
             The number of evaluations to estimate the evaluation time from
 
+        Returns
+        =======
+        log_likelihood_eval_time: float
+            The time (in s) it took for one likelihood evaluation
         """
 
         t1 = datetime.datetime.now()
@@ -442,15 +447,16 @@ class Sampler(object):
             )[:, 0]
             self.log_likelihood(theta)
         total_time = (datetime.datetime.now() - t1).total_seconds()
-        self._log_likelihood_eval_time = total_time / n_evaluations
+        log_likelihood_eval_time = total_time / n_evaluations
 
-        if self._log_likelihood_eval_time == 0:
-            self._log_likelihood_eval_time = np.nan
+        if log_likelihood_eval_time == 0:
+            log_likelihood_eval_time = np.nan
             logger.info("Unable to measure single likelihood time")
         else:
             logger.info(
                 f"Single likelihood evaluation took {self._log_likelihood_eval_time:.3e} s"
             )
+        return log_likelihood_eval_time
 
     def _verify_use_ratio(self):
         """

bilby/core/sampler/dynesty.py

@@ -239,8 +239,12 @@ class Dynesty(NestedSampler):
         self.nestcheck = nestcheck
 
         if self.n_check_point is None:
-            self.n_check_point = max(
-                int(check_point_delta_t / self._log_likelihood_eval_time / 10), 10
+            self.n_check_point = (
+                10
+                if np.isnan(self._log_likelihood_eval_time)
+                else max(
+                    int(check_point_delta_t / self._log_likelihood_eval_time / 10), 10
+                )
             )
         self.check_point_delta_t = check_point_delta_t
         logger.info(f"Checkpoint every check_point_delta_t = {check_point_delta_t}s")

bilby/core/sampler/dynesty_utils.py

@@ -300,6 +300,8 @@ class ACTTrackingRWalk:
         )
         reject += nfail
         blob = {"accept": accept, "reject": reject, "scale": args.scale}
+        iact = int(np.ceil(self.act))
+        thin = self.thin * iact
 
         if accept == 0:
             logger.debug(
@@ -314,11 +316,9 @@ class ACTTrackingRWalk:
                 "Unable to find a new point using walk: try increasing maxmcmc"
             )
             self._cache.append((current_u, current_v, logl, ncall, blob))
-        elif self.thin == -1:
+        elif (self.thin == -1) or (len(u_list) <= thin):
             self._cache.append((current_u, current_v, logl, ncall, blob))
         else:
-            iact = int(np.ceil(self.act))
-            thin = self.thin * iact
             u_list = u_list[thin::thin]
             v_list = v_list[thin::thin]
             logl_list = logl_list[thin::thin]

bilby/core/utils/__init__.py

@@ -6,6 +6,7 @@ from .constants import *
 from .conversion import *
 from .counter import *
 from .docs import *
+from .entry_points import *
 from .introspection import *
 from .io import *
 from .log import *

bilby/core/utils/entry_points.py

+import sys
+if sys.version_info < (3, 10):
+    from importlib_metadata import entry_points
+else:
+    from importlib.metadata import entry_points
+
+
+def get_entry_points(group):
+    """Return a dictionary of entry points for a given group
+
+    Parameters
+    ----------
+    group: str
+        Entry points you wish to query
+    """
+    return {
+        custom.name: custom for custom in entry_points(group=group)
+    }

bilby/gw/detector/__init__.py

@@ -239,7 +239,7 @@ def load_data_from_cache_file(
         An initialised interferometer object with strain data set to the
         appropriate data in the cache file and a PSD.
     """
-
+    import lal
     data_set = False
     psd_set = False
 

docs/citing-bilby.txt

+=======================================
+Acknowledging/Citing Bilby
+=======================================
+
+If you have used Bilby in your scientific work, please acknowledge us in your papers/proposals. 
+
+ .. code:: bibtex
+
+    @article{bilby_paper,
+        author = "Ashton, Gregory and others",
+        title = "{BILBY: A user-friendly Bayesian inference library for gravitational-wave astronomy}",
+        eprint = "1811.02042",
+        archivePrefix = "arXiv",
+        primaryClass = "astro-ph.IM",
+        doi = "10.3847/1538-4365/ab06fc",
+        journal = "Astrophys. J. Suppl.",
+        volume = "241",
+        number = "2",
+        pages = "27",
+        year = "2019"
+    }
+
+Additionally, if you used `bilby_pipe` or `parallel_bilby`, please cite appropriate papers mentioned below"
+
+ .. code:: bibtex
+
+    @article{bilby_pipe_paper,
+        author = "Romero-Shaw, I. M. and others",
+        title = "{Bayesian inference for compact binary coalescences with bilby: validation and application to the first LIGO\textendash{}Virgo gravitational-wave transient catalogue}",
+        eprint = "2006.00714",
+        archivePrefix = "arXiv",
+        primaryClass = "astro-ph.IM",
+        doi = "10.1093/mnras/staa2850",
+        journal = "Mon. Not. Roy. Astron. Soc.",
+        volume = "499",
+        number = "3",
+        pages = "3295--3319",
+        year = "2020"
+    }
+
+    @article{pbilby_paper,
+        author = "Smith, Rory J. E. and Ashton, Gregory and Vajpeyi, Avi and Talbot, Colm",
+        title = "{Massively parallel Bayesian inference for transient gravitational-wave astronomy}",
+        eprint = "1909.11873",
+        archivePrefix = "arXiv",
+        primaryClass = "gr-qc",
+        reportNumber = "LIGO Document P1900255-v1",
+        doi = "10.1093/mnras/staa2483",
+        journal = "Mon. Not. Roy. Astron. Soc.",
+        volume = "498",
+        number = "3",
+        pages = "4492--4502",
+        year = "2020"
+    }
+
+If you use any of the accelerated likelihoods like `ROQGravitationalWaveTransient`, `MBGravitationalWaveTransient`, `RelativeBinningGravitationalWaveTransient` etc., please cite the following papers in addition to the above.
+
+- `ROQGravitationalWaveTransient`
+ .. code:: bibtex
+
+     @article{roq_paper,
+        author = {Smith, Rory and Field, Scott E. and Blackburn, Kent and Haster, Carl-Johan and P\"urrer, Michael and Raymond, Vivien and Schmidt, Patricia},
+        title = "{Fast and accurate inference on gravitational waves from precessing compact binaries}",
+        eprint = "1604.08253",
+        archivePrefix = "arXiv",
+        primaryClass = "gr-qc",
+        reportNumber = "LIGO-DOCUMENT-NUMBER-P1600096, LIGO-P1600096",
+        doi = "10.1103/PhysRevD.94.044031",
+        journal = "Phys. Rev. D",
+        volume = "94",
+        number = "4",
+        pages = "044031",
+        year = "2016"
+    }
+
+- `MBGravitationalWaveTransient`
+ .. code:: bibtex
+
+    @article{mb_paper,
+        author = "Morisaki, Soichiro",
+        title = "{Accelerating parameter estimation of gravitational waves from compact binary coalescence using adaptive frequency resolutions}",
+        eprint = "2104.07813",
+        archivePrefix = "arXiv",
+        primaryClass = "gr-qc",
+        doi = "10.1103/PhysRevD.104.044062",
+        journal = "Phys. Rev. D",
+        volume = "104",
+        number = "4",
+        pages = "044062",
+        year = "2021"
+    }
+
+
+- `RelativeBinningGravitationalWaveTransient`
+ .. code:: bibtex
+
+    @article{relbin_paper,
+        author = "Cornish, Neil J.",
+        title = "{Fast Fisher Matrices and Lazy Likelihoods}",
+        eprint = "1007.4820",
+        archivePrefix = "arXiv",
+        primaryClass = "gr-qc",
+        month = "7",
+        year = "2010"
+    }   
+
+Please also ensure that you appropriately cite the following:
+
+- The sampler(s) used for your analysis
+- Any additional package(s) that you use to generate waveforms e.g. `lalsuite` (https://doi.org/10.7935/GT1W-FZ16)
+as well as the original scientific papers for each waveform.

docs/dynesty-guide.txt

@@ -38,8 +38,9 @@ new point from the constrained prior. These can be specified using the
    results, however, in some cases, a larger value may be required.
 
 2. :code:`sample="acceptance-walk"`: with this method, at each iteration all MCMC chains
-   are set to the same length. The specific length evolves during the run to yield an
-   average of :code:`naccept` accepted jumps during each chain. This method is well
+   are set to the same length. The specific length evolves during the run so that the
+   number of accepted steps follows a Poisson distribution with mean :code:`naccept`
+   during each chain. This method is well
    suited to parallelised applications, as each MCMC chain will run for the same
    amount of time. The value of :code:`naccept` should be tuned based on the
    application. For example, one could run a single analysis with
@@ -62,7 +63,9 @@ There are a number of keyword arguments that influence these sampling methods:
    for consistency with other codes. Default is 5000. If this limit is reached,
    a warning will be printed during sampling.
 
-#. :code:`proposals`: a list of the desired proposals. The allowed values are
+#. :code:`proposals`: a list of the desired proposals.
+   Each of these proposals can be used with any of the sampling methods described
+   above. The allowed values are
 
   * :code:`diff`: `ter Braak + (2006) <https://doi.org/10.1007/s11222-006-8769-1>`_
     differential evolution. This is the default for :code:`bound="live"` and
@@ -70,7 +73,9 @@ There are a number of keyword arguments that influence these sampling methods:
 
   * :code:`volumetric`: sample from an ellipsoid centered on the current point.
     This is the proposal distribution implemented in :code:`dynesty` and the
-    default for all other :code:`bound` options.
+    default for all other :code:`bound` options. This was the default proposal
+    distribution for :code:`Bilby<2`, however, in many applications it leads to
+    longer autocorrelation times and struggles to explore multi-modal distributions.
 
 Finally, we implement two custom :code:`dynesty.sampler.Sampler` classes to
 facilitate the differential evolution proposal and average acceptance tracking.

docs/faq.txt

-=========================
-Frequency Asked Questions
-=========================
+==========================
+Frequently Asked Questions
+==========================
 
 Plotting questions
 ------------------

docs/index.txt

@@ -9,6 +9,7 @@ Welcome to bilby's documentation!
    :caption: Contents:
 
    installation
+   citing-bilby
    code-overview
    examples
    basics-of-parameter-estimation
@@ -17,6 +18,7 @@ Welcome to bilby's documentation!
    samplers
    dynesty-guide
    bilby-mcmc-guide
+   plugins
    bilby-output
    compact-binary-coalescence-parameter-estimation
    transient-gw-data

docs/plugins.txt

+=======
+Plugins
+=======
+
+----------------
+Defining plugins
+----------------
+
+:code:`bilby` allows for additional customizations/extra features via plugins.
+This allows users to add new functionality without the need to modify the main
+:code:`bilby` codebase, for example to add a new sampler.
+
+To make your plugins discoverable to :code:`bilby`, you need to specify a plugin
+group (which :code:`bilby` knows to search for), a name for the plugin, and the
+python path to your function/class within your package metadata, see `here
+<https://packaging.python.org/en/latest/guides/creating-and-discovering-plugins/#using-package-metadata>`_
+for details. For example, if you have a package called :code:`mypackage` and
+you wish to add a plugin called :code:`my_awesome_plugin` within the group
+:code:`bilby.plugin`, you would specify the following in your `pyproject.toml
+<https://packaging.python.org/en/latest/guides/writing-pyproject-toml/>`_
+file::
+
+    [project.entry-points."bilby.plugin"]
+    my_awesome_plugin = "mypackage.plugin"
+
+Currently :code:`bilby` allows for the following plugin groups:
+
+- :code:`"bilby.samplers"`: group for adding samplers to :code:`bilby`. See :ref:`Sampler plugins` for more details.
+
+
+---------------
+Sampler plugins
+---------------
+
+Sampler plugins can specified via the :code:`"bilby.samplers"` group and these
+are automatically added to the 'known' samplers in :code:`bilby`.
+This allows users to add support for new samplers without having to modify the
+core :code:`bilby` codebase.
+Sampler plugins should implement a sampler class that in inherits from one of
+the following classes:
+
+- :py:class:`bilby.core.sampler.base_sampler.Sampler`
+- :py:class:`bilby.core.sampler.base_sampler.NestedSampler`
+- :py:class:`bilby.core.sampler.base_sampler.MCMCSampler`
+
+We provide a `template <https://github.com/bilby-plugins/sampler-template>`_
+for creating sampler plugins on GitHub.
+
+.. note::
+    When implementing a new sampler plugin, please avoid using a generic name for
+    the plugin (e.g. 'nest', 'mcmc') as this may lead to naming conflicts.
+
+
+Sampler plugin library
+----------------------
+
+This is a list of known sampler plugins. if you don't see your plugin listed
+here, we encourage you to open a
+`merge request <https://git.ligo.org/lscsoft/bilby/-/merge_requests/new>`_ to add it.
+
+- This could be your sampler
+
+
+--------------------------------
+Information for bilby developers
+--------------------------------
+
+Using plugins within bilby
+--------------------------
+
+Within :code:`bilby`, plugins are discovered with the
+:py:func:`bilby.core.utils.get_entry_points` function,
+and can be used throughout the :code:`bilby` infrastructure.
+
+Adding a new plugin group
+-------------------------
+
+If you want to add support for a new plugin group, please
+`open an issue <https://git.ligo.org/lscsoft/bilby/-/issues/new>`_
+to discuss the details with other developers.

examples/core_examples/alternative_samplers/linear_regression_pymc_custom_likelihood.py

@@ -12,6 +12,7 @@ import bilby
 import matplotlib.pyplot as plt
 import numpy as np
 import pymc as pm
+from bilby.core.sampler.pymc import Pymc
 
 # A few simple setup steps
 label = "linear_regression_pymc_custom_likelihood"
@@ -76,13 +77,11 @@ class GaussianLikelihoodPyMC(bilby.core.likelihood.GaussianLikelihood):
         ----------
         sampler: :class:`bilby.core.sampler.Pymc`
             A Sampler object must be passed containing the prior distributions
-            and PyMC3 :class:`~pymc3.Model` to use as a context manager.
+            and PyMC :class:`~pymc.Model` to use as a context manager.
             If this is not passed, the super class is called and the regular
             likelihood is evaluated.
         """
 
-        from bilby.core.sampler import Pymc
-
         if not isinstance(sampler, Pymc):
             return super(GaussianLikelihoodPyMC, self).log_likelihood()
 
@@ -116,15 +115,12 @@ class PyMCUniformPrior(bilby.core.prior.Uniform):
 
     def ln_prob(self, sampler=None):
         """
-        Change ln_prob method to take in a Sampler and return a PyMC3
+        Change ln_prob method to take in a Sampler and return a PyMC
         distribution.
 
-        If the passed argument is not a `Pymc3` sampler, assume that it is a
+        If the passed argument is not a `Pymc` sampler, assume that it is a
         float or array to be passed to the superclass.
         """
-
-        from bilby.core.sampler import Pymc
-
         if not isinstance(sampler, Pymc):
             return super(PyMCUniformPrior, self).ln_prob(sampler)
 

requirements.txt

@@ -10,3 +10,4 @@ dill
 tqdm
 h5py
 attrs
+importlib-metadata>=3.6; python_version < '3.10'

setup.py

@@ -82,7 +82,26 @@ setup(
         "console_scripts": [
             "bilby_plot=cli_bilby.plot_multiple_posteriors:main",
             "bilby_result=cli_bilby.bilby_result:main",
-        ]
+        ],
+        "bilby.samplers": [
+            "bilby_mcmc=bilby.bilby_mcmc.sampler:Bilby_MCMC",
+            "cpnest=bilby.core.sampler.cpnest:Cpnest",
+            "dnest4=bilby.core.sampler.dnest4:DNest4",
+            "dynesty=bilby.core.sampler.dynesty:Dynesty",
+            "dynamic_dynesty=bilby.core.sampler.dynamic_dynesty:DynamicDynesty",
+            "emcee=bilby.core.sampler.emcee:Emcee",
+            "kombine=bilby.core.sampler.kombine:Kombine",
+            "nessai=bilby.core.sampler.nessai:Nessai",
+            "nestle=bilby.core.sampler.nestle:Nestle",
+            "ptemcee=bilby.core.sampler.ptemcee:Ptemcee",
+            "ptmcmcsampler=bilby.core.sampler.ptmcmc:PTMCMCSampler",
+            "pymc=bilby.core.sampler.pymc:Pymc",
+            "pymultinest=bilby.core.sampler.pymultinest:Pymultinest",
+            "pypolychord=bilby.core.sampler.polychord:PyPolyChord",
+            "ultranest=bilby.core.sampler.ultranest:Ultranest",
+            "zeus=bilby.core.sampler.zeus:Zeus",
+            "fake_sampler=bilby.core.sampler.fake_sampler:FakeSampler",
+        ],
     },
     classifiers=[
         "Programming Language :: Python :: 3.9",

test/core/sampler/base_sampler_test.py

@@ -10,9 +10,11 @@ import numpy as np
 import bilby
 from bilby.core import prior
 
+loaded_samplers = {k: v.load() for k, v in bilby.core.sampler.IMPLEMENTED_SAMPLERS.items()}
+
 
 class TestSampler(unittest.TestCase):
-    def setUp(self):
+    def setUp(self, soft_init=False):
         likelihood = bilby.core.likelihood.Likelihood()
         likelihood.parameters = dict(a=1, b=2, c=3)
         delta_prior = prior.DeltaFunction(peak=0)
@@ -36,11 +38,16 @@ class TestSampler(unittest.TestCase):
             outdir=test_directory,
             use_ratio=False,
             skip_import_verification=True,
+            soft_init=soft_init
         )
 
     def tearDown(self):
         del self.sampler
 
+    def test_softinit(self):
+        self.setUp(soft_init=True)
+        self.assertTrue(hasattr(self.sampler, "_log_likelihood_eval_time"))
+
     def test_search_parameter_keys(self):
         expected_search_parameter_keys = ["c"]
         self.assertListEqual(
@@ -165,9 +172,7 @@ class GenericSamplerTest(unittest.TestCase):
 
     @parameterized.expand(samplers)
     def test_pool_creates_properly_no_pool(self, sampler_name):
-        sampler = bilby.core.sampler.IMPLEMENTED_SAMPLERS[sampler_name](
-            self.likelihood, self.priors
-        )
+        sampler = loaded_samplers[sampler_name](self.likelihood, self.priors)
         sampler._setup_pool()
         if sampler_name == "kombine":
             from kombine import SerialPool
@@ -179,7 +184,7 @@ class GenericSamplerTest(unittest.TestCase):
 
     @parameterized.expand(samplers)
     def test_pool_creates_properly_pool(self, sampler):
-        sampler = bilby.core.sampler.IMPLEMENTED_SAMPLERS[sampler](
+        sampler = loaded_samplers[sampler](
             self.likelihood, self.priors, npool=2
         )
         sampler._setup_pool()

test/core/sampler/cpnest_test.py

@@ -2,6 +2,7 @@ import unittest
 from unittest.mock import MagicMock
 
 import bilby
+import bilby.core.sampler.cpnest
 
 
 class TestCPNest(unittest.TestCase):
@@ -10,7 +11,7 @@ class TestCPNest(unittest.TestCase):
         self.priors = bilby.core.prior.PriorDict(
             dict(a=bilby.core.prior.Uniform(0, 1), b=bilby.core.prior.Uniform(0, 1))
         )
-        self.sampler = bilby.core.sampler.Cpnest(
+        self.sampler = bilby.core.sampler.cpnest.Cpnest(
             self.likelihood,
             self.priors,
             outdir="outdir",