Merge branch 'add_prior_tests' into 'master'

Add prior tests

See merge request Monash/tupak!196

CHANGELOG.md

@@ -19,6 +19,7 @@ Changes currently on master, but not under a tag.
 - Adds plotting of the prior on 1D marginal distributions of corner plots
 - Adds a method to plot time-domain GW data
 - Hyperparameter estimation now enables the user to provide the single event evidences
+- Prior and child classes now implement the \_\_eq\_\_ magic method for comparisons
 
 ### Changes
 - Fix construct_cbc_derived_parameters
@@ -27,6 +28,8 @@ Changes currently on master, but not under a tag.
 - PowerSpectralDensity structure modified
 - Fixed bug in get_open_data 
 - .prior files are no longer created. The prior is stored in the result object.
+- Changed the way repr works for priors. The repr can now be used to
+re-instantiate the Prior in most cases
 - Users can now choose to overwrite existing result files, rather than creating
   a .old file.
 

test/gw_prior_tests.py → test/gw_prior_test.py

@@ -12,11 +12,9 @@ class TestBBHPriorSet(unittest.TestCase):
         self.base_directory =\
             '/'.join(os.path.dirname(
                 os.path.abspath(sys.argv[0])).split('/')[:-1])
-        self.filename =\
-            self.base_directory + '/tupak/gw/prior_files/GW150914.prior'
+        self.filename = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'prior_files/binary_black_holes.prior')
         self.default_prior = tupak.gw.prior.BBHPriorSet(
-            filename=self.base_directory\
-                     + '/tupak/gw/prior_files/binary_black_holes.prior')
+            filename=self.filename)
 
     def tearDown(self):
         del self.prior_dict

tupak/core/prior_files/binary_black_holes.prior → test/prior_files/binary_black_holes.prior

-mass_1 = Uniform(name='mass_1', minimum=20, maximum=100)
-mass_2 = Uniform(name='mass_2', minimum=20, maximum=100)
-chirp_mass = Uniform(name='chirp_mass', minimum=25, maximum=100)
-total_mass =  Uniform(name='total_mass', minimum=10, maximum=200)
-mass_ratio =  Uniform(name='mass_ratio', minimum=0.125, maximum=1)
-symmetric_mass_ratio =  Uniform(name='symmetric_mass_ratio', minimum=8 / 81, maximum=0.25)
+# These are the default priors we use for BBH systems.
+# Note that you may wish to use more specific mass and distance parameters.
+# These commands are all known to tupak.gw.prior.
+# Lines beginning "#" are ignored.
+mass_1 = Uniform(name='mass_1', minimum=5, maximum=100)
+mass_2 = Uniform(name='mass_2', minimum=5, maximum=100)
+# chirp_mass = Uniform(name='chirp_mass', minimum=25, maximum=100)
+# total_mass =  Uniform(name='total_mass', minimum=10, maximum=200)
+# mass_ratio =  Uniform(name='mass_ratio', minimum=0.125, maximum=1)
+# symmetric_mass_ratio =  Uniform(name='symmetric_mass_ratio', minimum=8 / 81, maximum=0.25)
 a_1 =  Uniform(name='a_1', minimum=0, maximum=0.8)
 a_2 =  Uniform(name='a_2', minimum=0, maximum=0.8)
 tilt_1 =  Sine(name='tilt_1')
 tilt_2 =  Sine(name='tilt_2')
-cos_tilt_1 =  Uniform(name='cos_tilt_1', minimum=-1, maximum=1)
-cos_tilt_2 =  Uniform(name='cos_tilt_2', minimum=-1, maximum=1)
+# cos_tilt_1 =  Uniform(name='cos_tilt_1', minimum=-1, maximum=1)
+# cos_tilt_2 =  Uniform(name='cos_tilt_2', minimum=-1, maximum=1)
 phi_12 =  Uniform(name='phi_12', minimum=0, maximum=2 * np.pi)
 phi_jl =  Uniform(name='phi_jl', minimum=0, maximum=2 * np.pi)
-luminosity_distance =  UniformComovingVolume(name='luminosity_distance', minimum=1e2, maximum=5e3)
+luminosity_distance =  tupak.gw.prior.UniformComovingVolume(name='luminosity_distance', minimum=1e2, maximum=5e3)
 dec =  Cosine(name='dec')
 ra =  Uniform(name='ra', minimum=0, maximum=2 * np.pi)
 iota =  Sine(name='iota')
-cos_iota =  Uniform(name='cos_iota', minimum=-1, maximum=1)
-psi =  Uniform(name='psi', minimum=0, maximum=2 * np.pi)
+# cos_iota =  Uniform(name='cos_iota', minimum=-1, maximum=1)
+psi =  Uniform(name='psi', minimum=0, maximum=np.pi)
 phase =  Uniform(name='phase', minimum=0, maximum=2 * np.pi)

test/prior_test.py

@@ -4,6 +4,8 @@ import unittest
 from mock import Mock
 import numpy as np
 import os
+import shutil
+from collections import OrderedDict
 
 
 class TestPriorInstantiationWithoutOptionalPriors(unittest.TestCase):
@@ -38,6 +40,27 @@ class TestPriorInstantiationWithoutOptionalPriors(unittest.TestCase):
         expected_string = "Prior(name='test_name', latex_label='test_label', unit=None, minimum=0, maximum=1)"
         self.assertEqual(expected_string, self.prior.__repr__())
 
+    def test_base_prob(self):
+        self.assertTrue(np.isnan(self.prior.prob(5)))
+
+    def test_base_ln_prob(self):
+        self.prior.prob = lambda val: val
+        self.assertEqual(np.log(5), self.prior.ln_prob(5))
+
+    def test_is_in_prior(self):
+        self.prior.minimum = 0
+        self.prior.maximum = 1
+        val_below = self.prior.minimum - 0.1
+        val_at_minimum = self.prior.minimum
+        val_in_prior = (self.prior.minimum + self.prior.maximum)/2.
+        val_at_maximum = self.prior.maximum
+        val_above = self.prior.maximum + 0.1
+        self.assertTrue(self.prior.is_in_prior_range(val_at_minimum))
+        self.assertTrue(self.prior.is_in_prior_range(val_at_maximum))
+        self.assertTrue(self.prior.is_in_prior_range(val_in_prior))
+        self.assertFalse(self.prior.is_in_prior_range(val_below))
+        self.assertFalse(self.prior.is_in_prior_range(val_above))
+
 
 class TestPriorName(unittest.TestCase):
 
@@ -113,7 +136,8 @@ class TestPriorClasses(unittest.TestCase):
             tupak.gw.prior.UniformComovingVolume(name='test', minimum=2e2, maximum=5e3),
             tupak.core.prior.Sine(name='test', unit='unit'),
             tupak.core.prior.Cosine(name='test', unit='unit'),
-            tupak.core.prior.Interped(name='test', unit='unit', xx=np.linspace(0, 10, 1000), yy=np.linspace(0, 10, 1000) ** 4,
+            tupak.core.prior.Interped(name='test', unit='unit', xx=np.linspace(0, 10, 1000),
+                                      yy=np.linspace(0, 10, 1000) ** 4,
                                       minimum=3, maximum=5),
             tupak.core.prior.TruncatedGaussian(name='test', unit='unit', mu=1, sigma=0.4, minimum=-1, maximum=1),
             tupak.core.prior.TruncatedNormal(name='test', unit='unit', mu=1, sigma=0.4, minimum=-1, maximum=1),
@@ -169,9 +193,6 @@ class TestPriorClasses(unittest.TestCase):
     def test_probability_above_domain(self):
         """Test that the prior probability is non-negative in domain of validity and zero outside."""
         for prior in self.priors:
-            # skip delta function prior in this case
-            if isinstance(prior, tupak.core.prior.DeltaFunction):
-                continue
             if prior.maximum != np.inf:
                 outside_domain = np.linspace(prior.maximum + 1, prior.maximum + 1e4, 1000)
                 self.assertTrue(all(prior.prob(outside_domain) == 0))
@@ -179,9 +200,6 @@ class TestPriorClasses(unittest.TestCase):
     def test_probability_below_domain(self):
         """Test that the prior probability is non-negative in domain of validity and zero outside."""
         for prior in self.priors:
-            # skip delta function prior in this case
-            if isinstance(prior, tupak.core.prior.DeltaFunction):
-                continue
             if prior.minimum != -np.inf:
                 outside_domain = np.linspace(prior.minimum - 1e4, prior.minimum - 1, 1000)
                 self.assertTrue(all(prior.prob(outside_domain) == 0))
@@ -189,9 +207,6 @@ class TestPriorClasses(unittest.TestCase):
     def test_probability_in_domain(self):
         """Test that the prior probability is non-negative in domain of validity and zero outside."""
         for prior in self.priors:
-            # skip delta function prior in this case
-            if isinstance(prior, tupak.core.prior.DeltaFunction):
-                continue
             if prior.minimum == -np.inf:
                 prior.minimum = -1e5
             if prior.maximum == np.inf:
@@ -242,6 +257,201 @@ class TestPriorClasses(unittest.TestCase):
             else:
                 self.assertEqual('unit', prior.unit)
 
+    def test_eq_different_classes(self):
+        for i in range(len(self.priors)):
+            for j in range(len(self.priors)):
+                if i == j:
+                    self.assertEqual(self.priors[i], self.priors[j])
+                else:
+                    self.assertNotEqual(self.priors[i], self.priors[j])
+
+    def test_eq_other_condition(self):
+        prior_1 = tupak.core.prior.PowerLaw(name='test', unit='unit', alpha=0, minimum=0, maximum=1)
+        prior_2 = tupak.core.prior.PowerLaw(name='test', unit='unit', alpha=0, minimum=0, maximum=1.5)
+        self.assertNotEqual(prior_1, prior_2)
+
+    def test_eq_different_keys(self):
+        prior_1 = tupak.core.prior.PowerLaw(name='test', unit='unit', alpha=0, minimum=0, maximum=1)
+        prior_2 = tupak.core.prior.PowerLaw(name='test', unit='unit', alpha=0, minimum=0, maximum=1)
+        prior_2.other_key = 5
+        self.assertNotEqual(prior_1, prior_2)
+
+    def test_np_array_eq(self):
+        prior_1 = tupak.core.prior.PowerLaw(name='test', unit='unit', alpha=0, minimum=0, maximum=1)
+        prior_2 = tupak.core.prior.PowerLaw(name='test', unit='unit', alpha=0, minimum=0, maximum=1)
+        prior_1.array_attribute = np.array([1, 2, 3])
+        prior_2.array_attribute = np.array([2, 2, 3])
+        self.assertNotEqual(prior_1, prior_2)
+
+    def test_repr(self):
+        for prior in self.priors:
+            if isinstance(prior, tupak.core.prior.Interped):
+                continue  # we cannot test this because of the numpy arrays
+            elif isinstance(prior, tupak.gw.prior.UniformComovingVolume):
+                repr_prior_string = 'tupak.gw.prior.' + repr(prior)
+            else:
+                repr_prior_string = 'tupak.core.prior.' + repr(prior)
+            repr_prior = eval(repr_prior_string)
+            self.assertEqual(prior, repr_prior)
+
+
+class TestPriorSet(unittest.TestCase):
+
+    def setUp(self):
+        self.first_prior = tupak.core.prior.Uniform(name='a', minimum=0, maximum=1, unit='kg')
+        self.second_prior = tupak.core.prior.PowerLaw(name='b', alpha=3, minimum=1, maximum=2, unit='m/s')
+        self.third_prior = tupak.core.prior.DeltaFunction(name='c', peak=42, unit='m')
+        self.prior_dict = dict(mass=self.first_prior,
+                               speed=self.second_prior,
+                               length=self.third_prior)
+        self.prior_set_from_dict = tupak.core.prior.PriorSet(dictionary=self.prior_dict)
+        self.default_prior_file = os.path.join(os.path.dirname(os.path.realpath(__file__)),
+                                               'prior_files/binary_black_holes.prior')
+        self.prior_set_from_file = tupak.core.prior.PriorSet(filename=self.default_prior_file)
+
+    def tearDown(self):
+        del self.first_prior
+        del self.second_prior
+        del self.third_prior
+        del self.prior_dict
+        del self.prior_set_from_dict
+        del self.default_prior_file
+        del self.prior_set_from_file
+
+    def test_prior_set_is_ordered_dict(self):
+        self.assertIsInstance(self.prior_set_from_dict, OrderedDict)
+
+    def test_prior_set_has_correct_length(self):
+        self.assertEqual(3, len(self.prior_set_from_dict))
+
+    def test_prior_set_has_expected_priors(self):
+        self.assertDictEqual(self.prior_dict, dict(self.prior_set_from_dict))
+
+    # Removed for now as it does not create the outdir in the Gitlab CI
+    # def test_write_to_file(self):
+    #     outdir = 'prior_set_test_outdir'
+    #     label = 'test_label'
+    #     label_dot_prior = label + '.prior'
+    #     current_dir_path = os.path.dirname(os.path.realpath(__file__))
+    #     outdir_path = os.path.join(current_dir_path, outdir)
+    #     outfile_path = os.path.join(outdir_path, label_dot_prior)
+    #     self.prior_set_from_dict.write_to_file(outdir=outdir, label=label)
+    #     self.assertTrue(os.path.isdir(outdir_path))
+    #     self.assertTrue(os.listdir(outdir_path)[0] == label_dot_prior)
+    #     with open(outfile_path, 'r') as f:
+    #         expected_outfile = [
+    #             'speed = PowerLaw(alpha=3, minimum=1, maximum=2, name=\'b\', latex_label=\'b\', unit=\'m/s\')\n',
+    #             'mass = Uniform(minimum=0, maximum=1, name=\'a\', latex_label=\'a\', unit=\'kg\')\n',
+    #             'length = DeltaFunction(peak=42, name=\'c\', latex_label=\'c\', unit=\'m\')\n']
+    #         self.assertListEqual(sorted(expected_outfile), sorted(f.readlines()))
+    #     shutil.rmtree(outdir_path)
+
+    def test_read_from_file(self):
+        expected = dict(mass_1=tupak.core.prior.Uniform(name='mass_1', minimum=5, maximum=100),
+                        mass_2=tupak.core.prior.Uniform(name='mass_2', minimum=5, maximum=100),
+                        a_1=tupak.core.prior.Uniform(name='a_1', minimum=0, maximum=0.8),
+                        a_2=tupak.core.prior.Uniform(name='a_2', minimum=0, maximum=0.8),
+                        tilt_1=tupak.core.prior.Sine(name='tilt_1'),
+                        tilt_2=tupak.core.prior.Sine(name='tilt_2'),
+                        phi_12=tupak.core.prior.Uniform(name='phi_12', minimum=0, maximum=2 * np.pi),
+                        phi_jl=tupak.core.prior.Uniform(name='phi_jl', minimum=0, maximum=2 * np.pi),
+                        luminosity_distance=tupak.gw.prior.UniformComovingVolume(name='luminosity_distance',
+                                                                                 minimum=1e2, maximum=5e3),
+                        dec=tupak.core.prior.Cosine(name='dec'),
+                        ra=tupak.core.prior.Uniform(name='ra', minimum=0, maximum=2 * np.pi),
+                        iota=tupak.core.prior.Sine(name='iota'),
+                        psi=tupak.core.prior.Uniform(name='psi', minimum=0, maximum=np.pi),
+                        phase=tupak.core.prior.Uniform(name='phase', minimum=0, maximum=2 * np.pi)
+                        )
+        self.assertDictEqual(expected, self.prior_set_from_file)
+
+    def test_convert_floats_to_delta_functions(self):
+        self.prior_set_from_dict['d'] = 5
+        self.prior_set_from_dict['e'] = 7.3
+        self.prior_set_from_dict['f'] = 'unconvertable'
+        self.prior_set_from_dict.convert_floats_to_delta_functions()
+        expected = dict(mass=tupak.core.prior.Uniform(name='a', minimum=0, maximum=1, unit='kg'),
+                        speed=tupak.core.prior.PowerLaw(name='b', alpha=3, minimum=1, maximum=2, unit='m/s'),
+                        length=tupak.core.prior.DeltaFunction(name='c', peak=42, unit='m'),
+                        d=tupak.core.prior.DeltaFunction(peak=5),
+                        e=tupak.core.prior.DeltaFunction(peak=7.3),
+                        f='unconvertable')
+        self.assertDictEqual(expected, self.prior_set_from_dict)
+
+    def test_prior_set_from_dict_but_using_a_string(self):
+        prior_set = tupak.core.prior.PriorSet(dictionary=self.default_prior_file)
+        expected = dict(mass_1=tupak.core.prior.Uniform(name='mass_1', minimum=5, maximum=100),
+                        mass_2=tupak.core.prior.Uniform(name='mass_2', minimum=5, maximum=100),
+                        a_1=tupak.core.prior.Uniform(name='a_1', minimum=0, maximum=0.8),
+                        a_2=tupak.core.prior.Uniform(name='a_2', minimum=0, maximum=0.8),
+                        tilt_1=tupak.core.prior.Sine(name='tilt_1'),
+                        tilt_2=tupak.core.prior.Sine(name='tilt_2'),
+                        phi_12=tupak.core.prior.Uniform(name='phi_12', minimum=0, maximum=2 * np.pi),
+                        phi_jl=tupak.core.prior.Uniform(name='phi_jl', minimum=0, maximum=2 * np.pi),
+                        luminosity_distance=tupak.gw.prior.UniformComovingVolume(name='luminosity_distance',
+                                                                                 minimum=1e2, maximum=5e3),
+                        dec=tupak.core.prior.Cosine(name='dec'),
+                        ra=tupak.core.prior.Uniform(name='ra', minimum=0, maximum=2 * np.pi),
+                        iota=tupak.core.prior.Sine(name='iota'),
+                        psi=tupak.core.prior.Uniform(name='psi', minimum=0, maximum=np.pi),
+                        phase=tupak.core.prior.Uniform(name='phase', minimum=0, maximum=2 * np.pi)
+                        )
+        self.assertDictEqual(expected, prior_set)
+
+    def test_dict_argument_is_not_string_or_dict(self):
+        with self.assertRaises(ValueError):
+            tupak.core.prior.PriorSet(dictionary=list())
+
+    def test_sample_subset_correct_size(self):
+        size = 7
+        samples = self.prior_set_from_dict.sample_subset(keys=self.prior_set_from_dict.keys(), size=size)
+        self.assertEqual(len(self.prior_set_from_dict), len(samples))
+        for key in samples:
+            self.assertEqual(size, len(samples[key]))
+
+    def test_sample_subset_correct_size_when_non_priors_in_dict(self):
+        self.prior_set_from_dict['asdf'] = 'not_a_prior'
+        samples = self.prior_set_from_dict.sample_subset(keys=self.prior_set_from_dict.keys())
+        self.assertEqual(len(self.prior_set_from_dict) - 1, len(samples))
+
+    def test_sample_subset_with_actual_subset(self):
+        size = 3
+        samples = self.prior_set_from_dict.sample_subset(keys=['length'], size=size)
+        expected = dict(length=np.array([42., 42., 42.]))
+        self.assertTrue(np.array_equal(expected['length'], samples['length']))
+
+    def test_sample(self):
+        size = 7
+        np.random.seed(42)
+        samples1 = self.prior_set_from_dict.sample_subset(keys=self.prior_set_from_dict.keys(), size=size)
+        np.random.seed(42)
+        samples2 = self.prior_set_from_dict.sample(size=size)
+        self.assertEqual(samples1.keys(), samples2.keys())
+        for key in samples1:
+            self.assertTrue(np.array_equal(samples1[key], samples2[key]))
+
+    def test_prob(self):
+        samples = self.prior_set_from_dict.sample_subset(keys=['mass', 'speed'])
+        expected = self.first_prior.prob(samples['mass']) * self.second_prior.prob(samples['speed'])
+        self.assertEqual(expected, self.prior_set_from_dict.prob(samples))
+
+    def test_ln_prob(self):
+        samples = self.prior_set_from_dict.sample_subset(keys=['mass', 'speed'])
+        expected = self.first_prior.ln_prob(samples['mass']) + self.second_prior.ln_prob(samples['speed'])
+        self.assertEqual(expected, self.prior_set_from_dict.ln_prob(samples))
+
+    def test_rescale(self):
+        theta = [0.5, 0.5, 0.5]
+        expected = [self.first_prior.rescale(0.5),
+                    self.second_prior.rescale(0.5),
+                    self.third_prior.rescale(0.5)]
+        self.assertListEqual(sorted(expected), sorted(self.prior_set_from_dict.rescale(
+            keys=self.prior_set_from_dict.keys(), theta=theta)))
+
+    def test_redundancy(self):
+        for key in self.prior_set_from_dict.keys():
+            self.assertFalse(self.prior_set_from_dict.test_redundancy(key=key))
+
 
 class TestFillPrior(unittest.TestCase):
 
@@ -250,9 +460,9 @@ class TestFillPrior(unittest.TestCase):
         self.likelihood.parameters = dict(a=0, b=0, c=0, d=0, asdf=0, ra=1)
         self.likelihood.non_standard_sampling_parameter_keys = dict(t=8)
         self.priors = dict(a=1, b=1.1, c='string', d=tupak.core.prior.Uniform(0, 1))
-        self.priors = tupak.core.prior.PriorSet(self.priors)
-        self.default_prior_file = os.path.join('/'.join(os.path.dirname(__file__).split('/')[:-1]), 'tupak', 'gw',
-                                               'prior_files', 'binary_black_holes.prior')
+        self.priors = tupak.core.prior.PriorSet(dictionary=self.priors)
+        self.default_prior_file = os.path.join(os.path.dirname(os.path.realpath(__file__)),
+                                               'prior_files/binary_black_holes.prior')
         self.priors.fill_priors(self.likelihood, self.default_prior_file)
 
     def tearDown(self):
@@ -282,5 +492,22 @@ class TestFillPrior(unittest.TestCase):
         self.assertIsInstance(self.priors['ra'], tupak.core.prior.Uniform)
 
 
+class TestCreateDefaultPrior(unittest.TestCase):
+
+    def test_none_behaviour(self):
+        self.assertIsNone(tupak.core.prior.create_default_prior(name='name', default_priors_file=None))
+
+    def test_bbh_params(self):
+        prior_file = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'prior_files/binary_black_holes.prior')
+        prior_set = tupak.core.prior.PriorSet(filename=prior_file)
+        for prior in prior_set:
+            self.assertEqual(prior_set[prior], tupak.core.prior.create_default_prior(name=prior,
+                                                                                     default_priors_file=prior_file))
+
+    def test_unknown_prior(self):
+        prior_file = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'prior_files/binary_black_holes.prior')
+        self.assertIsNone(tupak.core.prior.create_default_prior(name='name', default_priors_file=prior_file))
+
+
 if __name__ == '__main__':
     unittest.main()

tupak/core/prior.py

@@ -12,6 +12,7 @@ from future.utils import iteritems
 from tupak.core.utils import logger
 from tupak.core import utils
 import tupak  # noqa
+import inspect
 
 
 class PriorSet(OrderedDict):
@@ -277,7 +278,6 @@ def create_default_prior(name, default_priors_file=None):
 
 
 class Prior(object):
-
     _default_latex_labels = dict()
 
     def __init__(self, name=None, latex_label=None, unit=None, minimum=-np.inf,
@@ -313,6 +313,20 @@ class Prior(object):
         """
         return self.sample()
 
+    def __eq__(self, other):
+        if self.__class__ != other.__class__:
+            return False
+        if sorted(self.__dict__.keys()) != sorted(other.__dict__.keys()):
+            return False
+        for key in self.__dict__:
+            if type(self.__dict__[key]) is np.ndarray:
+                if not np.array_equal(self.__dict__[key], other.__dict__[key]):
+                    return False
+            else:
+                if not self.__dict__[key] == other.__dict__[key]:
+                    return False
+        return True
+
     def sample(self, size=None):
         """Draw a sample from the prior
 
@@ -374,6 +388,20 @@ class Prior(object):
         """
         return np.log(self.prob(val))
 
+    def is_in_prior_range(self, val):
+        """Returns True if val is in the prior boundaries, zero otherwise
+
+        Parameters
+        ----------
+        val: float
+
+        Returns
+        -------
+        np.nan
+
+        """
+        return (val >= self.minimum) & (val <= self.maximum)
+
     @staticmethod
     def test_valid_for_rescaling(val):
         """Test if 0 < val < 1
@@ -394,38 +422,23 @@ class Prior(object):
     def __repr__(self):
         """Overrides the special method __repr__.
 
-        Should return a representation of this instance that resembles how it is instantiated
+        Returns a representation of this instance that resembles how it is instantiated.
+        Works correctly for all child classes
 
         Returns
         -------
         str: A string representation of this instance
 
         """
-        return self._subclass_repr_helper()
-
-    def _subclass_repr_helper(self, subclass_args=iter([])):
-        """Helps out subclass _repr__ methods by creating a common template
-
-        Parameters
-        ----------
-        subclass_args: list, optional
-            List of attributes in the subclass instance
-
-        Returns
-        -------
-        str: A string representation for this subclass instance.
-
-        """
+        subclass_args = inspect.getargspec(self.__init__).args
+        subclass_args.pop(0)
         prior_name = self.__class__.__name__
-        args = ['name', 'latex_label', 'unit', 'minimum', 'maximum']
-        args.extend(subclass_args)
 
         property_names = [p for p in dir(self.__class__) if isinstance(getattr(self.__class__, p), property)]
         dict_with_properties = self.__dict__.copy()
         for key in property_names:
             dict_with_properties[key] = getattr(self, key)
-
-        args = ', '.join(['{}={}'.format(key, repr(dict_with_properties[key])) for key in args])
+        args = ', '.join(['{}={}'.format(key, repr(dict_with_properties[key])) for key in subclass_args])
         return "{}({})".format(prior_name, args)
 
     @property
@@ -550,14 +563,8 @@ class DeltaFunction(Prior):
         float: np.inf if val = peak, 0 otherwise
 
         """
-        if self.peak == val:
-            return np.inf
-        else:
-            return 0
-
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['peak'])
+        at_peak = (val == self.peak)
+        return np.nan_to_num(np.multiply(at_peak, np.inf))
 
 
 class PowerLaw(Prior):
@@ -618,13 +625,12 @@ class PowerLaw(Prior):
         -------
         float: Prior probability of val
         """
-        in_prior = (val >= self.minimum) & (val <= self.maximum)
         if self.alpha == -1:
-            return np.nan_to_num(1 / val / np.log(self.maximum / self.minimum)) * in_prior
+            return np.nan_to_num(1 / val / np.log(self.maximum / self.minimum)) * self.is_in_prior_range(val)
         else:
             return np.nan_to_num(val ** self.alpha * (1 + self.alpha) /
                                  (self.maximum ** (1 + self.alpha) -
-                                  self.minimum ** (1 + self.alpha))) * in_prior
+                                  self.minimum ** (1 + self.alpha))) * self.is_in_prior_range(val)
 
     def ln_prob(self, val):
         """Return the logarithmic prior probability of val
@@ -638,19 +644,13 @@ class PowerLaw(Prior):
         float:
 
         """
-        in_prior = (val >= self.minimum) & (val <= self.maximum)
-
         if self.alpha == -1:
             normalising = 1. / np.log(self.maximum / self.minimum)
         else:
             normalising = (1 + self.alpha) / (self.maximum ** (1 + self.alpha) -
                                               self.minimum ** (1 + self.alpha))
 
-        return (self.alpha * np.log(val) + np.log(normalising)) + np.log(1. * in_prior)
-
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['alpha'])
+        return (self.alpha * np.log(val) + np.log(normalising)) + np.log(1. * self.is_in_prior_range(val))
 
 
 class Uniform(Prior):
@@ -732,10 +732,6 @@ class LogUniform(PowerLaw):
         if self.minimum <= 0:
             logger.warning('You specified a uniform-in-log prior with minimum={}'.format(self.minimum))
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self)
-
 
 class Cosine(Prior):
 
@@ -779,8 +775,7 @@ class Cosine(Prior):
         -------
         float: Prior probability of val
         """
-        in_prior = (val >= self.minimum) & (val <= self.maximum)
-        return np.cos(val) / 2 * in_prior
+        return np.cos(val) / 2 * self.is_in_prior_range(val)
 
 
 class Sine(Prior):
@@ -825,11 +820,11 @@ class Sine(Prior):
         -------
         float: Prior probability of val
         """
-        in_prior = (val >= self.minimum) & (val <= self.maximum)
-        return np.sin(val) / 2 * in_prior
+        return np.sin(val) / 2 * self.is_in_prior_range(val)
 
 
 class Gaussian(Prior):
+
     def __init__(self, mu, sigma, name=None, latex_label=None, unit=None):
         """Gaussian prior with mean mu and width sigma
 
@@ -875,10 +870,6 @@ class Gaussian(Prior):
     def ln_prob(self, val):
         return -0.5 * ((self.mu - val) ** 2 / self.sigma ** 2 + np.log(2 * np.pi * self.sigma ** 2))
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['mu', 'sigma'])
-
 
 class Normal(Gaussian):
 
@@ -964,13 +955,8 @@ class TruncatedGaussian(Prior):
         -------
         float: Prior probability of val
         """
-        in_prior = (val >= self.minimum) & (val <= self.maximum)
         return np.exp(-(self.mu - val) ** 2 / (2 * self.sigma ** 2)) / (
-            2 * np.pi) ** 0.5 / self.sigma / self.normalisation * in_prior
-
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['mu', 'sigma'])
+            2 * np.pi) ** 0.5 / self.sigma / self.normalisation * self.is_in_prior_range(val)
 
 
 class TruncatedNormal(TruncatedGaussian):
@@ -1020,10 +1006,6 @@ class HalfGaussian(TruncatedGaussian):
                                    name=name, latex_label=latex_label,
                                    unit=unit)
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['sigma'])
-
 
 class HalfNormal(HalfGaussian):
     def __init__(self, sigma, name=None, latex_label=None, unit=None):
@@ -1100,10 +1082,6 @@ class LogNormal(Prior):
     def ln_prob(self, val):
         return scipy.stats.lognorm.logpdf(val, self.sigma, scale=np.exp(self.mu))
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['mu', 'sigma'])
-
 
 class LogGaussian(LogNormal):
     def __init__(self, mu, sigma, name=None, latex_label=None, unit=None):
@@ -1175,10 +1153,6 @@ class Exponential(Prior):
     def ln_prob(self, val):
         return scipy.stats.expon.logpdf(val, scale=self.mu)
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['mu'])
-
 
 class StudentT(Prior):
     def __init__(self, df, mu=0., scale=1., name=None, latex_label=None,
@@ -1239,10 +1213,6 @@ class StudentT(Prior):
     def ln_prob(self, val):
         return scipy.stats.t.logpdf(val, self.df, loc=self.mu, scale=self.scale)
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['df', 'mu', 'scale'])
-
 
 class Beta(Prior):
     def __init__(self, alpha, beta, name=None, latex_label=None, unit=None):
@@ -1320,10 +1290,6 @@ class Beta(Prior):
         else:
             return -np.inf
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['alpha', 'beta'])
-
 
 class Logistic(Prior):
     def __init__(self, mu, scale, name=None, latex_label=None, unit=None):
@@ -1379,10 +1345,6 @@ class Logistic(Prior):
     def ln_prob(self, val):
         return scipy.stats.logistic.logpdf(val, loc=self.mu, scale=self.scale)
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['mu', 'scale'])
-
 
 class Cauchy(Prior):
     def __init__(self, alpha, beta, name=None, latex_label=None, unit=None):
@@ -1438,10 +1400,6 @@ class Cauchy(Prior):
     def ln_prob(self, val):
         return scipy.stats.cauchy.logpdf(val, loc=self.alpha, scale=self.beta)
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['alpha', 'beta'])
-
 
 class Lorentzian(Cauchy):
     def __init__(self, alpha, beta, name=None, latex_label=None, unit=None):
@@ -1522,10 +1480,6 @@ class Gamma(Prior):
     def ln_prob(self, val):
         return scipy.stats.gamma.logpdf(val, self.k, loc=0., scale=self.theta)
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['k', 'theta'])
-
 
 class ChiSquared(Gamma):
     def __init__(self, nu, name=None, latex_label=None, unit=None):
@@ -1551,6 +1505,14 @@ class ChiSquared(Gamma):
         Gamma.__init__(self, name=name, k=nu / 2., theta=2.,
                        latex_label=latex_label, unit=unit)
 
+    @property
+    def nu(self):
+        return int(self.k * 2)
+
+    @nu.setter
+    def nu(self, nu):
+        self.k = nu / 2.
+
 
 class Interped(Prior):
 
@@ -1595,6 +1557,13 @@ class Interped(Prior):
                        maximum=np.nanmin(np.array((max(xx), maximum))))
         self.__initialize_attributes()
 
+    def __eq__(self, other):
+        if self.__class__ != other.__class__:
+            return False
+        if np.array_equal(self.xx, other.xx) and np.array_equal(self.yy, other.yy):
+            return True
+        return False
+
     def prob(self, val):
         """Return the prior probability of val.
 
@@ -1620,10 +1589,6 @@ class Interped(Prior):
             rescaled = float(rescaled)
         return rescaled
 
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['xx', 'yy'])
-
     @property
     def minimum(self):
         """Return minimum of the prior distribution.
@@ -1716,7 +1681,3 @@ class FromFile(Interped):
             logger.warning("Can't load {}.".format(self.id))
             logger.warning("Format should be:")
             logger.warning(r"x\tp(x)")
-
-    def __repr__(self):
-        """Call to helper method in the super class."""
-        return Prior._subclass_repr_helper(self, subclass_args=['id'])