Update tests following distributions refactoring

The distributions refactoring moves the random variable sampling to
aesara. This relies on numpy and scipy random variables implementation.
So, now the only thing we care about testing is that the parametrization
on the PyMC side is sendible given the one on the Aesara side
(effectively the numpy/scipy one)

More details can be found on issue pymc-devs#4554
pymc-devs#4554

Author: matteo-pallini

pymc3/tests/test_distributions_random.py

@@ -24,13 +24,13 @@
 import pytest
 import scipy.stats as st
 
+from numpy.testing import assert_almost_equal
 from scipy import linalg
 from scipy.special import expit
 
 import pymc3 as pm
 
 from pymc3.aesaraf import change_rv_size, floatX, intX
-from pymc3.distributions.dist_math import clipped_beta_rvs
 from pymc3.distributions.shape_utils import to_tuple
 from pymc3.exceptions import ShapeError
 from pymc3.tests.helpers import SeededTest, select_by_precision
@@ -524,6 +524,76 @@ def test_dirichlet_random_shape(self, shape, size):
         assert pm.Dirichlet.dist(a=np.ones(shape)).random(size=size).shape == out_shape
 
 
+class TestCorrectParametrizationMappingPymcToScipy(SeededTest):
+    @staticmethod
+    def get_inputs_from_apply_node_outputs(outputs):
+        parents = outputs.get_parents()
+        if not parents:
+            raise Exception("Parent Apply node missing for output")
+        # I am assuming there will always only be 1 Apply parent node in this context
+        return parents[0].inputs
+
+    def test_pymc_params_match_rv_ones(
+        self, pymc_params, expected_aesara_params, pymc_dist, decimal=6
+    ):
+        pymc_dist_output = pymc_dist.dist(**dict(pymc_params))
+        aesera_dist_inputs = self.get_inputs_from_apply_node_outputs(pymc_dist_output)[3:]
+        assert len(expected_aesara_params) == len(aesera_dist_inputs)
+        for (expected_name, expected_value), actual_variable in zip(
+            expected_aesara_params, aesera_dist_inputs
+        ):
+            assert_almost_equal(expected_value, actual_variable.eval(), decimal=decimal)
+
+    def test_normal(self):
+        params = [("mu", 5.0), ("sigma", 10.0)]
+        self.test_pymc_params_match_rv_ones(params, params, pm.Normal)
+
+    def test_uniform(self):
+        params = [("lower", 0.5), ("upper", 1.5)]
+        self.test_pymc_params_match_rv_ones(params, params, pm.Uniform)
+
+    def test_half_normal(self):
+        params, expected_aesara_params = [("sigma", 10.0)], [("mean", 0), ("sigma", 10.0)]
+        self.test_pymc_params_match_rv_ones(params, expected_aesara_params, pm.HalfNormal)
+
+    def test_beta_alpha_beta(self):
+        params = [("alpha", 2.0), ("beta", 5.0)]
+        self.test_pymc_params_match_rv_ones(params, params, pm.Beta)
+
+    def test_beta_mu_sigma(self):
+        params = [("mu", 2.0), ("sigma", 5.0)]
+        expected_alpha, expected_beta = pm.Beta.get_alpha_beta(mu=params[0][1], sigma=params[1][1])
+        expected_params = [("alpha", expected_alpha), ("beta", expected_beta)]
+        self.test_pymc_params_match_rv_ones(params, expected_params, pm.Beta)
+
+    @pytest.mark.skip(reason="Expected to fail due to bug")
+    def test_exponential(self):
+        params = [("lam", 10.0)]
+        expected_params = [("lam", 1 / params[0][1])]
+        self.test_pymc_params_match_rv_ones(params, expected_params, pm.Exponential)
+
+    def test_cauchy(self):
+        params = [("alpha", 2.0), ("beta", 5.0)]
+        self.test_pymc_params_match_rv_ones(params, params, pm.Cauchy)
+
+    def test_half_cauchy(self):
+        params = [("alpha", 2.0), ("beta", 5.0)]
+        self.test_pymc_params_match_rv_ones(params, params, pm.HalfCauchy)
+
+    @pytest.mark.skip(reason="Expected to fail due to bug")
+    def test_gamma_alpha_beta(self):
+        params = [("alpha", 2.0), ("beta", 5.0)]
+        expected_params = [("alpha", params[0][1]), ("beta", 1 / params[1][1])]
+        self.test_pymc_params_match_rv_ones(params, expected_params, pm.Gamma)
+
+    @pytest.mark.skip(reason="Expected to fail due to bug")
+    def test_gamma_mu_sigma(self):
+        params = [("mu", 2.0), ("sigma", 5.0)]
+        expected_alpha, expected_beta = pm.Gamma.get_alpha_beta(mu=params[0][1], sigma=params[1][1])
+        expected_params = [("alpha", expected_alpha), ("beta", 1 / expected_beta)]
+        self.test_pymc_params_match_rv_ones(params, expected_params, pm.Gamma)
+
+
 class TestScalarParameterSamples(SeededTest):
     @pytest.mark.xfail(reason="This distribution has not been refactored for v4")
     def test_bounded(self):
@@ -535,20 +605,6 @@ def ref_rand(size, tau):
 
         pymc3_random(BoundedNormal, {"tau": Rplus}, ref_rand=ref_rand)
 
-    @pytest.mark.skip(reason="This test is covered by Aesara")
-    def test_uniform(self):
-        def ref_rand(size, lower, upper):
-            return st.uniform.rvs(size=size, loc=lower, scale=upper - lower)
-
-        pymc3_random(pm.Uniform, {"lower": -Rplus, "upper": Rplus}, ref_rand=ref_rand)
-
-    @pytest.mark.skip(reason="This test is covered by Aesara")
-    def test_normal(self):
-        def ref_rand(size, mu, sigma):
-            return st.norm.rvs(size=size, loc=mu, scale=sigma)
-
-        pymc3_random(pm.Normal, {"mu": R, "sigma": Rplus}, ref_rand=ref_rand)
-
     @pytest.mark.xfail(reason="This distribution has not been refactored for v4")
     def test_truncated_normal(self):
         def ref_rand(size, mu, sigma, lower, upper):
@@ -587,13 +643,6 @@ def ref_rand(size, alpha, mu, sigma):
 
         pymc3_random(pm.SkewNormal, {"mu": R, "sigma": Rplus, "alpha": R}, ref_rand=ref_rand)
 
-    @pytest.mark.skip(reason="This test is covered by Aesara")
-    def test_half_normal(self):
-        def ref_rand(size, tau):
-            return st.halfnorm.rvs(size=size, loc=0, scale=tau ** -0.5)
-
-        pymc3_random(pm.HalfNormal, {"tau": Rplus}, ref_rand=ref_rand)
-
     @pytest.mark.xfail(reason="This distribution has not been refactored for v4")
     def test_wald(self):
         # Cannot do anything too exciting as scipy wald is a
@@ -607,13 +656,6 @@ def ref_rand(size, mu, lam, alpha):
             ref_rand=ref_rand,
         )
 
-    @pytest.mark.skip(reason="This test is covered by Aesara")
-    def test_beta(self):
-        def ref_rand(size, alpha, beta):
-            return clipped_beta_rvs(a=alpha, b=beta, size=size)
-
-        pymc3_random(pm.Beta, {"alpha": Rplus, "beta": Rplus}, ref_rand=ref_rand)
-
     @pytest.mark.xfail(reason="This distribution has not been refactored for v4")
     def test_laplace(self):
         def ref_rand(size, mu, b):
@@ -648,20 +690,7 @@ def ref_rand(size, nu, mu, lam):
         pymc3_random(pm.StudentT, {"nu": Rplus, "mu": R, "lam": Rplus}, ref_rand=ref_rand)
 
     @pytest.mark.skip(reason="This test is covered by Aesara")
-    def test_cauchy(self):
-        def ref_rand(size, alpha, beta):
-            return st.cauchy.rvs(alpha, beta, size=size)
-
-        pymc3_random(pm.Cauchy, {"alpha": R, "beta": Rplusbig}, ref_rand=ref_rand)
 
-    @pytest.mark.skip(reason="This test is covered by Aesara")
-    def test_half_cauchy(self):
-        def ref_rand(size, beta):
-            return st.halfcauchy.rvs(scale=beta, size=size)
-
-        pymc3_random(pm.HalfCauchy, {"beta": Rplusbig}, ref_rand=ref_rand)
-
-    @pytest.mark.skip(reason="This test is covered by Aesara")
     def test_inverse_gamma(self):
         def ref_rand(size, alpha, beta):
             return st.invgamma.rvs(a=alpha, scale=beta, size=size)