Use pytensor-distributions for Normal logp, logcdf, logccdf and icdf

conda-envs/environment-alternative-backends.yml

@@ -21,6 +21,7 @@ dependencies:
 - pandas>=0.24.0
 - pip
 - pytensor>=3.2.2,<3.3
+- pytensor-distributions>=0.2.0
 - python-graphviz
 - networkx
 - rich>=13.7.1

conda-envs/environment-dev.yml

@@ -13,6 +13,7 @@ dependencies:
 - pandas>=0.24.0
 - pip
 - pytensor>=3.2.2,<3.3
+- pytensor-distributions>=0.2.0
 - python-graphviz
 - networkx
 - scipy>=1.4.1

conda-envs/environment-docs.yml

@@ -12,6 +12,7 @@ dependencies:
 - pandas>=0.24.0
 - pip
 - pytensor>=3.2.2,<3.3
+- pytensor-distributions>=0.2.0
 - python-graphviz
 - rich>=13.7.1
 - scipy>=1.4.1

conda-envs/environment-test.yml

@@ -16,6 +16,7 @@ dependencies:
 - pip
 - polyagamma
 - pytensor>=3.2.2,<3.3
+- pytensor-distributions>=0.2.0
 - python-graphviz
 - networkx
 - rich>=13.7.1

conda-envs/windows-environment-dev.yml

@@ -13,6 +13,7 @@ dependencies:
 - pandas>=0.24.0
 - pip
 - pytensor>=3.2.2,<3.3
+- pytensor-distributions>=0.2.0
 - python-graphviz
 - networkx
 - rich>=13.7.1

conda-envs/windows-environment-test.yml

@@ -17,6 +17,7 @@ dependencies:
 - pip
 - polyagamma
 - pytensor>=3.2.2,<3.3
+- pytensor-distributions>=0.2.0
 - python-graphviz
 - networkx
 - rich>=13.7.1

pymc/distributions/continuous.py

@@ -51,6 +51,7 @@
 from pytensor.tensor.random.op import RandomVariable
 from pytensor.tensor.random.utils import normalize_size_param
 from pytensor.tensor.variable import TensorConstant, TensorVariable
+from pytensor_distributions import normal as ptd_normal
 
 from pymc.distributions.custom import CustomDist
 from pymc.logprob.abstract import _logprob_helper
@@ -524,32 +525,29 @@ def support_point(rv, size, mu, sigma):
         return mu
 
     def logp(value, mu, sigma):
-        res = -0.5 * pt.pow((value - mu) / sigma, 2) - pt.log(pt.sqrt(2.0 * np.pi)) - pt.log(sigma)
         return check_parameters(
-            res,
+            ptd_normal.logpdf(value, mu, sigma),
             sigma > 0,
             msg="sigma > 0",
         )
 
     def logcdf(value, mu, sigma):
         return check_parameters(
-            normal_lcdf(mu, sigma, value),
+            ptd_normal.logcdf(value, mu, sigma),
             sigma > 0,
             msg="sigma > 0",
         )
 
     def logccdf(value, mu, sigma):
         return check_parameters(
-            normal_lccdf(mu, sigma, value),
+            ptd_normal.logsf(value, mu, sigma),
             sigma > 0,
             msg="sigma > 0",
         )
 
     def icdf(value, mu, sigma):
-        res = mu + sigma * -np.sqrt(2.0) * pt.erfcinv(2 * value)
-        res = check_icdf_value(res, value)
         return check_icdf_parameters(
-            res,
+            ptd_normal.ppf(value, mu, sigma),
             sigma > 0,
             msg="sigma > 0",
         )

requirements-dev.txt

@@ -17,6 +17,7 @@ pandas>=0.24.0
 polyagamma
 pre-commit>=2.8.0
 pymc-sphinx-theme>=0.16.0
+pytensor-distributions>=0.2.0
 pytensor>=3.2.2,<3.3
 pytest-cov>=2.5
 pytest>=3.0

requirements.txt

@@ -4,6 +4,7 @@ cloudpickle
 numpy>=1.25.0
 pandas>=0.24.0
 pytensor>=3.2.2,<3.3
+pytensor-distributions>=0.2.0
 rich>=13.7.1
 scipy>=1.4.1
 threadpoolctl>=3.1.0,<4.0.0