added normal logcdf func and new test domains

Author: Luke LB

pymc/distributions/continuous.py

@@ -85,6 +85,7 @@ def polyagamma_cdf(*args, **kwargs):
     check_parameters,
     clipped_beta_rvs,
     i0e,
+    log_diff_normal_cdf,
     log_normal,
     logpow,
     normal_lccdf,
@@ -743,6 +744,31 @@ def logp(value, mu, sigma, lower, upper):
 
         return logp
 
+    def logcdf(value, mu, sigma, lower, upper):
+        logcdf = log_diff_normal_cdf(mu, sigma, value, lower) - log_diff_normal_cdf(
+            mu, sigma, upper, lower
+        )
+
+        is_lower_bounded = not (
+            isinstance(lower, TensorConstant) and np.all(np.isneginf(lower.value))
+        )
+        is_upper_bounded = not (isinstance(upper, TensorConstant) and np.all(np.isinf(upper.value)))
+
+        if is_lower_bounded:
+            logcdf = pt.switch(value < lower, -np.inf, logcdf)
+
+        if is_upper_bounded:
+            logcdf = pt.switch(value <= upper, logcdf, 0.0)
+
+        if is_lower_bounded and is_upper_bounded:
+            logcdf = check_parameters(
+                logcdf,
+                pt.le(lower, upper),
+                msg="lower_bound <= upper_bound",
+            )
+
+        return logcdf
+
 
 @_default_transform.register(TruncatedNormal)
 def truncated_normal_default_transform(op, rv):

tests/distributions/test_continuous.py

@@ -37,6 +37,7 @@
     Circ,
     Domain,
     R,
+    Rminusbig,
     Rplus,
     Rplusbig,
     Rplusunif,
@@ -934,6 +935,11 @@ def scipy_logp(value, mu, sigma, lower, upper):
                 value, (lower - mu) / sigma, (upper - mu) / sigma, loc=mu, scale=sigma
             )
 
+        def scipy_logcdf(value, mu, sigma, lower, upper):
+            return st.truncnorm.logcdf(
+                value, (lower - mu) / sigma, (upper - mu) / sigma, loc=mu, scale=sigma
+            )
+
         check_logp(
             pm.TruncatedNormal,
             R,
@@ -961,6 +967,33 @@ def scipy_logp(value, mu, sigma, lower, upper):
             skip_paramdomain_outside_edge_test=True,
         )
 
+        check_logcdf(
+            pm.TruncatedNormal,
+            R,
+            {"mu": R, "sigma": Rplusbig, "lower": -Rplusbig, "upper": Rplusbig},
+            scipy_logcdf,
+            decimal=select_by_precision(float64=6, float32=1),
+            skip_paramdomain_outside_edge_test=True,
+        )
+
+        check_logcdf(
+            pm.TruncatedNormal,
+            R,
+            {"mu": R, "sigma": Rplusbig, "upper": Rplusbig},
+            ft.partial(scipy_logcdf, lower=-np.inf),
+            decimal=select_by_precision(float64=6, float32=1),
+            skip_paramdomain_outside_edge_test=True,
+        )
+
+        check_logcdf(
+            pm.TruncatedNormal,
+            R,
+            {"mu": R, "sigma": Rplusbig, "lower": -Rplusbig},
+            ft.partial(scipy_logcdf, upper=np.inf),
+            decimal=select_by_precision(float64=6, float32=1),
+            skip_paramdomain_outside_edge_test=True,
+        )
+
         # This is a regression test for #6128: Check that having one out-of-bound value
         # in an input array does not set all logp values to -inf
         dist = pm.TruncatedNormal.dist(mu=1, sigma=2, lower=0, upper=3)