add logs test

Author: frazane

scoringrules/__init__.py

@@ -3,7 +3,7 @@
 from scoringrules._brier import brier_score
 from scoringrules._crps import crps_ensemble, crps_lognormal, crps_normal
 from scoringrules._energy import energy_score
-from scoringrules._logs import logscore_normal
+from scoringrules._logs import logs_normal
 from scoringrules._variogram import variogram_score
 from scoringrules.backend import register_backend
 
@@ -15,7 +15,7 @@
     "crps_ensemble",
     "crps_normal",
     "crps_lognormal",
-    "logscore_normal",
+    "logs_normal",
     "brier_score",
     "energy_score",
     "variogram_score",

scoringrules/_logs.py

@@ -6,7 +6,7 @@
 ArrayLike = tp.TypeVar("ArrayLike", Array, float)
 
 
-def logscore_normal(
+def logs_normal(
     mu: ArrayLike,
     sigma: ArrayLike,
     observation: ArrayLike,
@@ -37,7 +37,7 @@ def logscore_normal(
     Examples
     --------
     >>> import scoringrules as sr
-    >>> sr.logscore_normal(0.1, 0.4, 0.0)
+    >>> sr.logs_normal(0.1, 0.4, 0.0)
     >>> 0.033898
     """
-    return srb[backend].logscore_normal(mu, sigma, observation)
+    return srb[backend].logs_normal(mu, sigma, observation)

scoringrules/backend/array.py

@@ -155,7 +155,7 @@ def variogram_score(
         out = self.np.sum((obs_diff - vfcts) ** 2, axis=(-2, -1))
         return out
 
-    def logscore_normal(
+    def logs_normal(
         self,
         mu: ArrayLike,
         sigma: ArrayLike,

scoringrules/backend/base.py

@@ -33,7 +33,7 @@ def crps_lognormal(
         """Compute the CRPS for the log normal distribution."""
 
     @abstractmethod
-    def logscore_normal(
+    def logs_normal(
         self,
         mu: ArrayLike,
         sigma: ArrayLike,

scoringrules/backend/gufuncs.py

@@ -42,7 +42,7 @@ def _crps_lognormal_ufunc(mulog: float, sigmalog: float, observation: float) ->
 
 
 @vectorize(["float32(float32, float32, float32)", "float64(float64, float64, float64)"])
-def _logscore_normal_ufunc(mu: np.ndarray, sigma: np.ndarray, observation: np.ndarray):
+def _logs_normal_ufunc(mu: np.ndarray, sigma: np.ndarray, observation: np.ndarray):
     ω = (observation - mu) / sigma
     return -np.log(_norm_pdf(ω) / sigma)
 
@@ -324,7 +324,7 @@ def _variogram_score_gufunc(forecasts, observation, p, out):
     "_crps_ensemble_qd_gufunc",
     "_crps_normal_ufunc",
     "_crps_lognormal_ufunc",
-    "_logscore_normal_ufunc",
+    "_logs_normal_ufunc",
     "_energy_score_gufunc",
     "_brier_score_ufunc",
     "_variogram_score_gufunc",

scoringrules/backend/numba.py

@@ -16,7 +16,7 @@
     _crps_lognormal_ufunc,
     _crps_normal_ufunc,
     _energy_score_gufunc,
-    _logscore_normal_ufunc,
+    _logs_normal_ufunc,
     _variogram_score_gufunc,
 )
 
@@ -87,13 +87,13 @@ def crps_lognormal(
         return _crps_lognormal_ufunc(mulog, sigmalog, observation)
 
     @staticmethod
-    def logscore_normal(
+    def logs_normal(
         mu: ArrayLike,
         sigma: ArrayLike,
         observation: ArrayLike,
     ) -> Array:
         """Compute the logarithmic score for a normal distribution."""
-        return _logscore_normal_ufunc(mu, sigma, observation)
+        return _logs_normal_ufunc(mu, sigma, observation)
 
     @staticmethod
     def energy_score(

tests/test_logs.py

@@ -0,0 +1,14 @@
+import numpy as np
+import pytest
+from scoringrules import _logs
+
+ENSEMBLE_SIZE = 51
+N = 100
+
+BACKENDS = ["numpy", "numba", "jax"]
+
+
+@pytest.mark.parametrize("backend", BACKENDS)
+def test_normal(backend):
+    res = _logs.logs_normal(0.1, 0.1, 0.0, backend=backend)
+    assert np.isclose(res, -0.8836466, rtol=1e-5)