Merge pull request #279 from ingmarschuster/patch-1

Bugfix powered_exponential.py

gpjax/kernels/stationary/powered_exponential.py

@@ -31,17 +31,19 @@
 
 @dataclass
 class PoweredExponential(AbstractKernel):
-    r"""The powered exponential family of kernels.
+    r"""The powered exponential family of kernels. This also equivalent to the symmetric generalized normal distribution.
 
-    Key reference is Diggle and Ribeiro (2007) - "Model-based Geostatistics".
+    See Diggle and Ribeiro (2007) - "Model-based Geostatistics".
+    and   
+    https://en.wikipedia.org/wiki/Generalized_normal_distribution#Symmetric_version
 
     """
 
     lengthscale: Union[ScalarFloat, Float[Array, " D"]] = param_field(
         jnp.array(1.0), bijector=tfb.Softplus()
     )
     variance: ScalarFloat = param_field(jnp.array(1.0), bijector=tfb.Softplus())
-    power: ScalarFloat = param_field(jnp.array(1.0))
+    power: ScalarFloat = param_field(jnp.array(1.0), bijector=tfb.Sigmoid())
     name: str = "Powered Exponential"
 
     def __call__(self, x: Float[Array, " D"], y: Float[Array, " D"]) -> ScalarFloat:

mkdocs.yml

@@ -89,7 +89,7 @@ plugins:
       execute: true
       allow_errors: false
       include: ["examples/*.py"]
-      ignore: ["examples/utils.py"]
+      ignore: ["examples/utils.py", "_statch/*.py", "scripts/*.py"]
     # binder: true
     # binder_service_name: "gh"
 #     binder_branch: "main"

tests/test_kernels/test_stationary.py

@@ -108,7 +108,7 @@ def test_initialization(self, fields: dict, dim: int) -> None:
             if field in ["variance", "lengthscale", "period", "alpha"]:
                 assert isinstance(meta[field]["bijector"], tfb.Softplus)
             if field in ["power"]:
-                assert isinstance(meta[field]["bijector"], tfb.Identity)
+                assert isinstance(meta[field]["bijector"], tfb.Sigmoid)
 
             # Trainability state
             assert meta[field]["trainable"] is True
@@ -225,7 +225,7 @@ class TestPeriodic(BaseTestKernel):
 class TestPoweredExponential(BaseTestKernel):
     kernel = PoweredExponential
     fields = prod(
-        {"lengthscale": [0.1, 1.0], "variance": [0.1, 1.0], "power": [0.1, 2.0]}
+        {"lengthscale": [0.1, 1.0], "variance": [0.1, 1.0], "power": [0.1, 0.9]}
     )
     params = {"test_initialization": fields}
     default_compute_engine = DenseKernelComputation