Develop

.github/workflows/unittests.yml

@@ -27,10 +27,7 @@ jobs:
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip
-          python -m pip install flake8 pytest
-          pip install -r requirements.txt
-          pip install -r requirements-tests.txt
-          pip install .
+          pip install .[tests]
       - name: Lint with flake8
         run: |
           # stop the build if there are Python syntax errors or undefined names
@@ -39,4 +36,4 @@ jobs:
           flake8 . --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics
       - name: Test with pytest
         run: |
-          pytest
+          pytest -s

README.md

@@ -359,6 +359,8 @@ before importing JAXNS.
 
 # Change Log
 
+7 Dec, 2024 -- JAXNS 2.6.7 released. Fix pip dependencies install.
+
 13 Nov, 2024 -- JAXNS 2.6.6 released. Minor improvements to plotting.
 
 9 Nov, 2024 -- JAXNS 2.6.5 released. Added gradient guided nested sampling. Removed `num_parallel_workers` in favour

docs/conf.py

@@ -12,7 +12,7 @@
 project = "jaxns"
 copyright = "2024, Joshua G. Albert"
 author = "Joshua G. Albert"
-release = "2.6.6"
+release = "2.6.7"
 
 # -- General configuration ---------------------------------------------------
 # https://www.sphinx-doc.org/en/master/usage/configuration.html#general-configuration

pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "jaxns"
-version = "2.6.6"
+version = "2.6.7"
 description = "Nested Sampling in JAX"
 readme = "README.md"
 requires-python = ">=3.9"
@@ -19,10 +19,7 @@ classifiers = [
     "Operating System :: OS Independent"
 ]
 urls = { "Homepage" = "https://github.com/joshuaalbert/jaxns" }
-
-[project.optional-dependencies]
-# Define the extras here; they will be loaded dynamically from setup.py
-notebooks = []  # Placeholders; extras will load from setup.py
+dynamic = ["dependencies", "optional-dependencies"]
 
 [tool.setuptools]
 include-package-data = true

requirements-tests.txt

@@ -1,4 +1,5 @@
 scikit-learn
 networkx
 psutil
-pytest
+pytest
+flake8

setup.py

@@ -12,6 +12,6 @@ def load_requirements(file_name):
     install_requires=load_requirements("requirements.txt"),
     extras_require={
         "examples": load_requirements("requirements-examples.txt"),
-    },
-    tests_require=load_requirements("requirements-tests.txt"),
+        "tests": load_requirements("requirements-tests.txt"),
+    }
 )

src/jaxns/framework/special_priors.py

@@ -25,7 +25,8 @@
     "Poisson",
     "UnnormalisedDirichlet",
     "Empirical",
-    "TruncationWrapper"
+    "TruncationWrapper",
+    "ExplicitDensityPrior",
 ]
 
 
@@ -72,6 +73,7 @@ def _quantile(self, U):
         sample = jnp.less(U, probs)
         return sample.astype(self.dtype)
 
+
 class Beta(SpecialPrior):
     def __init__(self, *, concentration0=None, concentration1=None, name: Optional[str] = None):
         super(Beta, self).__init__(name=name)
@@ -443,7 +445,8 @@ class Empirical(SpecialPrior):
     Represents the empirical distribution of a set of 1D samples, with arbitrary batch dimension.
     """
 
-    def __init__(self, *, samples: jax.Array, resolution: int = 100, name: Optional[str] = None):
+    def __init__(self, *, samples: jax.Array, support_min: Optional[FloatArray] = None,
+                 support_max: Optional[FloatArray] = None, resolution: int = 100, name: Optional[str] = None):
         super(Empirical, self).__init__(name=name)
         if len(np.shape(samples)) < 1:
             raise ValueError("Samples must have at least one dimension")
@@ -452,6 +455,17 @@ def __init__(self, *, samples: jax.Array, resolution: int = 100, name: Optional[
         if resolution < 1:
             raise ValueError("Resolution must be at least 1")
         samples = jnp.asarray(samples)
+        # Add 1 point for each support endpoint
+        endpoints = []
+        if support_min is not None:
+            endpoints.append(support_min)
+        if support_max is not None:
+            endpoints.append(support_max)
+        if len(endpoints) > 0:
+            samples = jnp.concatenate([samples, jnp.asarray(endpoints)])
+
+        resolution = min(resolution, len(samples) - 1)
+
         self._q = jnp.linspace(0., 100., resolution + 1)
         self._percentiles = jnp.reshape(jnp.percentile(samples, self._q, axis=-1), (resolution + 1, -1))
 

src/jaxns/framework/tests/test_prior.py

@@ -321,15 +321,15 @@ def test_forced_identifiability():
 
 
 def test_empirical():
-    samples = jax.random.normal(jax.random.PRNGKey(42), shape=(5, 2000), dtype=mp_policy.measure_dtype)
+    samples = jax.random.normal(jax.random.PRNGKey(42), shape=(2000,), dtype=mp_policy.measure_dtype)
     prior = Empirical(samples=samples, resolution=100, name='x')
-    assert prior._percentiles.shape == (101, 5)
+    assert prior._percentiles.shape == (101, 1)
 
     x = prior.forward(jnp.ones(prior.base_shape, mp_policy.measure_dtype))
-    assert x.shape == (5,)
+    assert x.shape == ()
     assert jnp.all(jnp.bitwise_not(jnp.isnan(x)))
     x = prior.forward(jnp.zeros(prior.base_shape, mp_policy.measure_dtype))
-    assert x.shape == (5,)
+    assert x.shape == ()
     assert jnp.all(jnp.bitwise_not(jnp.isnan(x)))
 
     x = prior.forward(0.5 * jnp.ones(prior.base_shape, mp_policy.measure_dtype))

src/jaxns/public.py

@@ -55,7 +55,7 @@ class NestedSampler:
     max_samples: Optional[Union[int, float]] = None
     num_live_points: Optional[int] = None
     num_slices: Optional[int] = None
-    s: Optional[int] = None
+    s: Optional[Union[int, float]] = None
     k: Optional[int] = None
     c: Optional[int] = None
     devices: Optional[List[xla_client.Device]] = None
@@ -70,9 +70,9 @@ def __post_init__(self):
         # Determine number of slices per acceptance
         if self.num_slices is None:
             if self.difficult_model:
-                self.s = 10 if self.s is None else int(self.s)
+                self.s = 10 if self.s is None else float(self.s)
             else:
-                self.s = 5 if self.s is None else int(self.s)
+                self.s = 5 if self.s is None else float(self.s)
             if self.s <= 0:
                 raise ValueError(f"Expected s > 0, got s={self.s}")
             self.num_slices = self.model.U_ndims * self.s