Merge branch 'main' into adding-blackjax-support

Author: zaxtax

.github/workflows/nightly.yml

@@ -0,0 +1,43 @@
+name: nightly
+
+on:
+  schedule:
+    - cron: "0 0 * * *"
+
+jobs:
+  build-and-publish-nightly:
+    name: Build source distribution
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v2
+        with:
+          fetch-depth: 0
+      - uses: actions/setup-python@v2
+        with:
+          python-version: 3.9
+      - name: Install dependencies
+        run: |
+          python -m pip install -U pip
+          python -m pip install build
+      - name: Build the sdist
+        run: python -m build --sdist .
+        env:
+          BUILD_PYMC_NIGHTLY: true
+      - name: Publish to PyPI
+        uses: pypa/[email protected]
+        with:
+          user: __token__
+          password: ${{ secrets.PYPI_TOKEN_PYMC_NIGHTLY }}
+  test-install-job:
+    needs: build-and-publish-nightly
+    runs-on: ubuntu-latest
+    steps:
+    - name: Set up Python
+      uses: actions/setup-python@v2
+      with:
+        python-version: 3.9
+    - name: Give PyPI a chance to update the index
+      run: sleep 240
+    - name: Install from PyPI
+      run: |
+        pip install pymc-nightly==$(grep 'version' pymc/__init__.py | awk '{print $3}' | tr -d '"').dev$(date +"%Y%m%d")

.github/workflows/pytest.yml

@@ -63,7 +63,6 @@ jobs:
             --ignore=pymc/tests/test_distributions_random.py
             --ignore=pymc/tests/test_idata_conversion.py
             --ignore=pymc/tests/test_smc.py
-            --ignore=pymc/tests/test_bart.py
             --ignore=pymc/tests/test_missing.py
 
           - |
@@ -77,7 +76,7 @@ jobs:
             pymc/tests/test_updates.py
             pymc/tests/test_transforms.py
             pymc/tests/test_smc.py
-            pymc/tests/test_bart.py
+            pymc/tests/test_mixture.py
 
           - |
             pymc/tests/test_parallel_sampling.py

RELEASE-NOTES.md

@@ -7,15 +7,13 @@ Instead update the vNext section until 4.0.0 is out.
 ⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠⚠
 -->
 
-## PyMC vNext (4.0.0b1 → 4.0.0b2 → 4.0.0b3 → 4.0.0)
+## PyMC vNext (4.0.0b1 → 4.0.0b2 → 4.0.0b3 → 4.0.0b4 → 4.0.0)
 ⚠ The changes below are the delta between the upcoming releases `v3.11.5` →...→ `v4.0.0`.
 
-### No-yet working features
+### Not-yet working features
 We plan to get these working again, but at this point their inner workings have not been refactored.
 - Timeseries distributions (see [#4642](https://github.com/pymc-devs/pymc/issues/4642))
-- Mixture distributions (see [#4781](https://github.com/pymc-devs/pymc/issues/4781))
-- Cholesky distributions (see WIP PR [#4784](https://github.com/pymc-devs/pymc/pull/4784))
-- Variational inference submodule (see WIP PR [#4582](https://github.com/pymc-devs/pymc/pull/4582))
+- Nested Mixture distributions (see [#5533](https://github.com/pymc-devs/pymc/issues/5533))
 - Elliptical slice sampling (see [#5137](https://github.com/pymc-devs/pymc/issues/5137))
 - `BaseStochasticGradient` (see [#5138](https://github.com/pymc-devs/pymc/issues/5138))
 - `pm.sample_posterior_predictive_w` (see [#4807](https://github.com/pymc-devs/pymc/issues/4807))
@@ -74,6 +72,7 @@ All of the above apply to:
   - In the gp.utils file, the `kmeans_inducing_points` function now passes through `kmeans_kwargs` to scipy's k-means function.
   - The function `replace_with_values` function has been added to `gp.utils`.
   - `MarginalSparse` has been renamed `MarginalApprox`.
+  - Removed `MixtureSameFamily`. `Mixture` is now capable of handling batched multivariate components (see [#5438](https://github.com/pymc-devs/pymc/pull/5438)).
 - ...
 
 ### Expected breaks
@@ -129,6 +128,7 @@ This includes API changes we did not warn about since at least `3.11.0` (2021-01
 - `softmax` and `log_softmax` functions added to `math` module (see [#5279](https://github.com/pymc-devs/pymc/pull/5279)).
 - ...
 
+
 ## Documentation
 - Switched to the [pydata-sphinx-theme](https://pydata-sphinx-theme.readthedocs.io/en/latest/)
 - Updated our documentation tooling to use [MyST](https://myst-parser.readthedocs.io/en/latest/), [MyST-NB](https://myst-nb.readthedocs.io/en/latest/), sphinx-design, notfound.extension,

docs/source/api.rst

@@ -15,7 +15,6 @@ API Reference
    api/smc
    api/backends
    api/data
-   api/bart
    api/ode
    api/tuning
    api/math

docs/source/api/bart.rst

@@ -8,4 +8,3 @@ Mixture
 
    Mixture
    NormalMixture
-   MixtureSameFamily

docs/source/api/distributions/mixture.rst

@@ -13,7 +13,7 @@
 #   limitations under the License.
 
 # pylint: disable=wildcard-import
-__version__ = "4.0.0b2"
+__version__ = "4.0.0b3"
 
 import logging
 import multiprocessing as mp
@@ -52,7 +52,6 @@ def __set_compiler_flags():
 from pymc import gp, ode, sampling
 from pymc.aesaraf import *
 from pymc.backends import *
-from pymc.bart import *
 from pymc.blocking import *
 from pymc.data import *
 from pymc.distributions import *

pymc/__init__.py

@@ -124,6 +124,15 @@ def _run_convergence_checks(self, idata: arviz.InferenceData, model):
             self._add_warnings([warn])
             return
 
+        elif idata.posterior.sizes["chain"] < 4:
+            msg = (
+                "We recommend running at least 4 chains for robust computation of "
+                "convergence diagnostics"
+            )
+            warn = SamplerWarning(WarningType.BAD_PARAMS, msg, "info")
+            self._add_warnings([warn])
+            return
+
         valid_name = [rv.name for rv in model.free_RVs + model.deterministics]
         varnames = []
         for rv in model.free_RVs:
@@ -139,44 +148,25 @@ def _run_convergence_checks(self, idata: arviz.InferenceData, model):
 
         warnings = []
         rhat_max = max(val.max() for val in rhat.values())
-        if rhat_max > 1.4:
+        if rhat_max > 1.01:
             msg = (
-                "The rhat statistic is larger than 1.4 for some "
-                "parameters. The sampler did not converge."
-            )
-            warn = SamplerWarning(WarningType.CONVERGENCE, msg, "error", extra=rhat)
-            warnings.append(warn)
-        elif rhat_max > 1.2:
-            msg = "The rhat statistic is larger than 1.2 for some " "parameters."
-            warn = SamplerWarning(WarningType.CONVERGENCE, msg, "warn", extra=rhat)
-            warnings.append(warn)
-        elif rhat_max > 1.05:
-            msg = (
-                "The rhat statistic is larger than 1.05 for some "
-                "parameters. This indicates slight problems during "
-                "sampling."
+                "The rhat statistic is larger than 1.01 for some "
+                "parameters. This indicates problems during sampling. "
+                "See https://arxiv.org/abs/1903.08008 for details"
             )
             warn = SamplerWarning(WarningType.CONVERGENCE, msg, "info", extra=rhat)
             warnings.append(warn)
 
         eff_min = min(val.min() for val in ess.values())
-        sizes = idata.posterior.sizes
-        n_samples = sizes["chain"] * sizes["draw"]
-        if eff_min < 200 and n_samples >= 500:
+        eff_per_chain = eff_min / idata.posterior.sizes["chain"]
+        if eff_per_chain < 100:
             msg = (
-                "The estimated number of effective samples is smaller than "
-                "200 for some parameters."
+                "The effective sample size per chain is smaller than 100 for some parameters. "
+                " A higher number is needed for reliable rhat and ess computation. "
+                "See https://arxiv.org/abs/1903.08008 for details"
             )
             warn = SamplerWarning(WarningType.CONVERGENCE, msg, "error", extra=ess)
             warnings.append(warn)
-        elif eff_min / n_samples < 0.1:
-            msg = "The number of effective samples is smaller than " "10% for some parameters."
-            warn = SamplerWarning(WarningType.CONVERGENCE, msg, "warn", extra=ess)
-            warnings.append(warn)
-        elif eff_min / n_samples < 0.25:
-            msg = "The number of effective samples is smaller than " "25% for some parameters."
-            warn = SamplerWarning(WarningType.CONVERGENCE, msg, "info", extra=ess)
-            warnings.append(warn)
 
         self._add_warnings(warnings)