Core fixes

pyproject.toml

@@ -23,7 +23,8 @@ classifiers = [
 ]
 requires-python = '>=3.11'
 dependencies = [
-  'easyscience @ git+https://github.com/easyscience/corelib.git@develop',
+  # 'easyscience @ git+https://github.com/easyscience/corelib.git@develop',
+  'easyscience @ git+https://github.com/easyscience/corelib.git@bayesian_extend',
   # 'easyscience',
   'scipp',
   'refnx',

src/easyreflectometry/analysis/bayesian.py

@@ -1332,8 +1332,9 @@ def load_posterior(path: str, skip: int = 0) -> 'PosteriorResults':
     """Reload a trace saved by :func:`save_posterior` into a
     :class:`PosteriorResults`.
 
-    The returned object's ``sampler_state`` can be fed back into
-    ``MultiFitter.mcmc_sample(..., resume_state=...)`` to extend the chain.
+    The returned object's ``sampler_state`` can be fed back into the core
+    ``Sampler`` (via ``Sampler.load_state(...)`` / ``Sampler.extend(...)``)
+    to extend the chain.
 
     :param path: File path prefix used in :func:`save_posterior`.
     :type path: str

src/easyreflectometry/fitting.py

@@ -10,6 +10,7 @@
 import scipp as sc
 from easyscience.fitting import AvailableMinimizers
 from easyscience.fitting import FitResults
+from easyscience.fitting import Sampler
 from easyscience.fitting.multi_fitter import MultiFitter as EasyScienceMultiFitter
 
 from easyreflectometry.data import DataSet1D
@@ -419,14 +420,22 @@
             y.append(y_eff)
             dy.append(weights)
 
-        # Delegate the actual BUMPS/DREAM sampling to the core MultiFitter
+        # Delegate the actual BUMPS/DREAM sampling to the core ``Sampler``.
+        # The core API moved from ``MultiFitter.mcmc_sample()`` to a dedicated
+        # ``Sampler`` class: construct it with the configured fitter and the
+        # bound data, then call ``sample()``. ``Sampler`` handles the
+        # multi-dataset reshaping internally.
         sampler_kwargs = {}
         if initializer is not None:
             sampler_kwargs['init'] = initializer
-        return self.easy_science_multi_fitter.mcmc_sample(
+
+        sampler = Sampler(
+            self.easy_science_multi_fitter,
             x=x,
             y=y,
             weights=dy,
+        )
+        results = sampler.sample(
             samples=samples,
             burn=burn,
             thin=thin,
@@ -435,6 +444,12 @@
             progress_callback=progress_callback,
             abort_test=abort_test,
         )
+        return {
+            'draws': results.draws,
+            'param_names': results.param_names,
+            'state': results.state,
+            'logp': results.logp,
+        }
 
     @property
     def chi2(self) -> float | None: