alchemist-nrel 0.2.1__py3-none-any.whl → 0.3.1__py3-none-any.whl

alchemist_core/__init__.py

@@ -10,6 +10,7 @@ Main Components:
     - ExperimentManager: Manage experimental data
     - Models: Surrogate modeling backends (sklearn, BoTorch)
     - Acquisition: Acquisition function strategies
+    - AuditLog: Reproducible audit trail for optimization decisions
 
 Example:
     >>> from alchemist_core import OptimizationSession, SearchSpace, ExperimentManager
@@ -26,10 +27,10 @@ Example:
     >>> # Get next experiment suggestion
     >>> next_point = session.suggest_next(acq_func="ei")
 
-Version: 0.2.0-dev (Core-UI Split)
+Version: 0.3.0-beta.1
 """
 
-__version__ = "0.2.0-dev"
+__version__ = "0.3.0b1"
 __author__ = "Caleb Coatney"
 __email__ = "caleb.coatney@nrel.gov"
 
@@ -37,6 +38,9 @@ __email__ = "caleb.coatney@nrel.gov"
 from alchemist_core.data.search_space import SearchSpace
 from alchemist_core.data.experiment_manager import ExperimentManager
 
+# Audit log and session metadata
+from alchemist_core.audit_log import AuditLog, SessionMetadata, AuditEntry
+
 # Event system
 from alchemist_core.events import EventEmitter
 
@@ -48,16 +52,19 @@ from alchemist_core.session import OptimizationSession
 
 # Public API
 __all__ = [
-    # High-level API (recommended entry point)
+    # Session API
     "OptimizationSession",
-    
     # Data structures
     "SearchSpace",
     "ExperimentManager",
-    
-    # Events and logging
+    # Audit log
+    "AuditLog",
+    "SessionMetadata", 
+    "AuditEntry",
+    # Event system
     "EventEmitter",
+    # Configuration
     "configure_logging",
     "get_logger",
-    "set_verbosity",
+    "set_verbosity"
 ]

alchemist_core/acquisition/botorch_acquisition.py

@@ -268,6 +268,7 @@ class BoTorchAcquisition(BaseAcquisition):
                     "batch_limit": batch_limit,
                     "maxiter": max_iter,
                     "ftol": 1e-3,  # More relaxed convergence criteria
+                    "factr": None, # Required when ftol is specified
                 }
             else:
                 # Standard parameters for other acquisition functions
@@ -392,10 +393,13 @@ class BoTorchAcquisition(BaseAcquisition):
         if self.batch_size > 1:
             result_points = []
             
+            # Use original feature names (before encoding)
+            feature_names = self.model.original_feature_names
+            
             # Convert each point in the batch to a dictionary with feature names
             for i in range(best_candidates.shape[0]):
                 point_dict = {}
-                for j, name in enumerate(self.model.feature_names):
+                for j, name in enumerate(feature_names):
                     value = best_candidates[i, j]
                     
                     # If this is a categorical variable, convert back to original value
@@ -418,8 +422,11 @@ class BoTorchAcquisition(BaseAcquisition):
             return result_points
         
         # For single-point results (q = 1)
+        # Use original feature names (before encoding)
+        feature_names = self.model.original_feature_names
+        
         result = {}
-        for i, name in enumerate(self.model.feature_names):
+        for i, name in enumerate(feature_names):
             value = best_candidates[0, i]
             
             # If this is a categorical variable, convert back to original value
@@ -448,10 +455,11 @@ class BoTorchAcquisition(BaseAcquisition):
                 return bounds_tensor
         
         # Get feature names from model to ensure proper order
-        if not hasattr(self.model, 'feature_names'):
-            raise ValueError("Model doesn't have feature_names attribute")
+        if not hasattr(self.model, 'original_feature_names'):
+            raise ValueError("Model doesn't have original_feature_names attribute")
         
-        feature_names = self.model.feature_names
+        # Use original_feature_names (before encoding) for bounds extraction
+        feature_names = self.model.original_feature_names
         
         # Get categorical variables
         categorical_variables = []
@@ -636,8 +644,9 @@ class BoTorchAcquisition(BaseAcquisition):
             best_candidate = best_x.cpu().numpy().reshape(1, -1)
             
         # Convert to dictionary and then to DataFrame
+        feature_names = self.model.original_feature_names
         result = {}
-        for i, name in enumerate(self.model.feature_names):
+        for i, name in enumerate(feature_names):
             value = best_candidate[0, i]
             
             # If this is a categorical variable, convert back to original value