ENH: Improve ref-tracking for group keys

pandas/core/groupby/grouper.py

@@ -944,12 +944,17 @@ def is_in_obj(gpr) -> bool:
         if not hasattr(gpr, "name"):
             return False
         if using_copy_on_write():
-            # For the CoW case, we need an equality check as the identity check
-            # no longer works (each Series from column access is a new object)
+            # For the CoW case, we check the references to determine if the
+            # series is part of the object
             try:
-                return gpr.equals(obj[gpr.name])
-            except (AttributeError, KeyError, IndexError, InvalidIndexError):
+                obj_gpr_column = obj[gpr.name]
+            except (KeyError, IndexError, InvalidIndexError):
                 return False
+            if isinstance(gpr, Series) and isinstance(obj_gpr_column, Series):
+                return gpr._mgr.references_same_values(  # type: ignore[union-attr]
+                    obj_gpr_column._mgr, 0  # type: ignore[arg-type]
+                )
+            return False
         try:
             return gpr is obj[gpr.name]
         except (KeyError, IndexError, InvalidIndexError):

pandas/core/internals/managers.py

@@ -11,6 +11,7 @@
     cast,
 )
 import warnings
+import weakref
 
 import numpy as np
 
@@ -258,6 +259,14 @@ def add_references(self, mgr: BaseBlockManager) -> None:
             # "Block"; expected "SharedBlock"
             blk.refs.add_reference(blk)  # type: ignore[arg-type]
 
+    def references_same_values(self, mgr: BaseBlockManager, blkno: int) -> bool:
+        """
+        Checks if two blocks from two different block managers reference the
+        same underlying values.
+        """
+        ref = weakref.ref(self.blocks[blkno])
+        return ref in mgr.blocks[blkno].refs.referenced_blocks
+
     def get_dtypes(self):
         dtypes = np.array([blk.dtype for blk in self.blocks])
         return dtypes.take(self.blknos)