BUG: Fixes regression in Series.pow with all-NA double[pyarrow] values in pandas 3.x (#62520)

Author: fandimeng-fm

doc/source/whatsnew/v3.0.0.rst

@@ -1226,6 +1226,7 @@ Other
 - Fixed bug in the :meth:`Series.rank` with object dtype and extremely small float values (:issue:`62036`)
 - Fixed bug where the :class:`DataFrame` constructor misclassified array-like objects with a ``.name`` attribute as :class:`Series` or :class:`Index` (:issue:`61443`)
 - Fixed regression in :meth:`DataFrame.from_records` not initializing subclasses properly (:issue:`57008`)
+- Fixed regression in :meth:`Series.pow` on Series with all-NA ``float64[pyarrow]`` values; this now returns Series with :class:`NA` values (:issue:`62520`)
 
 .. ***DO NOT USE THIS SECTION***
 

pandas/core/arrays/arrow/array.py

@@ -1051,7 +1051,7 @@ def _arith_method(self, other, op) -> Self | npt.NDArray[np.object_]:
         result = self._evaluate_op_method(other, op, ARROW_ARITHMETIC_FUNCS)
         if is_nan_na() and result.dtype.kind == "f":
             parr = result._pa_array
-            mask = pc.is_nan(parr).to_numpy()
+            mask = pc.is_nan(parr).fill_null(False).to_numpy()
             arr = pc.replace_with_mask(parr, mask, pa.scalar(None, type=parr.type))
             result = type(self)(arr)
         return result

pandas/tests/extension/test_arrow.py

@@ -3691,3 +3691,12 @@ def test_setitem_float_nan_is_na(using_nan_is_na):
         ser[2] = np.nan
         assert isinstance(ser[2], float)
         assert np.isnan(ser[2])
+
+
+def test_pow_with_all_na_float():
+    # GH#62520
+
+    s = pd.Series([None, None], dtype="float64[pyarrow]")
+    result = s.pow(2)
+    expected = pd.Series([pd.NA, pd.NA], dtype="float64[pyarrow]")
+    tm.assert_series_equal(result, expected)