Sa/v4 paper

pipeline/bundling/coadder.py

@@ -218,7 +218,7 @@ def bundle(self, bundle_id, map_dir, split_label=None, null_prop_val=None,
             indicating the inter-observation null split that observations
             belong to.
         abscal: dict
-            Nested dict in format {'ws0': {'f090': 1, 'f150': 1}, ...}.
+            Nested dict in format {'f090': {'ws0': 1, 'ws1': 1, ...}, ...}.
             Maps will be multiplied by the abscal factor.
         parallelizor: tuple
             (MPICommExecutor or ProcessPoolExecutor, as_completed_callable, num_workers)
@@ -276,7 +276,7 @@ def __init__(self, bundle_db, freq_channel, car_map_template, map_dir,
             Optional; label indicating the telescope wafer to include.
             If no wafer is provided, coadd maps made for all the wafers.
         abscal: dict
-            Nested dict in format {'ws0': {'f090': 1, 'f150': 1}, ...}.
+            Nested dict in format {'f090': {'ws0': 1, 'ws1': 1, ...}, ...}.
             Maps will be multiplied by the abscal factor.
         bundle_id: int
             Optional; ID corresponding to the bundle that observations belong
@@ -295,7 +295,8 @@ def __init__(self, bundle_db, freq_channel, car_map_template, map_dir,
 
         bundle_info = self._get_bundle_info(bundle_id, map_dir, null_prop_val=null_prop_val, split_label=split_label)
         self.fnames = list(bundle_info['filename'])
-        self.ws = bundle_info['weight'].to_numpy()
+        # copy(): pandas >= 3.0 returns a read-only view, but ws is scaled in place below
+        self.ws = bundle_info['weight'].to_numpy().copy()
 
         self.full_abscal = utils.get_abscal(abscal, bundle_info['wafer'], bundle_info['freq_channel'])
         self.ws *= self.full_abscal**-2  # ivar gets -2 powers of abscal