Aggregation with multiscale objects #215

Activity

1. LucaMarconato

   added

   method: aggregation 🔢

   on Apr 5, 2023
2. LucaMarconato commented on Sep 13, 2026

   

   LucaMarconato

   on Sep 13, 2026

   MemberAuthor

   More actions

   This whole message is AI-generated. The issue was automatically discovered and reported by an AI agent (Claude) during an autonomous bug hunt on the spatialdata code base. It has not been verified or triaged by a human yet; the needs: triage label is set so that a maintainer can confirm it. The reproduction script below was executed by the agent in an isolated environment (see Environment) and its output is pasted verbatim.

   I re-attempted to reproduce this issue on the current main branch (commit ccf1ea048d054b6624214bf618008a9f9ae223e0) as part of a triage pass over open issues, and confirmed it reproduces.

   Reproduction script (PEP 723 inline metadata; run with uv run repro.py, no other setup needed):

   # /// script
   # requires-python = ">=3.12"
   # dependencies = [
   #     "spatialdata @ git+https://github.com/scverse/spatialdata.git@ccf1ea048d054b6624214bf618008a9f9ae223e0",
   # ]
   # ///
   """Issue #215: aggregation with multiscale objects (multiscale images in `values`, multiscale labels in `by`)
   and check that multiscale labels are downscaled without interpolation."""
   import numpy as np
   
   from spatialdata import aggregate
   from spatialdata.models import Image2DModel, Labels2DModel
   
   rng = np.random.default_rng(0)
   img = rng.random((2, 32, 32))
   lab = np.zeros((32, 32), dtype=np.int32); lab[2:12, 2:12] = 5; lab[16:30, 16:30] = 9  # uint8 labels crash aggregate(), see #1229
   single_i, single_l = Image2DModel.parse(img), Labels2DModel.parse(lab)
   ms_i, ms_l = Image2DModel.parse(img, scale_factors=[2, 2]), Labels2DModel.parse(lab, scale_factors=[2, 2])
   
   ref = aggregate(values=single_i, by=single_l, agg_func="mean")["table"]
   ok = {}
   for name, v, b in [("ms image by labels", ms_i, single_l), ("image by ms labels", single_i, ms_l), ("ms image by ms labels", ms_i, ms_l)]:
       try:
           t = aggregate(values=v, by=b, agg_func="mean")["table"]
           dense = lambda X: X.toarray() if hasattr(X, 'toarray') else np.asarray(X)  # noqa: E731
           same = np.allclose(dense(t.X), dense(ref.X))
           ok[name] = same
           print(f"{name:22} -> OK, identical to single-scale result: {same}")
       except Exception as e:  # noqa: BLE001
           ok[name] = False
           print(f"{name:22} -> {type(e).__name__}: {str(e)[:100]}")
   
   # label values must be preserved (no interpolation) at lower scales
   vals = {k: np.unique(np.asarray(v["image"])).tolist() for k, v in ms_l.children.items()}
   print("unique label values per scale:", vals)
   no_interp = all(set(v) <= {0, 5, 9} for v in vals.values())
   if all(ok.values()) and no_interp:
       print("VERDICT: NOT REPRODUCED (multiscale values/by work and labels are downscaled without interpolation)")
   else:
       print("VERDICT: REPRODUCED (see failures above)")

   Environment: uv run --no-project --python 3.13, which resolves spatialdata from git+https://github.com/scverse/spatialdata.git@ccf1ea048d054b6624214bf618008a9f9ae223e0 (extra dependencies, if any, are pinned in the script's inline metadata above).

   Output (verbatim, only uv package-resolution/install log lines removed):

   ms image by labels     -> OK, identical to single-scale result: True
   image by ms labels     -> AttributeError: 'DataTree' object has no attribute 'dtype'
   ms image by ms labels  -> AttributeError: 'DataTree' object has no attribute 'dtype'
   unique label values per scale: {'scale0': [0, 5, 9], 'scale1': [0, 5, 9], 'scale2': [0, 5, 9]}
   VERDICT: REPRODUCED (see failures above)
   exit=0
   

   Agent's severity assessment: medium (see the priority: * label added to this issue).

   ---

   Automatically generated; discovered by an AI agent (Claude) and not yet reviewed by a human.
3. LucaMarconato

   added

   priority: medium

   needs: triage

   on Sep 13, 2026