UnboundLocalError when running DDF.py - 0 Jones-directions

[kendak333]

I'm trying to do some 3GC processing for a MeerKAT field and am coming across an issue when running DDF.py, which appears to be a problem with the output generated from MakeModel.py.

The command being run:
DDF.py ddf_imgDI_clustered.parset --Output-Clobber true

The parset file (in .txt form): ddf_imgDI_clustered.txt

The error:

- 11:12:39 - ClassStokes                  [0.2/0.2 8.9/8.9 0.0Gb] Correlations stored in Measurement set (and gridded) is labeled as ['XX', 'XY', 'YX', 'YY']
 - 11:12:39 - ClassStokes                  [0.2/0.2 8.9/8.9 0.0Gb] Stokes parameters required in FITS cube: ['I']
 - 11:12:39 - ClassFacetMachineTessel      [0.2/0.2 8.9/8.9 0.0Gb] Taking facet directions from Nodes catalog: /scratch/knowles/merghers/t1/out-1/continuum/image_3/J0438-5419_ddfsrcs.reg.ClusterCat.npy
 - 11:12:39 - ClassFacetMachineTessel      [0.2/0.2 8.9/8.9 0.0Gb]   There are 0 Jones-directions
 - 11:12:39 - ClassFacetMachineTessel      [0.2/0.2 8.9/8.9 0.0Gb] Saving Nodes catalog in ddf_DI_clustered.NodesCat.npy (Nfacets:0)
DDFacet has encountered an unexpected error. Dropping you into pdb for a post-mortem.
(This is because you're running with --Debug-Pdb set to 'auto' and --Log-Boring is off.)
Traceback (most recent call last):
  File "/home/knowles/venv-ddfkms/bin/DDF.py", line 461, in <module>
    main(OP, messages)
  File "/home/knowles/venv-ddfkms/bin/DDF.py", line 287, in main
    Imager.Init()
  File "/home/knowles/venv-ddfkms/lib/python3.8/site-packages/DDFacet/Imager/ClassDeconvMachine.py", line 313, in Init
    self.CreateFacetMachines()
  File "/home/knowles/venv-ddfkms/lib/python3.8/site-packages/DDFacet/Imager/ClassDeconvMachine.py", line 352, in CreateFacetMachines
    self.FacetMachine.appendMainField(ImageName="%s.image"%self.BaseName,**MainFacetOptions)
  File "/home/knowles/venv-ddfkms/lib/python3.8/site-packages/DDFacet/Imager/ClassFacetMachine.py", line 301, in appendMainField
    self.setFacetsLocs()
  File "/home/knowles/venv-ddfkms/lib/python3.8/site-packages/DDFacet/Imager/ClassFacetMachineTessel.py", line 334, in setFacetsLocs
    for iFacet, polygon0 in zip(range(len(LPolygon)), LPolygon):
UnboundLocalError: local variable 'LPolygon' referenced before assignment

The J0438-5419_ddfsrcs.reg.ClusterCat.npy file is generated via:

MakeModel.py --ds9PreClusterFile <ds9 region file> --NCluster=0 --BaseImageName <relevant 2GC image>

No errors are generated, other than the warning about Astropy read_ds9 depreciation. The region file looks like the following (I manually replaced the ellipse entries from CARTA with circle ones, but got the same issue):

# Region file format: DS9 CARTA 3.0.0
global color=green dashlist=8 3 width=1 font="helvetica 10 normal roman" select=1 highlite=1 dash=0 fixed=0 edit=1 move=1 delete=1 include=1 source=1
fk5
circle(69.147725970, -54.608262553, 34.7293") # color=#2EE6D6 width=2
circle(68.290231347, -53.727704663, 55.7437") # color=#2EE6D6 width=2
circle(69.416763670, -53.909809815, 51.7000") # color=#2EE6D6 width=2
circle(70.523180227, -54.248390134, 24.2000") # color=#2EE6D6 width=2
circle(69.949181960, -54.676278227, 28.6000") # color=#2EE6D6 width=2

I tried running MakeModel.py with --Ncluster=5 (same number as regions in the file) and got a different error (lack of DicoModel).
I'm not sure where I'm going wrong or whether this is a bug - having checked with colleagues, others are experiencing this issue for certain fields but not for others.

[ktrehaeven]

Hi. Also experiencing this issue.

[cyriltasse]

can you do an np.load("J0438-5419_ddfsrcs.reg.ClusterCat.npy") and see if anything intersting is in there?

[kendak333]

It's indeed empty:

Out[2]: 
array([],
      dtype=[('Name', 'S200'), ('ra', '<f8'), ('dec', '<f8'), ('SumI', '<f8'), ('Cluster', '<i8')])

I wondered if it was something the the ds9 file - after changing the header from # Region file format: DS9 CARTA 3.0.0 to # Region file format: DS9 version 4.1 (from my current install of ds9) and removing the # color=#2EE6D6 width=2 from the end of each region line I get a non-empty *ClusterCat.npy (see below). I still got an empty array if I leave in the # color=#2EE6D6 width=2 and only change the header.

Out[2]: 
array([(b'0', 1.20685554, -0.95309401, 1., 0),
       (b'1', 1.19188941, -0.93772537, 1., 1),
       (b'2', 1.21155107, -0.94090366, 1., 2),
       (b'3', 1.23086166, -0.94681305, 1., 3),
       (b'4', 1.22084355, -0.95428109, 1., 4)],
      dtype=[('Name', 'S200'), ('ra', '<f8'), ('dec', '<f8'), ('SumI', '<f8'), ('Cluster', '<i8')])

So looks like the issue is in the MakeModel.py if the region file is created in CARTA rather than in ds9.

[cyriltasse]

ok cool! :)