Opened 16 years ago
Closed 16 years ago
#29 closed defect (fixed)
Major problems when using subsections
Reported by: | MatthewWhiting | Owned by: | MatthewWhiting |
---|---|---|---|
Priority: | highest | Milestone: | |
Component: | Searching | Version: | 1.1.3 |
Severity: | major | Keywords: | |
Cc: |
Description
When using subsections, we can get objects of different sizes depending on the subsection used, even when the object is nowhere near the edge of the subsection (and so should not be affected by it).
I have attached a few results sets from different runs that use different subsections. Note the first detected object in particular (the Galactic emission), and how its values change for each run. File 1 goes up to z=700, file 2 goes up to z=512, file 3 up to z=1024 (with a specified subsection), and file 4 has no subsection given.
Attachments (4)
Change History (7)
Changed 16 years ago by
Attachment: | results2.txt added |
---|
Changed 16 years ago by
Attachment: | results1.txt added |
---|
Changed 16 years ago by
Attachment: | results3.txt added |
---|
Changed 16 years ago by
Attachment: | results4.txt added |
---|
comment:1 Changed 16 years ago by
Status: | new → assigned |
---|
comment:2 Changed 16 years ago by
Severity: | normal → major |
---|
comment:3 Changed 16 years ago by
Resolution: | → fixed |
---|---|
Status: | assigned → closed |
Note: See
TracTickets for help on using
tickets.
Solved -- I think. The key point here was that wavelet reconstruction was being done, and this will introduce a dependence, for each pixel, on all other pixels (nearby ones more so than far ones). Thus, the reconstruction will be slightly different for different sized subsections. This also explains why the Galactic object shows such large differences -- so many pixels adding up their tiny differences in reconstructed flux...
No such discrepancy was seen when no reconstruction was being done.
So there was no fault, as such, just a subtlety that comes out of the reconstruction process.