Status: offline
Registered: 03/20/2006
Posts: 3
|
While trying to run the image-subtraction based photometry routine mkdifflc.cl (external), I was given the following error ...ERROR: floating point invalid operation
"imexam (input=imname,frame=1, >> 'junk')"
line 78: ./mkdifflc.cl
called as: `mkdifflc ()'It killed the script immediately and evidently corrupted [b:57d36b08c2]imexam[/b:57d36b08c2] somehow. Now, the script called [b:57d36b08c2]imexam[/b:57d36b08c2] numerous times prior to this without error, and I've certainly worked [b:57d36b08c2]imexamine[/b:57d36b08c2] many times in the past also without error. After getting this error this morning, anytime I display an image in ds9, run [b:57d36b08c2]imexam[/b:57d36b08c2], and try to plot up a radial plot with 'r' i get ERROR: floating point invalid operationand [b:57d36b08c2]imexam[/b:57d36b08c2] quits. I can use the cursor commands 's, m, u, v, e, x, etc', but it doesn't like 'r' or ',' for some reason. I typically start ds9 from an xterm command line, start-up iraf in an xgterm, and display images from the iraf prompt. I've unlearned [b:57d36b08c2]imexam[/b:57d36b08c2], restarted iraf, restarted my computer (running mac os x.5) and nothing is bringing [b:57d36b08c2]imexam[/b:57d36b08c2] back to proper working order. Any suggestions?
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
The problem sounds data-specific (e.g. a region of very small/large/zero pixels), but there's not enough information to tell quite what's going on. If you can reproduce the problem outside of the script by calling IMEXAM directly, then it would help to see the image and the coord where you issue the command. A 'floating point invalid' is a last-resort error meaning only that some sort of FPE occurred, whether this is something the task should have caught depends on whether the script produced an input garbage image. Please post more information.-Mike
|
Status: offline
Registered: 11/11/2005
Posts: 728
|
Hello,I am aware that the profile fitting methods in imexam are unstable and can cause errors. When this happens is data dependent and is most likely to occur either with faint sources or out of focus sources. I would like to understand and trap these errors but have not had time to make it happen with my own data.So the first response is that your problems are probably due to bugs in the imexam profile fitting and the consequences are, unfortunately, as you describe. The second response is that if you can find a case that fails, ideally by first cutting out a small piece around an object rather than in a large image, I would be glad to trace the error and the mathematical problem and make a fix for future releases that makes imexam be better behaved. If you want to do this you can either put some sample data some place at your end that I can access or put data in our anonymous ftp area ftp://iraf.noao.edu/pub. Let me know if you do either of these. You can send mail directly to me at valdes@noao.edu.Yours,
Frank Valdes
|