Possible bug in craverage: floating point overflow error

emma

Hello,

When I run the following command on my linux machine (a "more /etc/redhat-release" gives "Red Hat Enterprise Linux Client release 5.2 (Tikanga)") I have no problems:

fitz · Posted: Wed Jul 06, 2011 9:39 pm Post subject:

Since it's a data-dependent problem then example FITS files would be required to reproduce it, otherwise there are no known issues with the task. Please upload the files to the anonftp at ftp://iraf.noao.edu/pub and I'll have a look.

emma · Posted: Wed Jul 06, 2011 9:48 pm Post subject:

Awesomely quick response time Very Happy

The file has been uploaded ... I called it test_craverage.fits.gz

Thanks Mike! Smile

emma · Posted: Wed Jul 06, 2011 9:58 pm Post subject:

Sorry Mike, I uploaded the wrong file ... I'm trying to upload a new one (with a different name) but it tells me

fitz · Posted: Wed Jul 06, 2011 9:58 pm Post subject:

You sure that's the right image? There is only one MEF extension, and it doesn't have a extname/extver. The error message might be bogus, but the problem may just be that there is no SCI or DQ extension in the image.

emma · Posted: Wed Jul 06, 2011 10:14 pm Post subject:

Ok, I got it, it's called test_craverage_new.fits this time Smile

fitz · Posted: Thu Jul 07, 2011 9:14 am Post subject:

I'm able to reproduce the problem, but it doesn't look trivial. Since this is Frank's task and he's more familiar with the algorithm I've asked him to take over, unfortunately he is away until next week (when I then leave). Somebody will get back to you, but post back if you don't hear anything. Thanks for the report.

emma · Posted: Thu Jul 07, 2011 4:02 pm Post subject:

That's great, thank you! Smile

valdes · IRAF Guru Joined: 11 Nov 2005 Posts: 677

Hi Emma,

I took a look at your example and I have not yet reproduced it. I am waiting to hear from Mike about how he reproduced the error.

But I wanted to also reply to comment on your usage. Craverage is primarily intended for direct images. It is probably not very appropriate for your spectral data. The cosmic rays outside the orders can be easily identified but they don't really have any impact on the later spectral processing. Within the orders a 2D CR detection method, such as craverage, is not a good choice. In fact in my running of this I see the central ridge line of the left 3 orders are totally being flagged as cosmic rays. As a hint: