iraf.net

I tried running the task scombine using the command:
[code:1:f1c3c227be]scombine input=*_H*L_2.fits output=TXS0827_HL_2.fits w1= 3776892.899268 dw=8.713374e-01 combine=average reject=none[/code:1:f1c3c227be]When I do I get the error:[code:1:f1c3c227be]ERROR: floating point invalid operation[/code:1:f1c3c227be]What can I do to fix it?
Just to test I also tried the task in pyraf and got this error:
[code:1:f1c3c227be]IrafError: Error running IRAF task scombine
IRAF task terminated abnormally
ERROR (502, "floating point invalid operation")[/code:1:f1c3c227be]

Single precision floating point number issues? DW is less than 10e-07 of W1 - 1 part in 10e07 is about the precision limit for 32-bit floating point. OTOH - if scombine does it's business in real*8 it must be something else. Just guessing, of course.Mike

If you're using v2.16 this is new problem and it would help to have some sample data (i.e. can you upload to the anonftp at ftp://iraf.noao.edu/pub) and know whether this is a linux or OSX machine. The 'w1' value does look unusually large and the comment about precision might be right, but I can't tell whether it is really an appropriate value without the header (what units are these?)If this is a v2.15 system then there was a patched SCOMBINE binary posted that fixed a similar problem, you can download this as ftp://iraf.noao.edu/iraf/v215/support/linux64 (or 'macintel' for the OSX binary).

I am using v2.15 and I have loaded the patched SCOMBINE and it work fine on another set of spectra that I have from the same instrument. I have uploaded a zip of 4 files (morrisonss_1.zip on the anonftp(morrisonss.zip was uploaded in error)) that I am trying to combine using my mac. the units are in angstroms. The header files are not very helpful because the files are the result of an rspectext.

I am unable to reproduce the error. However, I think you must be confused about how to do what you want. SCOMBINE is intended for spectra that have a wavelength scale. Your data don't appear to be spectra other than being 1D. The choice of parameters seem very strange. They result in trying to expand things enormously and certainly can run into arithmetic problems.Understanding what you want to do, explained in words, will let me help you find the right way to do what you want.Yours,
Frank Valdes

I am not getting there error now that I tried it again, but when I combine the spectra I get a spectra that is a flat line at Zero for the entire wavelength range. These are spectra with a wavelength scale in Angstroms. I am confused to why you think it is not and as to why you think my parameters are strange. I have 4 spectra (That are from Herschel and are thus brought into iraf from text files using rspectext.) that we are trying to combine to get higher signal to noise.

Never mind I figured out what you were talking about. I had an error reading it in to a fits that I had not noticed and once I fixed that it fixed all of my issues.Thanks for the help.