iraf.net

Hi Mike, Frank, and the IRAF team,

I'm running IRAF 2.16 (installed as part of the Ureka package) on a MAC OS 10.8.5. When running fxcor on a pair of spectra (object.fits and template.fits), I encounter the following error:

cl\$this->_split2($m[0]) fxcor object template
ERROR: arithmetic exception

The irafterm window opens in the usual way, with a small box on top (velocity-like numbers on the x-axis) and a large box below (pixel-like numbers on the x-axis). However, no cross-correlation velocity is found and the focus returns to the command line. I have printed my fxcor parameters here:

objects = object List of object spectra
template= template List of template spectra
(apertur= *) Apertures to be used
(cursor = ) Graphics input cursor

(continu= both) Continuum subtract spectra?
(filter = none) Fourier filter the spectra?
(rebin = smallest) Rebin to which dispersion?
(pixcorr= no) Do a pixel-only correlation?
(osample= *) Object regions to be correlated ('*' =\$this->_split2($m[0]) all)
(rsample= *) Template regions to be correlated
(apodize= 0.2) Apodize end percentage

(functio= gaussian) Function to fit correlation
(width = INDEF) Width of fitting region in pixels
(height = 0.) Starting height of fit
(peak = no) Is height relative to ccf peak?
(minwidt= 3.) Minimum width for fit
(maxwidt= 21.) Maximum width for fit
(weights= 1.) Power defining fitting weights
(backgro= 0.) Background level for fit
(window = INDEF) Size of window in the correlation plot
(wincent= INDEF) Center of peak search window

(output = ) Root spool filename for output
(verbose= long) Verbose output to spool file?
(imupdat= no) Update the image header?
(graphic= stdgraph) Graphics output device

(interac= yes) Interactive graphics?
(autowri= yes) Automatically record results?
(autodra= yes) Automatically redraw fit results?
(ccftype= image) Output type of ccf

(observa= kpno) Observation location database
(continp= ) Continuum processing parameters
(filtpar= ) Filter parameters pset
(keywpar= ) Header keyword translation pset

(mode = ql)

I have successfully run fxcor on other spectra using this installation and these settings. Strangely (?), I can successfully run fxcor on these images *if* object.fits OR template.fits is fed to fxcor as the object (and some other spectrum as the template), but neither object.fits or template.fits work as the template. I have uploaded object.fits and template.fits to the ftp://iraf.noao.edu/pub site.

Any advice you can offer on how to obtain velocities from fxcor for these spectra is much appreciated.

All the best,
Ian

Thanks for the data, it made things much simpler to diagnose.

The problem is that the task isn't getting a UT value from the header and the uninitialized value is causing the error. Normally the UT would come from the DATE-OBS keyword, but in these data the DATE-OBS contains only the date. The task then looks for the keyword specified by the keywpars.ut parameter (default = "UT") which also doesn't exist leading to the empty value.

The workaround is to set the keywpars.ut parameter to be "UT-TIME" for these images to define the start time. I'll modify the task for a future release to trap this error more cleanly.

Hi Mike,

That simple fix worked like a charm. I had flipped through all of those hidden parameter files and didn't pick up on that. Thanks for diagnosing and supplying the fix!

All the best,
Ian