iraf.net

Hi,I have a question about the task craverage of the imred/crutil package. Whenever I use the task craverage in a script the following error occurs sometimes:ERROR: floating point invalid operation
cl ()Here is an example code:int n
for(i=179;i<=179;i+=1)
{
for(n=1;n<=8;n+=1)
{
craverage (input="obj"//i//".fits["//n//"]", output="", crmask="crmask_"//i//"_"//n//".pl")
}
}thanks

A 'floating point invalid' is a generic message meaning some operation faulted on the machine, e.g. a divide by a really small number, and uncaught overflow, etc. You say it happens 'sometimes', is it always with the same image or mask? Does it happen if you run the task with that image alone instead of your loop? Does IMSTAT show anything 'unusual' about the pixel values and/or could you either upload the file to the anonftp at ftp://iraf.noao.edu/pub or post a URL so I can download it? If you unlearn the task do you find that some specific parameter setting causes the error?

at first thank you very much for your answear.imstat don't show anything unusual, the error happens quite randomly and only if using the task craverage in an script, when typing the command directly in the console it is working. I set the parameters now to default and the error does not occur any longer, but I am wondering why the previous setting only works in command mode. Also in using the script in the following way produces the error, randomly and not that often but like the same:craverage obj277.fits[im4] output=" " crmask="crmask_277_4.pl"prevoius settings:PACKAGE = crutil
TASK = craverageinput = obj181.fits[im6] List of input images
output = List of output images
(crmask = ) List of output cosmic ray and object masks
(average= ) List of output block average filtered images
(sigma = ) List of output sigma images(navg = 7) Block average box size
(nrej = 3) Number of high pixels to reject from the average
(nbkg = 5) Background annulus width
(nsig = 25) Box size for sigma calculation
(var0 = 0.) Variance coefficient for DN^0 term
(var1 = 0.) Variance coefficient for DN^1 term
(var2 = 0.) Variance coefficient for DN^2 term(crval = 1) Mask value for cosmic rays
(lcrsig = 10.) Low cosmic ray sigma outside object
(hcrsig = 5.) High cosmic ray sigma outside object
(crgrow = 0.) Cosmic ray grow radius(objval = 0) Mask value for objects
(lobjsig= 10.) Low object detection sigma
(hobjsig= 5.) High object detection sigma
(objgrow= 0.) Object grow radius
(mode = ql)************ default parameters -> no error *****************************PACKAGE = crutil
TASK = craverageinput = List of input images
output = List of output images
(crmask = ) List of output cosmic ray and object masks
(average= ) List of output block average filtered images
(sigma = ) List of output sigma images(navg = 5) Block average box size
(nrej = 0) Number of high pixels to reject from the average
(nbkg = 5) Background annulus width
(nsig = 25) Box size for sigma calculation
(var0 = 0.) Variance coefficient for DN^0 term
(var1 = 0.) Variance coefficient for DN^1 term
(var2 = 0.) Variance coefficient for DN^2 term(crval = 1) Mask value for cosmic rays
(lcrsig = 10.) Low cosmic ray sigma outside object
(hcrsig = 5.) High cosmic ray sigma outside object
(crgrow = 0.) Cosmic ray grow radius(objval = 0) Mask value for objects
(lobjsig= 10.) Low object detection sigma
(hobjsig= 5.) High object detection sigma
(objgrow= 0.) Object grow radius
(mode = ql)

Random crashes like this usually indicate a memory bug of some kind, and when running in a script loop the error gradually accumulates until at some point a fault occurs. If true, the you can put a 'flpr' at the bottom of the loop so CRAVERAGE restarts fresh each time.A second possibility is that something about the 'navg' or 'nrej' params (the only ones changed from defaults) is triggering a bug, perhaps because of specific data characteristics. If this were true, then you should be able to reproduce the problem from the commandline on the same image repeatedly. You might try changing only one of the parameters to see if you can determine which is the culprit.In any case, without being able to reproduce the problem here there's not much we can do to fix it. I'd need to know what OS and IRAF version you're using, and will probably need you to upload some data to the anonftp at ftp://iraf.noao.edu/pub once you find a reproducible case (e.g. does one image looped-over 200 times do it, or does it require 200 separate images?).