iraf.net

Edit: should have said I'm running IRAF V2.15 on Fedora 13.I'm running digiphot.apphot.fitpsf on an image to fit circular gaussians (radguass) to stars but find that the values it outputs are always too large when compared to imexam. The help says that when fitting a radgauss function it outputs RSIGMA, where FWHM=2.35*RSIGMA. However, the values it outputs are consistent with the HWHM as measured by other packages, which made me wonder if the output was really what the documentation claimed.Trawling through the source code, I think the task is actually outputting the HWHM, as the apsfradgauss.x file has the following lines:Before fitting, it comes up with an initial guess for par[4] as:
par[4] = (fwhmpsf ** 2 / 4.0)After fitting it converts this back as follows:
par[4] = sqrt (abs(par[4]))and then there doesn't appear to be any more conversion before it's printed. So if fwhmpsf was the correct FWHM, then par[4] will, after fitting, have the value of 0.5*fwhmpsf (since it was not multiplies by 4 before the square root was taken), and this is what gets printed to the screen or file.Hence I think the value that fitpsf outputs claiming to be RSIGMA is really HWHM (in pixels). This explains my results.Can anybody confirm?

Hello,I agree that the code seems to be producing the gaussian sigma for par[4]. The non-linear model being fits determines the variance; hence the square root after the fit is complete produces the sigma.Frank Valdes