Caveats

As with any chi2 fitting routine/procedure, DecompIR is dumb.  It will fit any data you provide it with the best fitting combination of a host-galaxy and AGN component, irrespective of whether this is physically justified or, indeed, necessary.  Therefore, anyone using DecompIR to analyse the infrared SEDs of galaxies should think carefully about how they interpret the results.  

I have only provided a chi2 fitting procedure; it is up to YOU to decide how to interpret the results produced by DecompIR.

In particular, you should consider the following points:

First, the good news:

After performing extensive tests we find that when the host-galaxy component can be well-constrained by the data, then DecompIR generally gives results that are consistent with other measures of AGN:host-galaxy contributions (e.g., emission line ratios etc; see Mullaney et al, 2011).  

What do I mean by well-constrained? 

-Emission at wavelengths > 90um would generally be dominated by the host galaxy, so FIR photometry at these long wavelengths is a good way to constrain the host galaxy component.

-Mid-infrared spectroscopy can also be used to constrain the host-galaxy component via the relative strengths of the PAH features (which are though to be produced in star-forming regions).  Including IR spectra in the input will generally produce reliable results.

-Constraining the host-galaxy component using other means (e.g., radio to star formation rate relation for star-forming galaxies) can also produce reliable results.

Now, the bad news:

When there is only poor constraints on the host-galaxy component, DecompIR tends to overpredict the AGN contribution to the infrared output of a galaxy.  In such cases, tests have shown that there is often no significant difference between fits that only include a host-galaxy component and those that combine a host-galaxy and an AGN component.  This is particularly clear when one considers all possible host-galaxy SEDs (i.e., one host-galaxy template alone may result in a poor fit, other ones may provide a good fit).  Only when an AGN+host-galaxy fit provides a significant improvement over all host-galaxy-alone fits, can one say that DecompIR requires an AGN component to fit the data (and I stress, this only means DecompIR requires an AGN component!).

We hope to release a follow-up study to Mullaney et al (2011) that goes into more details regarding the reliability of DecompIR under different conditions.

I would also like to stress:

Before publishing any results derived from DecompIR, I would recommend that you contact me directly to further explore the validity of your results.  My email address is:

(weird font and image used to confuse trawlers)