== de Vaucouleurs Model Parameters - Stack Photometry versus Full Force Photometery ==

Back in February we presented the first results of the new IPP programs psphotFullForce
[http://svn.pan-starrs.ifa.hawaii.edu/trac/ipp/wiki/fullforce_SAStest1].

There we presented plots that send to indicate that the full force results provided a much
better match to SDSS model fits than the psphotStack results for de Vaucouleurs models (dev model). There was some scepticism that
the difference was so great. In particular the validity of the [http://svn.pan-starrs.ifa.hawaii.edu/trac/ipp/attachment/wiki/fullforce_SAStest1/ddevrad_hist_comparison.jpg plot] 
was doubted.

It turns out that this scepticism was justified. The ddevrad values for the stack photometry were incorrect. The difference in units between the sdss value (arcseconds) and the IPP values (0.25 arcsecond pixels) was not accounted for. Upon further review we also discovered that the wrong stacks (label SAS.20130620) were used for that investigation rather than the SAS 32 stacks (label SAS.20130703).

In order to examine the dev model results in more detail we re-ran the full force pipeline on a subset of the SAS skycells. To improve statistics we changed
the system to select only dev models for forced photometry rather than the usual method of choosing the best of the 3 model fits. The skycells processed were near the center
of the SAS field: 332 < ra < 336, -3 < dec < 3.

Here are the differences between the SDSS model radius and the psphotStack and psphotFullForce model paramters for dev galaxies 
(devFrac_r == 1.0). Bright galaxies are those with SDSS r magnitude < 20.

{{{
ddevrad = devRad_r*0.25 - EXT_WIDTH_MAJ_dev
}}}


[[Image(ddevrad_stack.jpg)]] [[Image(ddevrad_forced.jpg)]]

Here are both sets of histograms on the same plot:

[[Image(ddevrad_comparison.jpg)]] 

We find that the lowest chi squared fit is shifted slightly to the left indicating that we get a larger radius. Interestingly this effect is larger for the bright objects.

If we plot the difference in pixels between the stack radius value and the best forced fit radius we get the following

[[Image(forcedevrad.jpg)]].

The sharp peak at 0.0 is due to the fact that the forced measurements are done in discrete steps.

Here is the plot of the difference between the sdss magnitude and the stack and forced values..

[[Image(ddevmag.jpg)]]

Apparently the larger radii are causing more flux to be included in the model magnitude.