From: Robert A. Knop Jr. (robert.a.knop@vanderbilt.edu)
Date: Fri Mar 14 2003 - 05:56:30 PST
One thing Rachel notes:
I remember doing this with STIS data and found TinyTim PSFs to be
consistently too narrow compared to empirical point source images - at
least for the models I was interested in.
I saw this too, long ago-- very frequenly I had a positive residual at
the center of my supernova fits. This went away when I started using
the "electron diffusion" matrix that I got from Andy, which blurred the
Tiny Tim profile out a little bit. After that, I started getting
excellent fits. If you look at the B&W figure from my poster on the
left of:
http://brahms.phy.vanderbilt.edu/~rknop/aasjan2002/
you can see that I'm getting unbiased residuals. I've got these plots
(somewhere) for all the other supernovae (although I have trouble
looking at them, because PGPLOT's greyscale routine does something
screwy that makes the version of ghostscript I'm using barf).
The chisquares for the fits all tended to be good (with one exception,
which is the one close to the host galaxy-- and there I'm pretty sure
it's a couple of high pixels right at the edge due to the background
ramping up beyond my model. I *think* for this SN I cut the patch size
down to 7x7 to avoid having to mess with the background where the galaxy
came in.)
-Rob
-- --Prof. Robert Knop Department of Physics & Astronomy, Vanderbilt University robert.a.knop@vanderbilt.edu
This archive was generated by hypermail 2.1.4 : Fri Mar 14 2003 - 05:56:31 PST