From: Robert A. Knop Jr. (robert.a.knop@vanderbilt.edu)
Date: Sat Apr 12 2003 - 20:54:38 PDT
On Sat, Apr 12, 2003 at 04:44:23PM -0700, Greg Aldering wrote:
> But, on page 14 of the paper, in the last paragraph you say "... and the
> host galaxy E(B-V) (measured from the peak color of the lightcurve)."
>
> In the preceding paragraph on that same page you say "... a template
> spectrum ... for each day ... must be modified ... to reflect dust
> extinction ... Reddening effects from dust were calculated given the
> E(B-V) parameter (measured from the lightcurve fits) (sic) for the host
> galaxy."
>
> To me the first says you are using the peak color to get E(B-V) and the
> other says you have a E(B-V) parameter that is used to correct the
> template at *all* lightcurve epochs.
Both are correct.
Snminuit returns R-I at peak. I then figure out what E(B-V) I need to
use with the O'Donnell law in the host galaxy in order to get that R-I
at peak for a supernova of the given stretch, redshift, and Milky Way
E(B-V). I then use that E(B-V) I found to generate the K-corrections
used at all epochs. (And then iterate those two steps iterate until the
thing reproduces itself.)
> If you are using E(B-V) as a parameter for all the lightcurve epochs,
> then the dispersion in U-B must affect all the lightcurve points. This
> will affect the stretch as well, and we don't even know whether that
> hurts or may even help.
Affect on the stretch will be secondary, and it's not obvious what the
effect will be. In a simple picture where supernovae that are UV bright
are always (say) 0.2 magnitudes brighter than a "normal" supernova, it
won't affect the stretch very much at all. (The whole lightcurve just
gets scaled up or scaled down, so the stretch maintains its value. It's
only the scaling of the lightcurve-- which we parametrize with the peak
flux-- that gets affected.) (There will be differences in the
incorrecteness of the K-corrections, but in this case I don't think
they'd be that huge.) On the other hand, if UV-bright supernovae are
bright only at max and return to the U-B colors of a "normal" supernova
sometime after max, then, yes, a simple stretch won't work in U. I
believe I've heard Peter and others assert that a simple stretch does
work in U, and if that's the case, then the "once UV bright, always UV
bright" case is more likely to be an approximation of what's going on.
> If so, then I would advocate letting the scatter of our own
> U-contaminated SNe tell us what intrinsic dispersion to use for the
> final Hubble-diagram points.
There's not nearly enough information to do that. We only have a few
supernovae at high enough redshifts that the U really contributes, and
there aren't nearly enough to do a sort of ridgeline analysis to get an
idea of an intrinsic color just from them. Any scatter could be a
scatter in the reddining too. The problem loops back upon itself
because you use the intrinsic color to get the host galaxy extinction,
but you'd need that host galaxy extinction in order to estimate anything
about the intrinsic colors. (With low-redshift supernovae studied in
multiple bands, you can use the other colors to constrain extinction
befor etalking about U-band.)
-Rob
-- --Prof. Robert Knop Department of Physics & Astronomy, Vanderbilt University robert.a.knop@vanderbilt.edu
This archive was generated by hypermail 2.1.4 : Sat Apr 12 2003 - 20:55:07 PDT