From: Robert A. Knop Jr. (robert.a.knop@vanderbilt.edu)
Date: Wed May 07 2003 - 10:19:31 PDT
RV=2.54 is what gives RB=3.5, and the latter is the value Phillips
quotes. Since we're generally doing these corrections in that region of
the spectrum (not always, but usually), I figured it was best to use an
RV that gave the RB for a supernova that matched the Phillips value.
(RV is a parameter in the extinction law which in fact is slightly
different from the total-to-selective extinction ratio for a supernova
integrated through the Bessell V filter.)
The primary fits using this -- and keeping the distinction bewteen the
primary and low-extinction subsets -- are at:
http://brahms.phy.vanderbilt.edu/~rknop/scp/hst/#rv254
Chisq/dof is generally 1.3. The chisq is too big generally due to
number of almost-outliers. With the extinction corrected fits, it's
mainly 95bd (overcorrected), 96bo (overcorrected), 92bs (too dim), and
to a lesser degree 9855 (still overcorrected, though not as bad). With
the lower RV, None of these are huge enough outliers that it's obvious
one should omit them from the fits.
With the low-extinction non-corrected fits, it's 92bp (too bright), 9569
(too dim), 9579 (too dim), with only 9569 >2.5.
If we can live with chisq/dof of 1.3, I think we should press forward
using these fits.
The only thing which is sad about the paper is that the extinction
systematic for non-extinction-corrected fits goes up to 0.08 compared to
what I used to think it would be. The two confidence intervals don't
align visually so nicely either, although they are consistent and
extinction corrected fits still have P(lambda>0) > 0.99. (See the plot
posted in Fit 15 at the above link.)
-Rob
-- --Prof. Robert Knop Department of Physics & Astronomy, Vanderbilt University robert.a.knop@vanderbilt.edu
This archive was generated by hypermail 2.1.4 : Wed May 07 2003 - 10:19:34 PDT