From: Saul Perlmutter (saul@lbl.gov)
Date: Thu Sep 16 2004 - 15:16:52 PDT
...Greg will have to corroborate my memory on this, but I believe that
we did know about this complication in doing the fitting correctly.
This can be checked by looking at that tar'ed collection of fits for the
P99 paper. --Saul
Alexander Conley wrote:
> I have now verified (using SN1998as) that snminuit behaves as I
> have stated.
>
> For a B only fit to 98as, tbmax = -12.572 which corresponds to
> the real date 50896.03. If I do a V band fit where tbmax is forced
> to be -12.572, the date of B maximum comes out to be 50913.82.
> Furthermore, a joint B and V fit with tbmax fixed is different that
> either of the individual fits, with the date of B maximum 50912.65.
>
> This procedure furthermore introduces a bias between the low
> and high redshift sample. With high quality data it is possible to
> tell that the date of V maximum is 1-2 days later than the date of
> B maximum, so the V fits will be systematically wrong by this
> amount. At high redshift there are likely to not be observations that
> can pick up on this difference, so the brightest B and V points will
> probably be the same, and therefore most of the high redshift SNe
> will not be affected by this problem.
>
> That is, I expect that the (B-V)max colors in P99 are systematically
> offset between the low and high redshift sample.
>
> Alex
>
This archive was generated by hypermail 2.1.4 : Thu Sep 16 2004 - 15:16:59 PDT