From: Don Groom (deg@panisse.lbl.gov)
Date: Tue Oct 21 2003 - 11:20:42 PDT
Rob is right, of course. It's always good to look at a serious snmin fit
carefull - the *.out file for example, and make sure the MINOS errors are
consistent with the parabolic errors. (In a just world they are usually a
little bigger). If this is to be an important fit, it's Good to diddle
with the snmin02_[SN name}.control file. The result should be insensitive
to the starting conditions, but I found at least one case where the
starting time had an enormous effect on the parabolic dt, while the MINOS
result remained stable. The fits were nearly unchanged, which is also
Good.
I'm not sure if you have the > 2 colors situation, but even if not the new
version snmin02 is a lot slicker and easier to use.
Don
On Mon, 20 Oct 2003, Robert A. Knop Jr. wrote:
> On Mon, Oct 20, 2003 at 01:35:20AM -0700, VAFadeyev@lbl.gov wrote:
> > The existence of the outliers is a bit worrisome;
> > also, there are cases with abonormally small
> > fit errors.
>
> That means that snminuit has broken down. It does that sometimes. You
> can generally fix that by futzing around with snminuit; different
> starting parameters, for example.
>
> -Rob
This archive was generated by hypermail 2.1.4 : Tue Oct 21 2003 - 11:20:58 PDT