Hoi,
in
http://www.slac.stanford.edu/~ursl/talks/121602
you can find a table with BRBR's and links to scans for various
alternative cut scenarios. Note: in the scans I show the full errors
(on the left) and the quadratic difference to the default (on the
right). Nothing fancy, I am only trying to observe trends, not to
minizime the #sigma.
Observations:
o Applying a lower cut on mm2 cures some of the low-mX behavior. The
B+ are perfect afterwards (one might argue that they are not in the
default, though they are obviously quite good already). The B0 are
corrected into the right direction, but not enough.
o Requiring a lower cut on mm2 *always* lowers BRBR (I played this
game only for the loosened upper cut at mm2 < 1.0).
o Loosening the mm2 cut to be mm2 < 1.0 drives BRBR(B+) down
o Fitting without depletion slightly increases the error, but does
not change much else.
o Q1 drives all BRBR up, most notably with mm2 < 0.5. Once a cut of
-1 < mm2 < 1 is applied, the situation is stable.
o The best stability for B0 (and B+) is for
fit-22: -1.0 < mm2 < 1.0 .and. -1 <= qtot <= 1
fit-32: no depl .and. mm2 < 1.0 .and. -1 <= qtot <= 1
The latter has a quite remarkable consistency between B0 and B+.
So this is a third way (after smearing and a variation) how to
systematically alter the mX scan behavior.
Cheers,
--U.
