From zutshi@fnal.gov Fri Jan 31 11:14:06 2003
Date: Fri, 31 Jan 2003 02:21:06 -0600
From: Vishnu Zutshi 
To: D0JETMET@listserv.fnal.gov
Subject: [D0JETMET] jets and noise

Please find the attached and the ensuing text to summarize the discussion
I had with Leslie last evening. This is partly my response to Emmanuel's
plots and partly an attempt to re-foist my prejudice.

fig0 == no. of jets in QCD40 no-noise (no zsup) events
fig1 == no. of jets in noise only (neutrino) events (2.0sigma zsup)
fig2 ==     "     "     "      "        "      "    (2.5sigma zsup)
fig3 == no. of jets in QCD40 files (2.0sigma zsup)
fig4 == "   "   "    "    "    "   (2.5sigma zsup)
fig5 == f90(jetpt>10) for QCD40 files  (2.5sigma zsup)
fig6 == "         "     "     "    "   (nonoise/nozsup)

So how do I interpret this:

a)fig 1 and 2 show what gaussian noise does in terms of jet creation. Point
to be noted is that 2.5sigma cut neutralizes jet creation due to gaussian
noise pretty well (as expected).
b)One might then expect that fig3 then should be appox. a sum of fig0 and
fig1. That it is.
c)Fig4 should can then be expected to be appox. the sum of fig0 and fig2
i.e. you expect fig. 4 to look very much like fig0. They do.
d) Given the evidence above my claim is that fig. 4 does not contain any
significant amount of pure noise jets.
e)However the f90 plot in fig5 does show a double bump with~10% of the
entries in the smaller bump as is the case with data processed with
2.5sigma.
f)Comparison to fig6 also shows that not only does the second bump
increase due to the presence of noise in the sorrunding environment the
lower peak also shifts significantly from about 0.35 to 0.5 (if memory
serves me the lower peak in data is also around 0.5?).

My conclusion is that with 2.5sigma suppression the no. of pure noise jets
is minimal (ofcourse the effect of noise on true jets is not) and that the
higher bump in f90 are actually true jets for the most part. Now data is
ofcourse somewhat more complicated but I believe the qualitative
conclusions should still hold.

Thanks
Vishnu