Search for B D proton X, DELPHI experiment

1
Search for B gt D proton X,DELPHI experiment

• Unfolding the true proton content.
• F. Mandl, J. MacNaughton
• PDA Nov. 2002
• Reported by J.M.

2
Contents of this part
- Introduction
- Checking MC treatment of PARTID
- Method for unfolding

• Results
• Present status and future plans
• what we do try, and what we
• do not try

3
Introduction
2 main problems for this analysis 1. use
information about vertices to suppress
backgrounds talks of Franz Mandl at previous
PDAs. Method of discriminant variables. 2.
Using PARTID to identify protons and dealing
with efficiencies and misidentification involved
in that. This talk.
4

• Motivation
• There are 2 motivations for doing this.
• The DELPHI MC ( very good one) contains a lot
• of info concerning about how PARTID works.
• Should be able to use this to unfold real protons
  from
• identified protons.
• 2. Greatly reduces our dependence on Monte Carlo
• for estimating the background in our
  signal-like events the last step in our
  analysis, but an important one. PARTID aspect of
  MC that we
• can test

5
Checking the PARTID in MC
To do this, use sample enriched in protons (
lambdas) and kaons and pions (D/D
decay chain). 79, 74 and 88 purities. Use
decays lambda to proton pi and D to D pi, D to K
pi.
6
We use the combined tag ( Peter Kluit )
(recommendation of N. Neufeld ), tight cuts,
info. from RICH ( Newtag) and dE/dx in TPC. I.e.
info. from Cerenkov counters and energy loss in
TPC. Amount of impurities should also be
well reproduced by MC recommendation of E.
Schyns. Fig. Comparing data, MC as function of
momentum.
7
Procedure for unfolding
First step Use MC (impurities removed) to get
prob.s for pi, K proton to be identified as pi,
K, proton 3x3 matrix, momentum dependent. (the
C matrix).
Then, ident. C x real or true Where
real or true and ident. are the vectors pi,
K, proton. Final step true C-1 x
ident. where ident. is measured, and we
have determined C from MC.
8
But, momentum dependence of the C matrix. Dealt
with this in 2 ways first way is to set up the C
matrix for each event, use the C matrix
corresponding to the momentum of that event, and
sum at the end. Second way, for each sample
average the C matrices over all events in that
sample Turns out to be a little different for
signal and M.C., but not very much. Because
momentum distributions of both samples are more
or less the same fortuneately
9
An advance comment on results Can be given in
either true proton space or event space and
then transform to upper limit ( or BR) by using
the appropriate efficiency. (We have both
eff.s) Calculation result comes ( this
formulation) in true proton space unfolding
takes into account both misid. and eff. of
PARTID. But, statistical errors natural feel
for that in event space so we chose to
transform first to event space, ( mult. by ratio
of eff. with, without PARTID)
10
Second advance comment on results The
calculation of stat. errors, taking into account
the stat. errors on C matrix elements. One would
think that this is easy, but in fact it is very
much non-standard and not trivial. Fortuneately
real experts on this subject in Vienna who
helped. Quite long calculation even after their
help. Ans Contributions of errors on C matrix
turned out to be negligible, but we did the
calculation to see that
11
Results After all cuts, 15 events in real data,
16 in M.C. - for protons ( one of the input to
this calculation). Want to remove the non-real
proton comp. by unfolding, then subtract out real
protons using M.C. 2 ways to get latter do also
by unfolding. Momentum dependent real data (
event space)- Real protons 8.25 /- 4.34 , M.C.
2.33 / 2.38 Same for Mom. average method Real
protons 9.61 /- 4.19, M.C. 1.90 /- 1.96 Doing
the arithmetic and averaging the 2 methods 7.71
/- 4.63 events, residual signal. Did 2
consistency checks M.C.- unfolding
results, both true proton component of M.C. and
total OK.
12
Finally, D proton, D antiproton
separately. Mostly D proton if B baryon
dominates, more of opposite if B meson
dominates. Answer D proton c.c. 3.16/-
2.68 residual events D antiproton c.c. 3.66
/- 3.86 residual events Good idea, but errors
too large to be conclusive. Does not look like B
baryon dominance.
13
Results - continued
Efficiencies Including PARTID eff. 2.6 All
except PARTID eff. 7.9 BR (B gt D p ( p-bar) X
(1.88 /- 1.30 /- 0.34) x 10-2 Upper limit
(and it is an upper limit) lt 4.5 x 10-2
14
Status future
My suggestion as to what we will not do extend
from data of 1994 run to 1995 run, add more D
decay channels, optimization, also add D not
coming from D - because we are now too busy with
BELLE.
Current status a few weeks ago we brought out (
released) a DELPHI note (finally!).
15
What we will try to do bring out a conference
report based on what is contained in the DELPHI
note. But without large investment in time.