Minutes of the NFS-meeting held on 9/8/94:
==========================================
 
In this meeting we had the following reports:
 
* Olaf Behnke   : K0b/K0 Normalisation study continued
* Cyril Hugonie : Systematics of the time resolution
 
 
K0b/K0 Normalisation study continued: (O.Behnke)
=====================================
 
* study effects for M1 and M2
  -> see bigger effects (which shall cancel when adding M1 + M2 !)
* split data into K0 momentum bins and make
   separate fits of the asymmetry and rates
   -> different normalisation respected
   -> results for phi00 consistent???
 
1. Effects for M1 and M2
For M1 and M2 it was checked how the K0B/K0 ratio
depends on:
neutral kaon kinematic
- PK0
- PK0Z/PK0
charged kaon kinematic
- PK+-
- PK+-Z/PK+-
charged pion kinematic
- PPI+-
- PPI+-Z/PPI+-
 
The data sample used for this study is our "normal"
data sample after the standard selection and cuts
described in note nfs number 8.
Only events with 6c-fit reconstructed lifetime between
0 and 4 ts are selected.
 
The plots show
that for separate M1 and M2 one gets almost always
some slope effects in the distribution of the K0b/K0 ratio
vs. kinematical quantity (e.g. PK+-). The slope effects
which are opposite for M1 and M2 are in the order of magnitude
of 10% to 40% (which means that the ratio changes like this
from start to end point of the kinematical quantity).
It is foreseen to expand this study for further
kinematical quantities like the distance of the
nearest shower to the charged Kaon track (->effects of fake shower
which could have different normalisation????).
One open question is, if one should correct the
data always to have the same statistics in M1 and in M2
in order for cancellation of slope effects.
At the moment the difference of M1 to M2 statistics
in our sample is about 3%.
 
2. Attempt to make separate asymmetries and fits:
The data sample was splitted into 7 regions of
K0 momentum of 50 MeV width from 400 Mev to 750 MeV.
The problems for fitting the rates and asymmetries
are:
- low statistics in some bins
- for each momentum bin one has to supply from MC (or fit to A./R.)
  * lifetime resolution
  * efficiency
  * background (shape)
 
One has to be careful not to enlarge the systematical
errors too much by making the splitting.
The present MC statistics is not high enough
for making any splitting easy in terms of statistics.
(we need more!!!)
 
The asymmetries and fitted curves are shown in plots.
In the three momentum bins 550-600, 600-650, 650-700
an asymmetry is clearly visible.
The results for fitted phi, alpha, background fraction
are plotted vs pk0 momentum.
For higher pk0 momentum the results
for phi00 are decreasing. Within the present systematic
the obtained result is nevertheless consistent with flat
phi00 vs pk0.
The obtained normalisations as function of pk0 agree
very well with what has been seen from the k0b/k0 ratio
vs pk0 plots.
The question now is how far one should really go into
splitting the data into groups (for instance in bins
of the K0-dipangle).
*
 
 
Systematics of the time resolution: (C.Hugonie)
===================================
 
BASIC IDEAS
 
   We took the resolution from the MC - applying the normal event selections
and cuts - as the true one so as to simulate reference rates and asymmetry,
with low statistic fluctuations (1e12 events). Then, we applyed different
tricks on the original resolution and fitted the CP parameters taking the
rates and the asymmetry calculated with this false resolution in comparison
with the previous ones to see the systematic shifts (especially on the phase).
   The two questions we would like to answer are mainly the following:
- How good has one to know the resolution in order to have minimum systematic
shifts on the phase ?
- Is it possible to find a nice parametrization for the resolution one could
eventually fit directly from the data ?
 
TAILS EFFECT
 
   We first tryed to cut symmetrically the tails of the resolution at different
times. The fitting were done on the asymmetry, between 0 and 20 TauS. For
eta00 fixed, the phase oscillates between 45.6 and 50.4 degrees (maximum for
a cut length = 10). For eta00 free, we get some very bad results for sharp
cuts (up to 75 deg for phi00 !).
 
WIDTH EFFECT
 
   We then multiply the width of the resolution by a scaling factor (0.75 -->
1.25). For a too thin resolution, the asymmetry is less dampened so that eta00
decreases. On the contrary, phi00 slightly increases. The effects are reversed
for a too broad resolution.
 
PARAMETRIZATIONS
 
   Two method were employed:
   1- We fitted directly the resolution with a chosen function. Then we fixed
the obtained parameters and fitted the phase taking this function as the false
resolution. This to know how good are such functions to describe the resolution.
   2- We fitted the resolution function paramters and the CP parameters all
together. This is what we should do to the real data if one of our functions
occured to be good.
   We tryed first a lorentzian, but the results were tragic, mainly because of
the too big tails. We then took 1 and 2 gaussian. The results are not so bad
when eta00 is fixed (unless correlation problems give crazy results) especially
with the first method. There is almost no difference between 1 and 2 gaussians.
Finally we tryed a non standard function:
             TR(t) = a1*exp((t-a2/a3)**2/(1+a4*|t|**a5))
By this trick we added tails on a usual gaussian. The Results are very good for
the phase (shifts within +-0.5 deg). We managed also to fit eta00 with such a
parametrization.
 
CONCLUSION
 
   What appeared during this work is that we cannot expect to fit eta00 -except
with our last function. At the opposite, the phase seems to behave nicely .Yet
we should not draw any definite conclusion since all our fits had unreallistic
boundaries. We will therefore redo some of our fits with better boundaries.
We will try also to investigate on the Fourier transform of the rates, which
could allow us to get rid of the time resolution in the rebuild asymmetry and
to study the efficiency effect.
 
*
 
Next meeting:
=============
 
 ***********************************************************************
* We agreed to have the next NFS-meeting on Tuesday 23/8/94 at 10.30 h  *
* in the CPLEAR meeting room.                                  =======  *
 ***********************************************************************