Results of Lambda Reconstruction in Selex
1.1 The ability to reconstruct lambda decays and the efficiency
estimation were tested with Selex Embed program, currently used
as Selex Monte Carlo program (see Embed in Selex at Work).
For the decay in in vertex spectrometer I generated flat decay
length distribution in the range 0.5 to 20 cm in order
to obtain the efficiency versus lambda momentum as shown
in fig.1 .
For the decay in M1 spectrometer I used fixed momentum of 90 GeV/c
to obtain the efficiency versus decay coordinate Z fig.2 .
Both distributions were calculated using standard tracking program
'all.tseg' and default V0 setup. One should consider the outcome of
the Embed analysis as very rough estimation because the accurate
parameters depend on particular physics you intend to investigate.
Nevertheless, you might hope to have the V0 efficiency not less than
60% in vertex and not less then 25% in M1 spectrometer.
2.0 Lambda reconstruction in real life was done for 1M sample (filtered
interactions, charm_run_10783, 1M of the events). The decays can be
classify using properties of proton and pion tracks:
- track origin in Vertex (vx) or M1 (m1) spectrometer;
- measured pion momentum (2p) or unmeasured one (1p);
The proton momentum supposes to be measured always.
Two modes of reconstructions were examined: default V0 mode and soft
cut mode (set on v0 soft). See V0 description for more details.
2.1 The mass distributions for lambda and anti-lambda selected in soft
cut mode in Vertex spectrometer are shown in fig.3a and in fig.3b
respectively.
one million filtered interactions
---------------------------------
number of events mass mean value GeV/cc rms
lambda 2798 +- 55 1.116 +- 0.0003 0.0014
anti-lambda 167 +- 14 1.116 +- 0.0009 0.0012
The whole set of histograms is presented in fig.4 .
I see two interesting features of lambdas in Vertex.
First, the miss distance of the proton with primary vertex is
small (<200 mkm), so lambda itself (proton) could be responsible
for a charm hit in the filter.
Second, the number of lambdas with unmeasured pion momentum is very
low (<5%). We do not have tracking tools to catch not primary tracks
without guidance from m2 or m1 spectrometer.
We could do this for example, using Peter's space points algorithm.
2.1 The mass distributions for lambdas selected in M1 spectrometer with
2 measured momenta with default V0 setup and soft cut setup are
given in fig.5 and in fig.6 respectively.
Both distributions were fitted with 3 Gaussian curves and simple
threshold function. We need several Gaussian because of the nature
of M1 spectrometer: Silicon, proportional chambers, drift chambers
with different space resolution (50 mkm - 3 mm).
lambda yield (2p) from one million filtered interactions
--------------------------------------------------------
default cut events total: 19636
number of events mass mean value GeV/cc rms
Gaussian 1 8636 +- 344 1.115 +- 0.08e-3 0.0072
Gaussian 2 1385 +- 183 1.115 +- 0.08e-3 0.0018
Gaussian 3 6577 +- 1137 1.115 +- 0.08e-3 0.1209
soft cut events total: 10952
number of events mass mean value GeV/cc rms
Gaussian 1 6891 +- 304 1.115 +- 0.07e-3 0.0076
Gaussian 2 776 +- 134 1.115 +- 0.07e-3 0.0021
Gaussian 3 4239 +- 1164 1.115 +- 0.07e-3 0.1197
First look, we see about 10000 lambda decays in M1 spectrometer when
the momenta of proton and pion were measured (7650 in soft setup).
The Gaussian 3 seems to be a background because of huge rms, but
this must be checked.
In addition we have about 5000 lambda candidates when pion momentum
was not measured (see complete histogram set for soft cut in fig.7 ).
The decay range in Vertex is about 10 cm, where we found about 3000
lambdas. The decay range in M1 spectrometer is about 600 cm and the
efficiency is 3 times less (say, 4 due to exponent in decay length).
We lost some more events with 2 unmeasured momenta in m1 than in vx
because M1 and M2 magnets are different.
All together, we could expect roughly about 45000 decays,
and we have 3 times less.
If we say that a half of charm filter decisions were done
due to protons from lambda decay in vertex, then we have
the agreement in numbers (for filter rate 1:7). This hypothesis
can be checked in the analysis of not filtered interactions.
Finally, for filtered interaction we have
3000 lambdas in Vertex
10000 lambdas in M1 with 2p
5000 lambdas in M1 with 1p
total 18000 lambdas per 1 Million interactions.
vmatveev@fnal.gov