Tracking Efficiency for $\tau$-ID

release: t01.45.00, p08

• Postscript version (somewhat nicer)

$\tau$ Samples:

single $\tau$
(8000 events 0 mb
6000 2.5 poisson events)

Z $\tau$ $\tau$, 2.5 poisson min bias
(5000 events)

$t \bar{t}$, 2.5 poisson min bias
(5000 events, about 900 $\tau$s)

Tracking efficiency: definition

look for charged $\tau$ daughters in event
for every charged daugher look for closest reconstructed track ( $\eta-\phi$ match)
track are only matched once (important for 3 prong tau decays)
if the match lies within a cone of 0.02 in $\eta-\phi$ space, the track is called ``found''

Check: look if 1./pT makes sense

1./pT - conclusions

For the Z $\rightarrow \tau \tau$ sample

if $\Delta \eta-\phi$ (track-prong) < 0.02 $\rightarrow$ 1./pT_prong-1./pT_track peaks at 0
if $\Delta r$ > 0.02 $\rightarrow$ no match in 1./pT

look at the MC particle closest of the track which is not the prong:
peak for the not-matched tracks $\rightarrow$ they do come from another particle
the number of tracks with a delta 1./pT of less 0.2
right-wrong distinction at $\rightarrow$<0.2 :
right: 95%, wrong: 7%
right-wrong distinction at $\rightarrow$<0.02 :
right: 98%, wrong: 10%

$\rightarrow$ Conclusion: matching using only $\rightarrow$ works well

Tracking efficiency as function of track pT
$\rightarrow$ no obvious dependence

Location of losses

non-efficiency (big box == 100 % lost)
red curves: indicate where tracks should have 8 CFT hits

conclusions:
1 prong is mainly lost in non-efficient tracking region
3 prongs are lost everywhere

Lost as other tracks are close?

only a slight difference for found and missed

sometimes tracks with other particles close by are found
where close == close in $\eta-\phi$
Event display for central events where all 3 prongs had a pT over 1.5 GeV
found only one

found all

lost all

Tracking efficiency in the different samples as function of min bias events

NB: for all samples in all min bias bins: no obvious pT dependence

Track finding efficiency:

prong 90 %
3 prong 72 %

per track!
independent of sample, of pT, of #min bias events...

probability to find 3 prong (with right 3 tracks) $\rightarrow$ about 40%

This can be seen in the plots below, esp in the lower left plot (matching with tracks).
Only a bit above 40% of all 3 prongs are found as 3 prongs
If all tracks in the event were found, this plot would like the plot on the right hand side.

Conclusions:
This is clearly a problem and will effect studies on track matching as well.
QCD rejection using the higher track multiplicity is affected, too.