K charged WG activity report (part I)

KGM 14/11/03 V.Patera

K charged WG activity report (part I)

K : first absolute BR evaluation

K0 decay: selection efficiency study

Ratio k/K0: analisys refinement

Kl4, Ke3 ; analysis on going

K decay vertex eff. under way

Background on K tag: first look

K tracking efficiency measurement

DE/DX : test of new hardware setup

P.Branchini, E. De Lucia, P. De Simone, E.Gorini, M.Martemianov, L.Passalacqua, M.Primavera, B.Sciascia, A.Ventura, R.Versaci, V.Patera

Next talk

This talk


dE/dx upgrade

• The ADC system has been completely revised and upgraded

• ADC integration time stretched from 1.8 to 3.1 s to collect most of the released charged in the cell

• Unefficiency in collecting charge released at short distance from the wire fixed

• Data with cosmic and magnetic field finally collected last week (!!)

• Analisys in progress: preliminary results on efficiency and resolution


dE/dx : ADC signal

RED = 2.2 s (short)BLUE = 3.1 s (long)BLACK = 1.8 s (old)

(adc count/cm)

(adc count/cm)

New dE/dx distribution from cosmic taken with two different ADC gate value (2.2 and 3.1 s) compared with 2002 cosmic data (1.8 s)


dE/dx : efficiency

The longer gate improved the ADC sampling efficiency by more then 30%

coscos

NA

DC

/Nhi

t

NOW

OLD


dE/dx : average energy loss

NEW OLD

The comparison between old 2002 cosmic and the newly collected cosmic shows a substantial agreement of the B.B. energy loss


dE/dx : resolution

The resolution with respect the number of samples in the truncated mean is improved by 30% @ 10 samples and by 25% @ 20 samples ( Nsampletrmean = 0.82 NsampleADC). At Nsample >20 the resolutionis almost flatten noise ?

OLDNOW

Prototype resolution

Nsampletrmean Nsampletrmean

Res

olut

ion


dE/dx : noise?

To test the noise hypothesys we use the pedestal runs. We sum the charge

over a road trough 58 planes that fakes a track. The RMS of this sum is

compared to the incoherent sum of the i for each ADC of the road. We find a

very little difference….

This amount of noise cannot be responsible for the resolution behaviour.(Maybe the noise analysis should not be done on ped runs ??)


dE/dx upgrade summary

The efficiency loss of the ADC sample has been fixed The resolution with respect to the T.M. sample is now

better by 30 % for track with 20-30 hits For track with a given hit number the gain is twofold: more

ADC samples and better resolution with respect of the T.M. sampling

Yet the prototype resolution has not been achieved There is a flat behaviour in the resolution at high sample

value The analysis is in progress


K tracking efficiency & geometrical acceptance : K(pK,K)

We use the tag in the handle emisphere to have in the signal

emisphere a “pure” beam of K+(K-)

The signal is flagged as Kaon with standard cut on momentum and

IP distance

Background to the signal is mainly due to early 3 body decay of the

K, where a low momentum doughter mimic a K coming from the IP

We use the minimum distance between the candidate K track and

the extrapolated K track from the handle as check parameter

The shape of the r distribution for background is taken from MC

“ “ for signal is taken from MC and

from double tagged event


K(pK,K) signal vs background

K definition cuts :1) q opposite to the “handle”2) 70 < PK < 130 MeV3) Rpca < 10 cm4) -20 < zpca < 20 cm

Once found a “candidate” K we compute the distance of closest approch between the first hit of its track and the track extrapolated from the handle:

Handle: K+ track with

tagged decay

K- extrapolated

K- candidate

We monitor the background contamination of the signal looking at the tracks minimum distance r computed at the point of closest aproach.


K track eff. = fit to r

r (cm)

BLUE K from MCRED K from 2 tagGREEN bck from MC

The fit to the r distribution between the candidate track and the extrapolated track is made using MC or 2 tag shape for the signal and MC for the background shape

K- extrapolated

K- with tag decay

K+ with tag decay

Signal shape from DATA:


K shape uncertainties

The r distribution in the K region is slightly overestimated by the fit with K shape from MC and underestimated by the fit with the K shape from 2 tag. The differences between the 2 fits gives the sistematic on the K shape

Fit – signal : MC shape

Fit – signal : 2tag shape

r (cm)

r (cm)r (cm)

signal


K- versus time and shape

systematic

We check the stability of K versus time. The 2001-2002 data were divided in chunk of 6 pb-1 each. The two different results account for the 2 different shape choice for the K contribution.

K-

IntLum/6 (pb-1)

2001

2002

Handle : K+

Signal : K-

BLUE = MC shapeRED = 2Tag shape

Systematic due to the shape uncertainty 2x10-3


K-

IntLum/5 (pb-1)

K sistematic : tag bias

Systematic on the K tracking eff. can also be due to what happen in the opposite emisphere. Thus we measured the tracking efficiency with respect to the kind of tagged decay of the other K ( the “handle”)

BLUE: K0

RED : KBLACK: all tag

The maximum bias found on K was between opposite tag : (7.5 ± 2.0) x 10-4


K+ with respect K and pK

K-

bin

Pbin

We divide the K in 6 bin in the range 30< K <90 and the K momentum in 6 bin in the range 70< pK<130 (Mev/c)

bin = 10 degPbin = 10 MeV/c

Nev

ents

bin

Pbin

pK 100 MeV/c 900


K: average 2001-2002 values

We show the value of the for both charge, relative to the full 2001 2002 data set, based on the dst production version 16 and 15, with the corresponding evaluated systematic errors

2001-2002 K(%) stat(%) shape (%) tag (%)

K+ 41.84 0.04 0.15 0.18

K- 40.77 0.04 0.15 0.17

MC : K+ = 46.98 % truth = 46.97 % MC : K

+ = 46.27 % truth = 46.27 %

Still missing a quantitative evaluation of the Tag background… seems to be negligible even at few per mill level, but we are working on it.


K+ vs K

- The nuclear interactions of K- in the beam pipe and in the DC wall reduce K

- in comparison to K

+ by more than 1 %

IntLum/6pb-1

K

BLUE = K+

RED = K-

IntLum/6pb-1

K

K

The K integrated over and pK is (10.4 ± 4) x10-3


K- nuclear interactions from K (??)

PnuclK C1 + C2/sin

1

22

1

If we assume that K is totally due to K- nuclear interactions, then it should contain a costant term due to BP, plus a bigger term proportional to 1/sin due to DC wall (and BP)

Work in progress…it is just a first look

K

cos


K summary

The K tracking efficiency times the geometrical acceptance K has been measured using the tag

tecnique at fraction of % level

The K has been measured independently for positive

and negative K

The sistematics due to the uncertainty on shape of the signal and due to tag bias have been evaluated

The K has been measured versus the time in step of

6pb-1

A memo is in preparation


First look at tag background evaluation

The use of the K+(K-) tag decay ( K and K0) allow us to select a pure K-(K+) beam. Eventual pollution of the tag reflects in a systematic underestimation of the absolute BR measured on the other emisphere. We made a first attempt to estimated this background using a sample of 4 pb-1 of 2002 data

We assumed that the background fraction in the events with one tag decay is small. There is no background in the events where both K+ and K- undergo a tag decay (double tagged events) We compare the single and double tag kinematic distribution: the differences can be due to the background ( and , to some extent, to slightly different acceptance ) The statistical power of this analysis is limited by the rate of double tagged decay in K+K- events ( 10% of the total in the stream)


Tag bck: Kinematic variables

The control variables was chosen both in the lab and in the center of mass

frame:

1. Momentum of the K charged secondary in the K frame2. Angle between the K flight path and the charged secondary in the K

frame3. Angle between the charge secondary and the K in the lab4. Number of clusters associated at the K decay product ( ≤1 for K and

≤3 for K0)5. Energy of the cluster associated to the charged secondary6. Time of flight of the charged secondary

Only the shape can be compared due to the different yelds of single and double tag events


Charged secondary momentum in K frame

Red = difference of the 2 histoBlue = statistic uncertainty

Normalized comparison between single and double tag events

Linearscale

Logscale

Mev/cMev/c

Mev/c


Cos() between K and secondary in K frame


Linearscale

Logscale


Cos() between K and secondary in lab frame

Linearscale

Logscale


Number of secondary cluster associated

Ncluster ≤1 for K

Ncluster ≤3 for K0


Linearscale

Logscale


Energy of the cluster associated to the charged secondary

Linearscale

Logscale

MeV MeV

MeV


Time of flight of the charged secondary

Linearscale

Logscale

ns

ns ns


What about dE/dx ?

dE/dx is a very powerful PID for K± but:

1) Was not available in 2001 data2) Due to low efficiency of ADC sampling in 2002 the DE/dx

measurement select the K with 900

KTruncated mean (count/cm)

KaonsNon Kaons

K with dE/dxK with nodE/dx


Background statistic estimator

To build a conservative background estimator I have to measure the deviation from statistic fluctuation of the difference of the two sets of histos. We define:

n) = abs [ his2tag(n) – his1tag(n) ]

For each bin I consider the quantity (n) = n) - (n) . This variable gives

the deviation of (n) from the statistical fluctuation and is > 0 if the bin is bigger then statistica fluctuation and < 0 is underfluctuate. The sum over all the bins of (n) is a upperlimit to the background.

n)2 = ( his1(n))2 + (his2(n))2

For bin n


First results on 3 pb-1 of 2002

Variable ii ii ii

Pseccm 0.0219 0.0160 0.0059

coscm 0.0258 0.0301 -0.0043

coslab 0.0249 0.0298 -0.0050

Ncluass 0.0070 0.0038 0.0032

Eass 0.0318 0.0745 -0.0427

Tofass 0.0285 0.0491 -0.0205

Positive Tag


First results on 3 pb-1 of 2002

Variable ii ii ii

Pseccm 0.0303 0.0161 0.0142

coscm 0.0280 0.0303 -0.0024

coslab 0.0267 0.0300 -0.0032

Ncluass 0.0085 0.0038 0.0047

Eass 0.0331 0.0715 -0.0384

Tofass 0.0304 0.0386 -0.0083

Negative Tag


Backgroung on negative tag?

The difference between the 1 tag and the 2 tag distribution settles on the signal region.. True background ???

Conclusion:

There is no evidence for a clear background contamination in the single tag events, at least at % level

We are working out a robust statistic estimator for the background level (or limit)

Work in progress..

K charged WG activity report (part I)

Documents

adc samples

adc sampling efficiency

adc signalred

resolution behaviour

noise analysis

better resolution

resolutionthe resolution

fixedthe resolution