─юъєьхэЄ тч Є шч ъ¤°р яюшёъютющ ьр°шэ√. └фЁхё юЁшушэры№эюую фюъєьхэЄр : http://zebu.uoregon.edu/~uochep/talks/talks04/cpotter12-04.pdf
─рЄр шчьхэхэш : Tue Dec 21 19:58:30 2004
─рЄр шэфхъёшЁютрэш : Tue Oct 2 11:50:30 2012
╩юфшЁютър:
╧юшёъют√х ёыютр: m 13

With Hadronic Tags

б

The PEP-II/BaBar B-Factory

Run: 2397850

Date Taken: Wed Jan

б

1 02:02:03.799610000 1997 PST?

HER: 8.990 GeV, LER: 3.112 GeV

Chris Potter and David Strom University of Oregon

в

Collaboration Meeting, 7 December 2004 н p.1/25

Analysis Overview
Event selection: A tag meson decaying in a hadronic mode is fully reconstructed using the semiexclusive (where ). We use reconstruction: RecoilAnalysis/baseClass.cc for the interface to the semiexclusive ntuples.
же д йи з б ╕ б в !

Events with two remaining tracks identify the decays to single track tau modes: or . This accounts for 51% of all tau pair modes.
' #( #

в

%$ #

& "

Event remainder kaon, electron, muon and neutral pion multiplicity consistent with a tau pair decay are required. Kinematic correlations in momenta and remaining unassigned neutral energy are exploited by a neural network analysis. Changes since the last presentation (October AWG Meeting): Incorporated 1M SP5 and SP6 signal generic Monte Carlo events. Included a new parameter (the cosine of the angle between the tag which dramatically reduces the background. and the rest of the event)

Evaluated our systematic errors, in particular for the signal decay model. Extablished the limit setting procedure and optimized the cut selection for best upper limit. Released BAD 542 version 4. Please provide feedback!

Collaboration Meeting, 7 December 2004 н p.2/25

%$ #

Data and Simulation Samples
й
б

Data Sample Run 1 Run 2 Run 3 Run 4 Runs 1-4

и

иж

(

е

ж

ие

з

б

Events

вб

б╕

д

╕

!

!

!

" !
!

!

!
!

!

$ (' &% $ ) 1 " 1 $ #

We analyzed Runs 1 through 4 data (startup to 2004 summer shutdown) in the BSemiExcl skim, fb of integrated luminosity. The simulation corrections recommended by the which yielded tracking, PID and neutrals groups. We employ the PID killing tables obtained from Release-14 control samples, provided by the particle identification working group in $BFROOT/physicstools/pid/tables. We smear the photon energy resolution obtained in the EMC by resampling the energy from , and then scale the resampled energy up by the a Gaussian with mean and width using factors prescribed by the neutral working group. factor
'%

The tracking efficiency task force recommends an averaged flat efficiency reduction for tracking efficiency systematic error reconstruction of loose tracks, which corresponds to per GoodTracksLoose.

0

й

Collaboration Meeting, 7 December 2004 н p.3/25

Simulation Sample
е ж ие з

Simulation Sample B0->tau+tau- +CC (FSR) (Run1) B0->tau+tau- +CC (FSR) (Run2) B0->tau+tau- +CC (FSR) (Run3) B0->tau+tau- +CC (FSR) (Run4) B0B0bar generic BpBm generic ccbar generic uds generic B0B0bar generic BpBm generic ccbar generic uds generic Signal Cocktail Background Cocktail

SP Cycle SP5 SP5 SP5 SP6 SP5 SP5 SP5 SP5 SP6 SP6 SP6 SP6 SP4 SP4
!
!

иж

й

Events

#

и

!

#
!

и

"

#

и

!

#
!

и

!

!
!

!

"

" ! !
!

! !

"
!

" й
б
в

и

" "
╕

и

!
б

Collaboration Meeting, 7 December 2004 н p.4/25

б

б

б

б

Tag
Particle
б

Reconstruction
SemiExclUser Reconstructed Decay
д ╕

з ╕

и

╕

б

б

и

б

в

в

б

и╕

в

и $
в

и

и

и

и

и

и

'

$

$

$

б

в

в

б

в

╕

и

и

и

в

$

и

$

$

б

б

б

б

$

'

и

б

и

б

$

$

$

$

$

$

$

$

$

$

$

б

б

б

б

б

б

$

б

$

$

$

$

$

$

б

б

б

$

%

%

%

$

$

$

$

$

$

$

The reconstruction technique employs the SemiExclApp package. It proceeds by and then adds additional pions and kaons until the system is in a reconstructing a seed signal region.
же д
д

╕

$

$

╕

з

From among the many reconstructed candidates in a single event, the candidate whose mode is reconstructed with the highest integrated purity is selected as the tag . If the . candidate is a charged the event is rejected. We reject modes with purity less than

б

$

б

б

The semiexclusive n-tuples and the purities have been generated by the package IslBrecoilUserApp (which employs SemiExclApp) in the IslBrecoilUser package by others in analysis-21. Collaboration Meeting, 7 December 2004 н p.5/25

1

б

Fitting Procedure
The combinatorial and continuum background are subtracted by fitting the distribution. GeV are considered Peaking events lying in the signal region defined by tag signal.

╕

б

!

╕

$

The fits are maximum likelihood fits performed with the ROOT v3.10/01 interface to the MINUIT v94.1 package.

Two probability density functions are used in constructing the overall fitting function, a single Argus function and a Crystal Ball function:

╕

з

д

з

д

е

д

е

╕

(1)

ж

ж

е

и

й

0

0

й

$

0

з

%

й

д д з
е

в

з

& д

$

и

з "!
б

е

й

й

й

д

б

(2)

е

$

з

$б

$

и

д

з

е

&

%

д

й

з
б

б

е

#

й

й

д

The Argus function models the continuum and combinatorial background. In all is fixed at half the nominal center-of-momentum beam energy cases the Argus cutoff GeV). The Crystal Ball function models the signal peak with ( a Gaussian above a cutoff defined by and a tail with a power for modelling radiative losses from neutral pion decays.

╕

ж

е

0

$

0

з

д

б

е

$

и

$б

0

$

Collaboration Meeting, 7 December 2004 н p.6/25

з

й

б

The
Data Runs 1-4 & SP5/6 MC Before Preselection Data Mean: 1.9452 B0B0bar Mean: 1.9510 Data 25000 RMS: 0.0707 B0B0bar RMS: 0.0697

б

в

Mass and Tag
18000 16000
Data Runs 1-4 & SP5/6 MC Before Preselection Data Mean: -0.0064 B0B0bar Mean: -0.0057 Data RMS: 0.0298 B0B0bar RMS: 0.0276

20000 14000 12000 15000 10000 10000 8000 6000 5000 4000 2000 0 1.85 1.9 1.95 2 2.05 0 -0.1 -0.05 0 0.05 0.1

m D(*)
╕

E

rec
д

At left, the reconstructed mass of the seed or used in reconstructing the tag . At right, the of the tag . Only the peaking components of data (dots), generic neutral (solid) and the signal cocktail (dashed) are plotted.

╕е

Collaboration Meeting, 7 December 2004 н p.7/25

$

Tag
8000 7000 6000 5000 4000 3000 2000 1000 0 5.2

Yield
Data Runs 1-4 Tag B Yield Signal Region: 1193078 Four Argus Fit: 868996 9000 Data Peak: 324082

Runs 1-4 & SP5/6 MC Tag B Yield Data Signal Region: 1193078 9000 MC Signal Region: 895803 Data Peak: 297275

8000 7000 6000 5000 4000 3000 2000 1000

5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29 5.3 mES [GeV]

0 5.2

5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29 5.3 mES [GeV]

We added the combinatorial (nonpeaking) component of the neutral distribution with the continuum and charged distributions. The normalization has been selected to minimize the GeV. mean square error in the region

We also fitted the background distribution with the sum of four Argus functions in a region far below the peak. The four Argus shape parameters are free, but the cutoff is fixed at GeV. The fit is performed only in the region parameter , well below the peak tail.

вб ╕

д из ж е и й из ж йж

вб ╕

д из ж е и й б из ж й

вб ╕

из ж

й

The discrepancy is approximately

, and is used as the systematic error on the tag

Collaboration Meeting, 7 December 2004 н p.8/25

yield.

Event Preselection
90000 80000 35000 70000 30000 60000 25000 50000 40000 30000 20000 10000 0 0 2 4 6 8 10 12 N 20000 15000 10000 5000 0
GTVL Data Runs 1-4 & SP5/6 MC Before Preselection Data Mean: 4.6493 B0B0bar Mean: 4.5394 Data RMS: 1.9315 B0B0bar RMS: 1.9123

40000

Data Runs 1-4 & SP5/6 MC Before Preselection Data Mean: 7.0391 B0B0bar Mean: 7.5444 Data RMS: 3.3544 B0B0bar RMS: 3.5277

0

5

10

15

N

20
GPL

At left, the very loose track muliplicity on the signal side. At right, the loose photon muliplicity on the signal side. The distributions in data and Monte Carlo simulation do not match. Therefore two requirements are imposed on the samples before beginning this analysis: two very loose track candidates reconstructed opposite the tag four or fewer loose photons reconstructed opposite the tag

After imposing these requirements, the Monte Carlo simulation and data distributions came into better agreement. Collaboration Meeting, 7 December 2004 н p.9/25

Distributions after Preselection
Data Runs 1-4 Preselection Signal Region: 15930 Argus: 6436 1800 Crystal Ball: 9494

B0B0bar Generic Preselection Signal Region: 28884 Argus: 3313 3000 Crystal Ball: 25571

1600 2500 1400 2000 1500 800 1000 500 200 0 5.2
0

1200 1000

600 400

5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29 5.3 mES [GeV]
0

0 5.2

5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29 5.3 mES [GeV]

B + - Cocktail Preselection Signal Region: 19807 2500 Argus: 148 Crystal Ball: 19659

B + - Generic Preselection Signal Region: 451 Argus: 44 Crystal Ball: 407

100 2000 80 1500 60 1000 40 500

20

0 5.2

5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29 5.3 mES [GeV]

0 5.2

Collaboration Meeting, 7 December 2004 н p.10/25

5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29 5.3 mES [GeV]

Background Rejection
7000 6000 5000 4000 3000 3000 2000 2000 1000 0 1000
Data Runs 1-4 & SP5/6 MC Data Mean: 0.0002 B0B0bar Mean: -0.0135 Data RMS: 0.7983 B0B0bar RMS: 0.7827

6000

Data Runs 1-4 & SP5/6 MC Data Mean: 0.1652 B0B0bar Mean: 0.1970 Data RMS: 0.3913 B0B0bar RMS: 0.4219

5000

4000

-3

-2

-1

0

1

2

3 Q

0 -0.5
GTL

0

0.5

1

1.5

2

2.5

3

N

3.5
KML

5000

Data Runs 1-4 & SP5/6 MC Data Mean: 0.1788 B0B0bar Mean: 0.1968 Data RMS: 0.4523 B0B0bar RMS: 0.4793

3500 3000

Data Runs 1-4 & SP5/6 MC Data Mean: 0.4001 B0B0bar Mean: 0.3242 Data RMS: 0.6392 B0B0bar RMS: 0.5643

4000 2500 3000 2000 1500 2000 1000 1000 500 0 -0.5 0 -0.5

0

0.5

1

1.5

2

2.5

3

N

3.5
KsD

0

0.5

1

1.5

2

2.5

3

N

3.5
KlET

Clockwise from top left, the net loose track charge on the signal side, loose kaon, default and tight multiplicity on the signal side after all preceding requirements in the analysis chain have been imposed.
б

Collaboration Meeting, 7 December 2004 н p.11/25

Signal Selection:
Data Runs 1-4 & SP5/6 MC Data Mean: 0.4439 B0B0bar Mean: 0.4700 Data RMS: 0.7166 B0B0bar RMS: 0.6897

and

бв

де

╕

600 500

Data Runs 1-4 & SP5/6 MC Data Mean: 0.1824 B0B0bar Mean: 0.2024 Data RMS: 0.2732 B0B0bar RMS: 0.2899

50

40 400 30 300 20 200 100 0 -1 10

-0.8

-0.6

-0.4

-0.2

-0

0.2

0.4

0.6

0.8 1 cos(B,ROE)

0

0

0.5

1

1.5 E GPL -E

pDM

2 [GeV]

On the left, the cosine of the angle between the reconstructed tag and the rest of the event. On the right, remaining neutral energy after neutral pion reconstruction. Preceding requirements in the analysis chain have been imposed. The angle between the tag momentum and the rest of the event momentum (all tracks and photons) provides a discriminant between signal and background. Signal side momentum in the signal samples is nearly isotropically distributed.

The existence of any neutral energy which is not reconstructed in a neutral pion from the signal tau pair or in the reconstructed tag is a strong indication that unreconstructed neutral pions Collaboration Meeting, 7 December 2004 н p.12/25 are present.

ж

Signal Selection: Decay Mode
Selection Mode 0: 1: 2: 3: 4:
' ' '

BR [?]

Efficiency
!

Composition
!

вб

в╕

ед

╕

д

ж

#

"!

д

д

na na
б в

з

╕

!

#

'

GeV GeV GeV

!

!

!

"!

!

&

!

#

$

#

б

na

!

!

&

#

$

б

5:

!

!

&

#

$

Branching ratio, signal mode requirements, reconstruction efficiencies and sample purity with respect to all tau pair decay modes in the signal cocktail with a tag reconstructed with the exclusive method. The reconstruction is verified with Monte Carlo truth. In order to best exploit correlations in selection mode, candidate tau daughter momenta and remaining neutral energy, five parameters were used as inputs in a neural network analysis.

Collaboration Meeting, 7 December 2004 н p.13/25

в

в

Signal Selection: Neural Network
B0B0bar Generic mode>-1 Signal Region: 334 45 Argus: 17 Crystal Ball: 317 40

14 12 10

Data Runs 1-4 & SP5/6 MC Data Mean: 0.0000 B0B0bar Mean: 0.4933 Data RMS: 0.0000 B0B0bar RMS: 0.2054 - - - EvtGen ... Tauola

35 30 25

8

20
6

15 10 5 0 5.2 5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29 5.3 mES [GeV]

в

4 2 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9 1 NN Out

The selection for . At left, the distribution in Monte Carlo simulation. At right, the neural network output in Monte Carlo simulation. The analysis is still blind in data.
"и в "
╕

The neural network was implemented on the Stuttgart Neural Network Simulator (SNNSv4.2) topology. We separated each of the signal and background cocktail Monte in a Carlo simulation into two disjoint subsamples, one for purposes of training the neural network and the other for evaluating the neural network performance.

й

The truth tagged sample provided the neural network training patterns and the BRecoUser tagged sample provided the evaluation patterns. We stopped the neural network training after Collaboration the minimum root mean square error (RMSE) obtained. Meeting, 7 December 2004 н p.14/25

й

Signal Selection: NN Inputs
40 35 30 25 20 15 10 5 0 0 1 2 3 4 5 mode
Data Runs 1-4 & SP5/6 MC Data Mean: 0.0000 B0B0bar Mean: 1.2385 Data RMS: 0.0000 B0B0bar RMS: 0.5753 - - - EvtGen ... Tauola Data Runs 1-4 & SP5/6 MC Data Mean: 0.0000 B0B0bar Mean: 1.9057 Data RMS: 0.0000 B0B0bar RMS: 1.3918

22 20 18 16 14 12 10 8 6 4 2 0 -1 -0.5 0 0.5

Data Runs 1-4 & SP5/6 MC Data Mean: 0.0000 B0B0bar Mean: -0.4880 Data RMS: 0.0000 B0B0bar RMS: 0.5279 - - - EvtGen ... Tauola

1 cos( +, -)
Data Runs 1-4 & SP5/6 MC Data Mean: 0.0000 B0B0bar Mean: 0.1097 Data RMS: 0.0000 B0B0bar RMS: 0.0766

12

18 16 14 12 10

10

8

6 8 4 6 4 2 2 0 0 0 p
+

0.5

1

1.5

2

2.5

0

0.05

0.1

0.15

0.2 E

GPL

0.25 0.3 -EpDM [GeV]

Inputs to the neural network for the selection. Clockwise from top left are the signal multiplicity mode, cosine subtended by tau daughter candidate momenta, remaining neutral energy and magnitude of the tau daughter candidate momentum. The analysis is still blind in data.
"и в "

в

Collaboration Meeting, 7 December 2004 н p.15/25

в
$
!

й
%

"!#

$ '

д
╕

й

е й и д
$

же з
и и и

$ и з

$

д
и и

же в д ╕в б

Requirement

Preselection

й

з

!

GeV
! ! ! !
!

!

!

!

!

"

!

!

Peak Yield

"
!

!
!

! ! !

B0B0bar Monte Carlo

! !

Efficiency

!

!

blind blind blind blind blind

!

!

"!

!

Efficiencies: Data and B0B0bar

Peak Yield

!

Data Runs 1-4

"

!

!

Efficiency

blind

"

!

!

Collaboration Meeting, 7 December 2004 н p.16/25

в
$
!

й
%

!#

$ '

д
╕

й

е й и д
$

е з
и и и

$ и з

$

д
и и

же в д вб

Tag B Yield

Requirement

Preselection

й

з

!

GeV
" !
!

! !

!

!

!

!
!

"

!

Efficiencies: Signal

"
! !

"

!

!

Peak Yield

!

!

"
!

!

" ! "

!

Signal Cocktail

Efficiency

"

" "

"

!
!

! !
!

!

"

"

"

Peak Yield

! !
!

"

!

!

!

Signal Generic

!

!

! !

!

Efficiency

Collaboration Meeting, 7 December 2004 н p.17/25

Control Sample: NN Output
entries/bin
Data Runs 1-4 mode>-1: Control Sample Signal Region: 57 12Argus: 12 Crystal Ball: 45

10

Data Runs 1-4 & SP5/6 MC Data Mean: 0.4590 B0B0bar Mean: 0.4566 Data RMS: 0.1992 B0B0bar RMS: 0.1861

10

8

8
6

6 4
4

2 0 5.2

2

5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29 5.3 mES [GeV]

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9 1 NN Out

The selection for the extra control sample. At left, the distributions in data. At right, the neural network output in data and Monte Carlo samples. Only the peaking components of data (dots), B0B0bar (solid) and the signal cocktail (dashed) are plotted.

╕

The observed number of events in the control sample is . From Monte Carlo simulation, , yielding a relative the expected number of events in the control sample is . Modulo error, the observed number of events in the extra control discrepancy of sample agrees with the expected number.

!

1

Collaboration Meeting, 7 December 2004 н p.18/25

Control Sample: Inputs
entries/bin entries/bin 30
Data Runs 1-4 & SP5/6 MC Data Mean: 2.0172 B0B0bar Mean: 2.0795 Data RMS: 1.4202 B0B0bar RMS: 1.2856

14 12 10 8 6

Data Runs 1-4 & SP5/6 MC Data Mean: -0.3652 B0B0bar Mean: -0.4127 Data RMS: 0.5897 B0B0bar RMS: 0.5504

25

20

15

10 4 5 2 0

0

0

1

2

3

4

5 mode
Data Runs 1-4 & SP5/6 MC Data Mean: 1.1543 B0B0bar Mean: 1.0411 Data RMS: 0.5590 B0B0bar RMS: 0.5548

-1

-0.5

0

0.5

1 cos( +, -)
Data Runs 1-4 & SP5/6 MC Data Mean: 0.0957 B0B0bar Mean: 0.1098 Data RMS: 0.0858 B0B0bar RMS: 0.0824

entries/bin

entries/bin

12

18 16 14 12 10

10 8 6 4 2

8 6 4 2

0 0

0.5

1

1.5

2 p

2.5 [GeV] +

0

0

0.05

0.1

0.15

0.2 E

GPL

0.25 0.3 -EpDM [GeV]

Clockwise from top left are the signal multiplicity mode, cosine subtended by tau daughter candidate Collaboration Meeting, 7 December 2004 momenta, remaining neutral energy and magnitude of the tau daughter candidate momentum. н p.19/25

Systematic Errors
е

д

1

з

е

д

1

з

е

д
вб вб

Systematic Effect MC Stat and Fit Parameters Fit PDF Choice Electron Particle ID Muon Particle ID Kaon Particle ID Photon Energy Resolution Signal Decay Model Tracking Efficiency Remaining Photon Energy Total Error Relative systematic errors
йи
е

вб

б

б

жз

б

д

з

д

з

д

з

д

ед ╕

ед ╕

ед ╕

ед ╕

з

д

'

0

'

0

'

д

'

!

д

!

"

!

!

!

!

!

!

!

!

!

for the signal cocktail, signal generic and B0B0bar efficiencies.

Remaining neutral energy systematic: relative discrepancy between control sample observation and expectation.

Signal decay systematic: relative diffference between EvtGen and Tauola kinematic parameters. Collaboration Meeting, 7 December 2004 н p.20/25

б

!

!

д

жз

1

з

Limit Setting Procedure
в

1

б

To obtain the confidence level upper limit on the branching ratio on the number of observed events corresponding upper limit

в

#

й

б

в

б

#

в

б

й

и

б╕

д

'╕

б

#

#

д╕

й

й

б

в

б

, we determine the

в

$

ед ╕

д

where follows a Poisson distribution. We do so by finding such that We follow the approach taken by Cousins and Highland (and Barlow) and imagine , , and fluctuate according to their parent distributions. in which

. experiments

з

з

б

вб

е

й

б

#

ж

$

б╕

д

╕

б

We first fix a low hypothesis , , and .

д╕

б

'

for the branching ratio. Then for each experiment, we sample is calculated. If experiments are carried

й

б╕

д

╕

б

After such experiments, the fraction of experiments for which , the hypothesis is increased and another it is larger than out.
б
и

'

ед ╕

б

б

1

The procedure repeats until the fraction of experiments with . tolerance,
1

б

б

и

is, to within a small

й

Collaboration Meeting, 7 December 2004 н p.21/25

!

'

ж

!

'

ж

б

и

#

й

и

Upper Limit Optimization (1)
The optimization proceeds by sampling each cut value from a uniform distribution on an interval appropriate for the parameter. The systematic errors on the signal efficiency and background expectation for this set of cut values is computed. Assuming that the number of observed data events is equal to the background expectation, the is determined. upper limit
в

#

й

б

We sampled experiments

sets of cut values. For each set of cut values, the number of trial is .

The cut values which minimize the upper limit over the sampled set are

б
и

╕

!#

$

$

$

в

и

й

з

д

$

з

б в╕

! $

!

д

$

з

б в╕

й
$

Collaboration Meeting, 7 December 2004 н p.22/25

Upper Limit Optimization (2)
0.002 0.0019 0.0018 0.0017 0.0016 0.0015 0.0014 0.0013 0.0012 0.0011 0.1 0.2 0.3 0.4 0.5 0.6 0.002 0.0019 0.0018 0.0017 0.0016 0.0015 0.0014 0.0013 0.0012 0.0011 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

0.002 0.0019 0.0018 0.0017 0.0016 0.0015 0.0014 0.0013 0.0012 0.0011 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3

0.002 0.0019 0.0018 0.0017 0.0016 0.0015 0.0014 0.0013 0.0012 0.0011 0.2 0.3 0.4 0.5 0.6 0.7

в

1

The confidence level upper limit plotted versus cut values for (clockwise from top left) lower bound for the neural network output, upper bound for unassigned photon energy, upper bound for and lower bound for .

д

з

д

#

й

б

$

б

в╕

б

$

╕

в

$

Collaboration Meeting, 7 December 2004 н p.23/25

$

з

Results
в

We are still blind for the last three analysis chain requirements. Below, the upper limit . We assume the values of outcomes

б

for a range

!

'

ж

!

!

!"

б╕

д

б

╕

и

б

'

ед ╕

и

и

!

The range of

below does not cover many outcomes. See BAD 542 v4 for
в

!

!

'

ж

б

в

ж

й

ж

!

'

#

!

'

#

!

й

б

"

!

!

!

!

!

,

!

!

!

б

'

ед ╕

и

и

Collaboration Meeting, 7 December 2004 н p.24/25

#

coverage.

й

Outlook
We will fix knowm problems and incorporate feedback received this week. We will then release version 5 of BAD 542, hopefully early next week. We hope to have an AWG BAD reading next week. We would like to move the analysis to the Review Committee as soon as we have AWG approval.

Collaboration Meeting, 7 December 2004 н p.25/25