8/16/2004 ICHEP2004, Beijing

Rare Kaon Decay Results From E949 At BNL: K+

Shaomin CHENTRIUMF, Canada

Motivation is clear and simple

(sin2sinJ/

Ks

Use K+ and K0 to measure the CPV related elements

To construct the K unitarity triangle and confirm the “golden” relation which is valid in the SM and MFV.

A.J. Buras et.al hep-ph/0405132

Experiment is challenging

Weapons needed to suppress background by 1010

in order to see the signal.

Process EventsK 6343000000K 2113000000K 55000000Beam bkg 25000000Knp, Kl-

46000

K 1 Where can we find such a “WMD”?

Standard Model: Br(K+ 10-10

BNL-E949 detector -- Rejecting beam backgrounds

e+

Beam backgrounds include pion scattering, kaon decay-in-flight, and charge-exchange reactions.

cerenkov

K cerenkov

B4

K

Top half of side view

Charge exchange

Beam 1

Beam 2

K cluster

cluster

K decay

Target fibres

BNL-E949 detector -- Powerful and redundant particle ID

Resolutions:P/P ~ 1.1%; R/R~ 3.0%; Rejectionsfor ~for photon with 4 sr coverage.

Top half of end view

-ID from its decay chain.

dE/dxR/P

Photon veto

E787

E949

This is a team effort!

16 institutes and universities in 4 countries.

…more than 60 physicists from 6 countries.

BNL/FNAL/SBU/UNM, U.S.AIHEP/INR, Russia Fukui/KEK/Kyoto/NDA/Osaka, Japan TRIUMF/UA/UBC, Canada

Many thanks to BNL-AGS people

Before data taking

After data taking

Twice the instantaneous intensity

Platinum target used in 2002

Unique Strategy for data analysis

Full data set

1/3 data 2/3 data

Beam bkg

Background study

Background estimate

Step 1

Step 2

Step 3

Step 4

Open the box Step 5E787/E949: we bought a one-way ticket,-no way back after opening the box!

Cuts tuningE787: <0.1 eventE949: <0.5 event

Background suppression

CutBkg

Kinematics cuts(p/R/E)

Particle ID

Photon veto

Timing cuts

K

K

Beams

Knp, Kl-

How to achieve a reliable background estimate?

Blind analysis in tuning the cuts. Bifurcated analysis in estimating backgrounds.

1. Form two independent cuts.2. Estimate the background.3. Check the correlation.

Backgrounds inside the box

Items E949 E787Nk(1012) 1.8 5.9K 0.044 ± 0.005 0.062 ± 0.045K 0.216 ± 0.023 0.034 ± 0.007Beams 0.024 ± 0.010 0.025 ± 0.016Knp, Kl-

0.014 ± 0.003 0.025 ± 0.008

Total bkg (evts) 0.298 ± 0.026 0.146 ± 0.049Errors are statistical only.

What could be our nightmares?

Correlation between two bifurcated cuts.

A 0.3 event background estimated, but more than 3 events observed!

Appearance of pathology backgrounds.

Underestimating background because…

For example…

It can happen if an observable is used in both CUT1 and CUT2 categories, spoiling the independence of the bifurcated analysis.

It rarely happens but can appear thru a loophole. Hand-scanning the events outside the box is the only way to spot the problem.

Example1: Correlation between two cuts

Loosen PV x KIN

Obs. Pred.

20 x 20 3 1.820 x 50 8 3.050 x 50 22 11.1

Loosen PV x KIN

Obs. Pred.

20 x 20 4 4.920 x 50 9 12.450 x 50 22 31.1

A new geometrical cut that gained 10% more rejection again 0 was found to

cause trouble in the 1/3 outside-the-box study: N(obs.)~2.0 N(pred.)

retune PV

Reject any event with a photon in opposite side.

Reason: a new cut was developed that used the direction of the + to place a tighter requirement again photons Remedy: remove this cut & retune PV.

Example 2: Pathology background

K+

K+

B4

Target

Pions scattered back into the target overlaying a non-decaying kaon can make a fake signal.

x

y

zr

Reason: track reconstruction only down to the kaon decay vertex.Remedy: check the hits in the opposite direction of the track.

Readout fibre

Backgrounds outside the box

Motivations:1) To check the correlations;2) To estimate systematic errors.

Bkg. C Prob. 0.85 0.17 1.15 0.67 1.06 0.40

Fit Nobs. = c Npred.

Npred.

Nobs.

Deviations of c from unity give the systematic errors.

Acceptance

Br(P.D.G value is 0.211±0.001

Phase space and nuclear interaction effects from Monte Carlo.Acceptance loss due to PID, photon veto, timing and other kinematics cuts directly measured from data.

E949 acceptance: 0.0022 ± 0.002E787 acceptance: 0.0020 ± 0.002

Verification from B(measurements.

Open the box and guess what?

We see one new event!

E949

Branching Ratio Result

To make best use of our knowledge of the signal and the background distributions inside the box, we define cells inside the box.

E949 E787

Candidate E949A E787A E787C

S/b 0.9 50 7

W 0.48 0.98 0.88

Br ( 1.47 )10-10

Bi: background of cell containing candidate.Si : Br Ai NK.

NK: Stopped K+’s.Ai: acceptance in a cell.Br: branching ratio.W: S/(S + b).

Impact on CKM matrix

More data needed for reducing the uncertainty.New neutral kaon experiments (for example KOPIO) expected for further justification.

Many thanks to A. Höcker & J. Ocariz et alThe CKM fitter group, hep-ph/0406184

Summary

A successful run of E949 with a factor of two more beam intensity as in E787.

A new candidate event observed for K+ , leading to a combined result from E787/E949,

Br(K+ 10-10

~2Br(K+ SM, if (sin2sinJ/Ks.

Statistical fluctuation or evidence of new physics?

hep-ex/0403036