Low energy e + e –  hadrons in Novosibirsk I.Logashenko On behalf of the CMD-2 and SND Collaborations Budker Institure of Nuclear Physics (Novosibirsk, Russia) Boston University (Boston, USA) 10 th International Workshop On Tau Lepton Physics Novosibirsk, Russia, Sep 22 – 25, 2008

Cross-section e + e -  hadrons Measurement of the cross-section e+e-  hadrons in VEPP-2M energy range is interesting for: measurement of R(s) measurement of parameters of light vector mesons ρ, ω, φ, ρ’, ρ’’, ω’, ω’’ comparison with spectral functions of the hadronic tau decays At low s R(s) has to be measured. The value and the error of the hadronic contribution to muon’s (g-2) are dominated by low energy R(s).

R(s) measurements at low s VEPP-2M Babar/Belle (ISR) KLOE (ISR) VEPP-2000 At VEPP-2M the cross-sections of each final state are measured exclusively

VEPP-2M collider VEPP-2M collider: GeV in c.m., L  /cm 2 s at 1 GeV Detectors CMD-2 and SND: 70 pb -1 collected in

CMD-2 detector

SND detector 1 - beam pipe, 2 - drift chambers, 3 - coincidence counter, 4 - fibre light guide, 5 - PMTs, 6 - NaI(Tl) crystals, 7 – phototriodes, 8 - iron absorber, 9 - muon tubes, cm iron plate, 11 - muon counters, 12 - magnetic lenses, 13 - bending magnets

Example of CMD-2 and SND events e + e -  π + π - in CMD-2e + e -  K + K - in SND

Overview of the results

How cross-section is measured Luminosity L is measured using Bhabha scattering at large angles Efficiency  is calculated via Monte Carlo + corrections for imperfect detector Radiative correction  accounts for ISR effects only All modes except 2  Ratio N(2  )/N(ee) is measured directly  detector inefficiencies are cancelled out Virtually no background Analysis does not rely on simulation (sort of) Radiative corrections account for ISR and FSR effects Formfactor is measured to better precision than L (1-2%) 22

e+e-π+π-e+e-π+π-

Experimental data 96 95, ,98 98,2000 CMD-2 SND Events signature: two back-to-back tracks, originated near the interaction point Data sample includes: e+e-, μ+μ-, π+π-, cosmic muons There is almost no background at √s <1 GeV Data were taken in 6 separate runs between 1994 and 2000

Event separation (CMD-2) e/  /  separation using particles momentum can measure N(  )/N(ee) and compare to QED <0.6 GeV Momentum Energy >0.6 GeV e/  /  separation using energy deposition N(  )/N(ee) is fixed according to QED Momentum Energy Likelihood minimization:

Event separation (SND) Event separation is based on neural network: 7 input parameters: energy deposition in each layer for both clusters and polar angle 2 hidden layers 20 neurons each 1 output parameter – R e/π Trained on simulated events Checked on experimental 3π and e+e- events E 1 (1,2,3), E 2 (1,2,3),Θ R e/π Distribution by separation parameter Misidentification ~ %

Radiative corrections ISR+FSR ISR+FSR+VP Radiation terms Vacuum polarization CMD-2 uses custom Monte- Carlo generator to calculate RC ee, ,  final states: 1  at large angle, multiple  ’s along initial or final particles (≤0.2%) CMD-2 calculation is consistent with independent calculations (BHWIDE, KKMC) SND uses BHWIDE for ee final state and CMD-2 generator for ,  final states

Comparison with other calculations of the radiative corrections Comparison with BHWIDEComparison with KKMC

Reconstruction efficiency correction (CMD-2) All selected final states produce very similar signal in the drift chamber. In the first order, the reconstruction efficiency is cancelled out from the formfactor calculation Due to drift chamber malfunction, the correction for reconstruction efficiency reach up to 4% percent for 1998 data set. Measured ratio ε ππ /ε ee

Pion formfactor - results SND 3.2% 1.3% CMD2 0.7% 0.6% (95)/ 0.8% (98) % SND only CMD2 only

Internal Cross-checks Δ(95-98)≈0.7%±0.5% CMD2, 2 independent scans of rho-region CMD2 vs SND √s<0.52 GeV Δ(SND-CMD2)≈1.2%±3.6% CMD2 vs SND 0.6<√s<1.0 GeV Δ(SND-CMD2)≈-0.53%±0.34% CMD2 √s<0.52 GeV

Other modes

“Narrow” resonances Mass and width are measured to ≈0.1 MeV, Γ ee to ≈2-3%

Cross-section e+e-  4π Systematic error: Systematic error ≈5-7% Efficiency determination gives main contribution to the systematic error SND = 8% CMD2= 15% (discrepancy 15-25%) problem with efficiency determination! CMD2-reanalysis preliminary = 8%

Cross-section e+e-  3π Systematic error: ≈6% (>1GeV) 1-2% on omega 2.5% on phi Fit:

Cross-section e+e-  2K CMD2 SND Systematic error ≈5-8% Systematic error ≈3-9% Systematic error is ≈2-3% at φ resonance CMD2(prel.) SND

Overview of the results 1.0%~ % 0.6% % 1.5% ~ % 1--2% % 2.0% Systematic error: Total error: Error of R(s)

Future measurements at VEPP-2000

VEPP-2000 storage ring Up to 2 GeV c.m. energy Factor >10 in luminosity L=10 31 cm -2 c -1, √s=1.0 GeV L=10 32 cm -2 c -1, √s=2.0 GeV ≈100 1/pb per detector per year Status: construction is finished had first beams and collisions 2009 – start of experiments

CMD-3 Detector 1 – vacuum tube, 2 – drift chamber, 3 – calorimeter BGO (680 crystals), 4 – Z–chamber, 5 – CMD-3 superconducting solenoid, 6 – calorimeter LXe (400 liters), 7 – calorimeter CsI (1152 crystals), 8 – magnet yoke, 9 – solenoids of VEPP-2000 Advantages compared to CMD-2: new drift chamber with x2 better resolution better tracking thicker barrel calorimeter better separation LXe calorimeter - much better spatial resolution for γ’s - shower profile

SND – beam pipe 2 – tracking system 3 – aerogel 4 – NaI(Tl) crystals 5 – phototriodes 6 – muon absorber 7–9 – muon detector 10 – focusing solenoid Advantages compared to “old” SND: new system - cherenkov counter (n=1.05, 1.13) e/π separation E<450 MeV π/K separation E<1 GeV new drift chamber better tracking better determination of solid angle

How can we improve measurement of R? We expect to: measure 2  mode to %, 4  mode to 2% overall improvement in R precision by factor 2-3 We expect to get the following improvements: high statistics! (x10-x100) – current measurement is still statistics-limited radiative corrections to 0.1% - add photon jet + large angle γ measure radiative tails and compare them to calculations (ISR?) luminosity to 0.5%, use γγ in addition to Bhabha for cross-check much better separation – smaller systematic error, try to measure e+e-  μ+μ- precise trigger efficiency monitoring better drift chambers – higher resolution, efficiencies

Examples of improvements (CMD-3) CMD-3 CMD-2 Separation by energy deposition Separation by momentum Because of better resolution of the drift chamber, the separation by momentum should work up to √s ≈ 0.65 GeV CMD-2 2*0.26 GeV CMD-3 2*0.32 MeV 4σ4σ √s/2, GeV

Conclusion CMD-2 and SND data analyses are nearly complete. Cross sections of all major modes of e+e-  hadrons are measured at energy range √s < 1.4 GeV. These are the best direct (energy- scan) measurements at the moment. There is good agreement between Novosibirsk results, in particular: pion formfactor CMD-2 (94,95) vs CMD-2 (98), CMD-2 vs SND. Over the last few years new indirect high precision measurements of R were performed: new tau-decay data and ISR. The question of agreement between different methods is still open. New more precise measurement of R is scheduled at VEPP-2000.

Backup slides

Pion formfactor calculation Master formula: σ B – Born cross-section (F π =1) δ – radiative correction ε – reconstruction efficiency Δ N – correction for nuclear interactions Δ D – correction for decay in flight Δ bg – correction for e+e-  3π,4π,2K background Δ corr – correction for E +  E - correlation Nππ/Nee is measured, other values are calculated:

What is really measured? Definition of  (e + e -  hadrons) depends on the application Hadron spectroscopy: vacuum polarization (VP) is the part of the cross-section (“dressed”), final state radiation (FSR) is not “Bare” cross-section used in R: vice versa – FSR is the part of the cross-section, VP is not Measured number of events include VP and part of FSR allowed by the event selection CMD-2 published 2 cross-sections e + e -  2  : radiative correction take into account part of FSR, allowed by the event selection (thus remove FSR completely from the measured cross-section); VP is left untouched. Used to get rho-meson mass, width, … VP is removed, all FSR is added. Used for R calculation VP FSR

Systematic errors Source of errorCMD-2 √s<1 GeV SNDCMD-2 √s>1.0 GeV Event separation %0.5% % Fiducial volume0.2%0.8% % Energy calibration %0.3% % Efficiency correction0.2%-0.5%0.6% % Pion losses (decay, NI)0.2% Other0.2%0.5% % Radiative corrections %0.2% % Total %1.3% %