Belle preprint 2014-21 KEK preprint 2014-37 ¯ Measurement of the direct CP asymmetry in B → Xs+dγ decays with a lepton tag L. Pes´antez,3 P. Urquijo,35 J. Dingfelder,3 A. Abdesselam,55 I. Adachi,12,9 K. Adamczyk,44 H. Aihara,60 S. Al Said,55,26 K. Arinstein,4 D. M. Asner,47 V. Aulchenko,4 T. Aushev,37,21 R. Ayad,55 S. Bahinipati,14 A. M. Bakich,54 V. Bansal,47 E. Barberio,35 V. Bhardwaj,40 B. Bhuyan,15 A. Bobrov,4 A. Bondar,4 G. Bonvicini,65 A. Bozek,44 M. Braˇcko,33,22 T. E. Browder,11 D. Cˇervenkov,5 V. Chekelian,34 A. Chen,41 B. G. Cheon,10 K. Chilikin,21 R. Chistov,21 K. Cho,27 V. Chobanova,34 Y. Choi,53 D. Cinabro,65 J. Dalseno,34,57 Z. Doleˇzal,5 5 Z. Dra´sal,5 A. Drutskoy,21,36 D. Dutta,15 S. Eidelman,4 H. Farhat,65 J. E. Fast,47 T. Ferber,7 O. Frost,7 1 0 V. Gaur,56 N. Gabyshev,4 S. Ganguly,65 A. Garmash,4 D. Getzkow,8 R. Gillard,65 Y. M. Goh,10 B. Golob,31,22 2 J. Haba,12,9 J. Hasenbusch,3 H. Hayashii,40 X. H. He,48 A. Heller,24 T. Horiguchi,59 W.-S. Hou,43 n M. Huschle,24 T. Iijima,39,38 K. Inami,38 A. Ishikawa,59 R. Itoh,12,9 Y. Iwasaki,12 I. Jaegle,11 D. Joffe,25 u T. Julius,35 K. H. Kang,29 E. Kato,59 T. Kawasaki,45 C. Kiesling,34 D. Y. Kim,52 J. B. Kim,28 J. H. Kim,27 J K. T. Kim,28 M. J. Kim,29 S. H. Kim,10 Y. J. Kim,27 B. R. Ko,28 P. Kodyˇs,5 S. Korpar,33,22 P. Kriˇzan,31,22 1 P. Krokovny,4 B. Kronenbitter,24 T. Kuhr,24 T. Kumita,62 A. Kuzmin,4 Y.-J. Kwon,67 J. S. Lange,8 2 I. S. Lee,10 Y. Li,64 L. Li Gioi,34 J. Libby,16 D. Liventsev,12 P. Lukin,4 D. Matvienko,4 K. Miyabayashi,40 ] H. Miyata,45 R. Mizuk,21,36 G. B. Mohanty,56 A. Moll,34,57 H. K. Moon,28 E. Nakano,46 M. Nakao,12,9 x e T. Nanut,22 Z. Natkaniec,44 M. Nayak,16 C. Ng,60 N. K. Nisar,56 S. Nishida,12,9 S. Ogawa,58 S. Okuno,23 - S. L. Olsen,51 C. Oswald,3 P. Pakhlov,21,36 G. Pakhlova,21 C. W. Park,53 H. Park,29 T. K. Pedlar,32 R. Pestotnik,22 p e M. Petriˇc,22 L. E. Piilonen,64 E. Ribeˇzl,22 M. Ritter,34 A. Rostomyan,7 M. Rozanska,44 Y. Sakai,12,9 S. Sandilya,56 h L. Santelj,12 T. Sanuki,59 Y. Sato,38 V. Savinov,49 O. Schneider,30 G. Schnell,1,13 C. Schwanda,18 A. J. Schwartz,6 [ K. Senyo,66 O. Seon,38 M. E. Sevior,35 V. Shebalin,4 C. P. Shen,2 T.-A. Shibata,61 J.-G. Shiu,43 B. Shwartz,4 3 A. Sibidanov,54 F. Simon,34,57 Y.-S. Sohn,67 A. Sokolov,19 E. Solovieva,21 M. Stariˇc,22 M. Steder,7 T. Sumiyoshi,62 v U. Tamponi,20,63 N. Taniguchi,12 G. Tatishvili,47 Y. Teramoto,46 K. Trabelsi,12,9 M. Uchida,61 T. Uglov,21,37 2 0 Y. Unno,10 S. Uno,12,9 Y. Usov,4 C. Van Hulse,1 P. Vanhoefer,34 G. Varner,11 A. Vinokurova,4 V. Vorobyev,4 7 M. N. Wagner,8 B. Wang,6 C. H. Wang,42 M.-Z. Wang,43 P. Wang,17 Y. Watanabe,23 K. M. Williams,64 1 E. Won,28 J. Yamaoka,47 S. Yashchenko,7 Y. Yook,67 Z. P. Zhang,50 V. Zhilich,4 V. Zhulanov,4 and A. Zupanc22 0 . 1 (The Belle Collaboration) 0 5 1University of the Basque Country UPV/EHU, 48080 Bilbao 1 2Beihang University, Beijing 100191 v: 3University of Bonn, 53115 Bonn i 4Budker Institute of Nuclear Physics SB RAS and Novosibirsk State University, Novosibirsk 630090 X 5Faculty of Mathematics and Physics, Charles University, 121 16 Prague ar 6University of Cincinnati, Cincinnati, Ohio 45221 7Deutsches Elektronen–Synchrotron, 22607 Hamburg 8Justus-Liebig-Universit¨at Gießen, 35392 Gießen 9The Graduate University for Advanced Studies, Hayama 240-0193 10Hanyang University, Seoul 133-791 11University of Hawaii, Honolulu, Hawaii 96822 12High Energy Accelerator Research Organization (KEK), Tsukuba 305-0801 13IKERBASQUE, Basque Foundation for Science, 48013 Bilbao 14Indian Institute of Technology Bhubaneswar, Satya Nagar 751007 15Indian Institute of Technology Guwahati, Assam 781039 16Indian Institute of Technology Madras, Chennai 600036 17Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049 18Institute of High Energy Physics, Vienna 1050 19Institute for High Energy Physics, Protvino 142281 20INFN - Sezione di Torino, 10125 Torino 21Institute for Theoretical and Experimental Physics, Moscow 117218 Typeset by REVTEX 1 22J. Stefan Institute, 1000 Ljubljana 23Kanagawa University, Yokohama 221-8686 24Institut fu¨r Experimentelle Kernphysik, Karlsruher Institut fu¨r Technologie, 76131 Karlsruhe 25Kennesaw State University, Kennesaw GA 30144 26Department of Physics, Faculty of Science, King Abdulaziz University, Jeddah 21589 27Korea Institute of Science and Technology Information, Daejeon 305-806 28Korea University, Seoul 136-713 29Kyungpook National University, Daegu 702-701 30E´cole Polytechnique F´ed´erale de Lausanne (EPFL), Lausanne 1015 31Faculty of Mathematics and Physics, University of Ljubljana, 1000 Ljubljana 32Luther College, Decorah, Iowa 52101 33University of Maribor, 2000 Maribor 34Max-Planck-Institut fu¨r Physik, 80805 Mu¨nchen 35School of Physics, University of Melbourne, Victoria 3010 36Moscow Physical Engineering Institute, Moscow 115409 37Moscow Institute of Physics and Technology, Moscow Region 141700 38Graduate School of Science, Nagoya University, Nagoya 464-8602 39Kobayashi-Maskawa Institute, Nagoya University, Nagoya 464-8602 40Nara Women’s University, Nara 630-8506 41National Central University, Chung-li 32054 42National United University, Miao Li 36003 43Department of Physics, National Taiwan University, Taipei 10617 44H. Niewodniczanski Institute of Nuclear Physics, Krakow 31-342 45Niigata University, Niigata 950-2181 46Osaka City University, Osaka 558-8585 47Pacific Northwest National Laboratory, Richland, Washington 99352 48Peking University, Beijing 100871 49University of Pittsburgh, Pittsburgh, Pennsylvania 15260 50University of Science and Technology of China, Hefei 230026 51Seoul National University, Seoul 151-742 52Soongsil University, Seoul 156-743 53Sungkyunkwan University, Suwon 440-746 54School of Physics, University of Sydney, NSW 2006 55Department of Physics, Faculty of Science, University of Tabuk, Tabuk 71451 56Tata Institute of Fundamental Research, Mumbai 400005 57Excellence Cluster Universe, Technische Universita¨t Mu¨nchen, 85748 Garching 58Toho University, Funabashi 274-8510 59Tohoku University, Sendai 980-8578 60Department of Physics, University of Tokyo, Tokyo 113-0033 61Tokyo Institute of Technology, Tokyo 152-8550 62Tokyo Metropolitan University, Tokyo 192-0397 63University of Torino, 10124 Torino 64CNP, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061 65Wayne State University, Detroit, Michigan 48202 66Yamagata University, Yamagata 990-8560 67Yonsei University, Seoul 120-749 (Dated: June23, 2015) 2 Abstract We report the measurement of the direct CP asymmetry in the radiative B¯ → X γ decay using a data sample of s+d (772±11)×106 BB¯ pairs collected at the Υ(4S) resonance with the Belle detector at the KEKB asymmetric-energy e+e− collider. The CP asymmetry is measured as a function of the photon energy threshold. For E∗ ≥ 2.1 GeV, where E∗ is the γ γ photon energy in the center-of-mass frame, we obtain ACP(B¯ → Xs+dγ) = (2.2±3.9±0.9)%, consistent with the Standard Model prediction. PACSnumbers: 11.30.Er,13.25.Hw The radiative electroweak transitions b → sγ and the first Belle measurement of A (B¯ → X γ). We CP s+d b → dγ proceed via flavor-changing neutral currents in- profit from the large data sample available at Belle to volvingloopdiagrams. Thesedecaysaresensitivetopos- achieve a higher statistical precision. siblecontributionsfromnewheavyparticlesoccurringin The states X include resonant contributions such s+d the loop, which modify the branching fractions and CP- asK∗(892),ρandω,andnon-resonantcontributions. In violating effects predicted in the Standard Model (SM). order to be sensitive to all X states, the selection is s+d The decay rates, including QCD corrections, can be ex- basedon the high-energy-photonsignature of the transi- pressedbyaneffective Hamiltonianandcalculatedusing tion, i.e. the radiated photon is the only reconstructed the Operator Product Expansion approach. In the lead- particle from the B¯ → X γ decay. While this ap- s+d ingandnext-to-leadingorderlogarithmicapproximation, proachdoes not exclude explicitly possible contributions the branching fractions and CP asymmetries are pro- from B¯ → X γ or B¯ → X γ decays, such contributions c u portional to the dipole operators P7 and P8 [1]. New are very small in the SM [10] and will be neglected in physics effects would modify the corresponding Wilson this analysis. To tag the signal B flavor,we use the fact coefficients C7 and C8. that B mesons are produced in pairs from the reaction The CP asymmetry (ACP) in B¯ → Xs+dγ decays is e+e− → Υ(4S) → BB¯. The flavor of the signal B me- defined as: sonis determined by taggingthe flavorof the other B in ACP(B¯ →Xs+dγ)≡ ΓΓ((BB¯¯ →→XXss++ddγγ))−+ΓΓ((BB →→XXss¯¯++dd¯¯γγ)), ttlehhpeetoseenvmecnihltea,prugtsoeinniingcsadeemcchailayerpgotefodtnhilceepdotteohcnaeyr(seB,a.µreT)dhcieorneBcstislflytaevrneotlrawtaeintdhd. (1) where Γ(B¯ → X γ) represents the decay rate of the SincetheexpectedCP violationisverysmallandpre- s+d B0 or B− meson into the radiative final state. In the cisely calculable, all effects that could bias the measure- following, charge-conjugate states are included implic- ment must be carefully quantified. A measurement bias itly. The X states represent all possible hadronic isintroducediftheselectionprocedure,trackreconstruc- s+d final states derived from b → sγ or b → dγ transi- tion, or particle identification favors a particular charge. tions. TheSMpredictsA forthethesetwotransitions These effects are quantified in different control samples. CP in the ranges −0.6% ≤ ACP(B¯ → Xsγ) ≤ 2.8% and In this analysis, we also test the independence of ACP −62% ≤ A (B¯ → X γ) ≤ 14% [2]. Even though the with respect to the choice of cutoff energy,by measuring CP d individual CP-violating effects could be large, the CP- it as a function of the photon energy threshold. violating contributions cancel when both are considered Thisanalysisusesthe711 fb−1 samplerecordedatthe inclusively due to CKM unitarity, and the theory errors Υ(4S) resonance by the Belle experiment at the KEKB cancel almost perfectly except for small U-spin breaking storage ring [11], containing (772±11)×106 BB¯ pairs. corrections [3], additionally, the inclusive asymmetry is An89 fb−1samplerecordedatacenter-of-mass(CM)en- insensitivetothechoiceofphotonenergycutoff[4]. This ergy 60 MeV below the resonance is used to study con- precise SM prediction of A (B¯ → X γ)=0, serves tinuum background (e+e− → qq¯, where q = u,d,s,c); CP s+d asa cleantestfor new CP-violatingphases actinginthe the former sample is denoted on-resonance and the lat- decays. New physics (NP) scenarios such as supersym- ter off-resonance. The Belle detector is a large-solid- metricmodelswithminimalflavorviolationpredictA angle magnetic spectrometer that consists of a silicon CP (B¯ →X γ)uptoalevelof+2%. InmoregenericNP vertex detector (SVD), a 50-layer central drift chamber s+d scenarios, the asymmetries A (B¯ → X γ) and A (CDC), an array of aerogel threshold Cerenkov counters CP s CP (B¯ →X γ) do not cancel and A (B¯ →X γ) is the (ACC),abarrel-likearrangementoftime-of-flightscintil- d CP s+d mostsensitiveobservable,withvaluesaslargeas10%[3]. lationcounters(TOF),andanelectromagneticcalorime- Previous measurements of A (B¯ → X γ) have ter comprised of CsI(Tl) crystals (ECL) located inside CP s+d been performedby CLEO [5] and BaBar [6] and are sta- a super-conducting solenoid coil that provides a 1.5 T tistically limited. Belle has performed a measurement magneticfield. Anironflux-returnlocatedoutsideofthe of the inclusive branching fraction [7]. The asymmetry coilis instrumentedto detect K0 mesons andto identify L A (B¯ → X γ) has been measured separately as the muons (KLM). The detector is described in detail else- CP s sum of exclusive decays [8, 9]. In this letter, we present where [12]. 3 Monte Carlo (MC) simulation samples were gener- malized transverse deviations between the track and the ated to study continuum background, BB¯ decays and KLM hits associated to it. The polar angle requirement B¯ →X γ signal events. The size of the BB¯ MC sample for muons is 25◦ ≤θ ≤145◦. s µ isequivalenttotentimestheintegratedluminosityofthe After this initial selection, the sample is dominated data. The sizeofthe continuumMCsamplecorresponds by continuum background,which constitutes 77% of the to the integrated luminosity of the on-resonancesample. total yield; the signal component amounts only to 1% ThegenerationofthesignalB¯ →Xsγ decaysfollowsthe as can be seen in Fig. 1(a). To suppress the contin- theoretical prediction of the Kagan-Neubert model [13] uumbackground,weuseaBoostedDecisionTree(BDT), with parameters mb =4.574 GeV and µ2π =0.459 GeV2 thatistrainedtoachievethebestdiscriminationbetween representing the b-quark mass and mean kinetic energy. continuum and signal events. Eighteen kinematic, event The signal sample contains 2.6 million B¯ →Xsγ events, shape and isolation variables are used as input for the which corresponds roughly to five times the number ex- BDT: eleven Fox-Wolfram moments [15], constructed in pected in data. The BB¯ and B¯ → Xsγ MC samples threesets inwhich(1)allparticlesinthe eventareused, included B0-B¯0 mixing. (2)thesignalphotonisexcludedand(3)bothsignalpho- Inthis analysis,trackspassingveryfarfromthe inter- ton and tag lepton are excluded; the magnitude and di- action point or compatible with a low-momentum par- rectionoftheevent’sthrustvector;the distancebetween ticle reconstructed multiple times as it spirals in the the photon cluster and the closest extrapolated position CDC are excluded. For photons, minimum energies of of a charged particle at the ECL surface; the angle be- 100 MeV, 150 MeV and 50 MeV,respectively, are re- tween the directions of the photon and tag lepton; the quired in the forward, backward and barrel regions of RMS width of the photon cluster; the scalar sum of the ECL, defined in Ref. [12]. These requirements suppress transverse momenta of all reconstructed particles; and low-energy photons resulting from particle interactions the square of the missing four-momentum, calculated as with detector material or the beam pipe. All particles thedifferencebetweenthetotalbeamenergyandthemo- areusedtocalculatekinematicandtopologicalvariables. mentaofallreconstructedparticles. The BDTis trained Thesignalphotoncandidatesareselectedasconnected using continuum and B¯ → X γ MC samples. The se- s ◦ clusters of ECL crystals in the polar angle 32.2 ≤θ ≤ lection criterion on the BDT output classifier variable is γ ◦ ∗ 128.7 withaCMenergy1.4GeV≤E ≤4.0GeV. The chosen to minimize the expected statistical uncertainty γ polar angle is measured from the z axis that is collinear on A . The BDT classifier distribution and selection CP with the positron beam. The ratio of the energy deposit criterionareshowninFig.2. Thephotonspectrumafter inthecentral3×3crystalstothatinthecentral5×5crys- continuumsuppressionis showninFig. 1(b) for MC and tals must be larger than 90%. Photons from the decays on-resonance data, in this plot we include statistical un- π0(η)→γγarerejectedbyusingavetobasedonthepho- certainties and systematic uncertainties that come from ton energy, polar angle and the reconstructed diphoton calibration and normalization factors, that cancel in the mass,asdescribedinRef.[14]. Thesignalregionincludes measurement of A . CP photons with CM energy 1.7 GeV ≤ E∗ ≤ 2.8 GeV; the After the selection, in the region 1.7 GeV ≤ E∗ ≤ γ γ ∗ ∗ sidebands E < 1.7 GeV and E > 2.8 GeV are used 2.8 GeV, we find 21400 (21608) events tagged with a γ γ to study the normalization of BB¯ and continuum back- positive(negative)leptonintheon-resonancesampleand ground components, respectively. 2623±140 (2728± 143) events tagged with a positive Theleptoncandidatesusedfortagging(taglepton)are (negative) lepton in the off-resonance sample. The off- reconstructed as tracks in the SVD and CDC. We limit resonance events are corrected as they have, on average, the impact parameters along the z axis to |dz| ≤ 2 cm lowerparticleenergiesandmultiplicitiesduetothelower and dr ≤ 0.5 cm, require at least one hit in the SVD, CMenergy. Additionally,theoff-resonanceyieldisscaled and choose a momentum range in the CM frame of to take into account the difference in luminosities and 1.10 GeV ≤ p∗ ≤ 2.25 GeV. The upper-momentum cross-sections. ℓ bound reduces continuum background as it is near the Thesignalfractionis21.2%whilethecontinuumback- kinematic limit for leptons from B decays. The lower ground fraction is 12.4%. The BB¯ background contains bound ensures that most of the selected leptons origi- photons from several processes. The dominant sources nate directly from a B meson, which is important for arephotonsfromπ0 →γγ decays,whichmakeup49.5% flavortagging. Electroncandidatesareidentifiedbycon- ofthetotalyieldandphotonsfromη →γγ,contributing structing a likelihoodratio basedon the matching of the 7.9%. Photons from beam background are 2.2% of the cluster in the ECL and the extrapolated track, the ratio total contribution. Electrons and hadrons misidentified between its energy and momentum, the shower shape in as photons are small contributions of 0.8% and 0.2%, the ECL,the energylossinthe CDC,andthe lightyield respectively. Other photons, mainly from decays of ω, ′ inthe ACC.The polaranglerequirementforelectrons is η and J/ψ mesons, and bremsstrahlung, including final ◦ ◦ 18 ≤ θ ≤ 150 . Muon identification uses a likelihood state radiation [16], comprise the remaining 5.8%. The e ratiodeterminedfromtherangeofthetrackandthenor- B¯ →X γ signalisobtainedbysubtractingthecontin- s+d 4 ×106 Β1403 0.7 Data 9 Data 0.6 (a) MC signal 8 (b) MC signal BB MC BB MC V V 7 e 0.5 Continuum e Continuum G G 1 1 6 Total error 0. 0.4 0. 5 es / 0.3 es / 4 Entri 0.2 Signal × 50 Entri 3 2 0.1 1 0 0 1.5 2 2.5 3 3.5 4 1.5 2 2.5 3 3.5 4 Eγ* (GeV) Eγ* (GeV) FIG. 1: Photon energy spectrum in the CM frame showing on-resonance data, off-resonance data for continuum, and MC simulation. Thespectrumisshown(a)beforeand(b)aftercontinuumsuppression. In(a),theMCsignalisadditionallyplotted scaled a factor of fifty to show its expected position. In (b), the MC error includes statistical and systematic uncertainties coming from calibration and normalization factors that cancel in themeasurement of ACP. as a correction factor. MC signal Somebackgroundcomponentshaveanon-vanishingdi- n BB MC rectCP asymmetrythatcouldimpactourmeasurement. atio 10-1 Continuum Most have negligible contributions to the decay rate ex- aliz cept for B → Xsη decays, which comprises 1.2% of the m 10-2 rate according to the MC prediction, with a branching nor fractionB(B →Xsη)=(cid:0)26.1±3.0−+21..19+−47..01(cid:1)×10−5 and ary 10-3 a CP asymmetry ACP(B → Xsη) = (−13±5)% mea- bitr sured by Belle [17]. The MC is corrected to model this Ar effect properly. 10-4 The B¯ →X γ photon energy spectrum for positive s+d andnegativetaggedeventsaftersubtractingalltheback- -1 -0.5 0 0.5 1 BDT output ground is shown in Fig. 3. The measured asymmetry, Ameas, is calculated using Eq. (1) expressed in terms of CP N+−N− FIG.2: OutputoftheBDT.Thecontinuumdistributioncor- the charge-flavorcorrelation: Ameas = . Here, responds to off-resonance data. The vertical line denotes the CP N++N− minimum requirement on this variable. N+ andN− representthe totalnumber ofeventstagged by a positive or negative lepton for a given photon en- ergy threshold. The energy thresholds range from 1.7 to uum and BB¯ contributions. The BB¯ background com- 2.2 GeV. ponents are calibrated using data, as described below. The measured values must be corrected due to pos- sible asymmetries in the BB¯ background that is sub- All corrections and calibrations applied to MC and off- resonance data are determined and performed indepen- tracted(Abkg)andpossibleasymmetriesinthedetection ofleptons A . An additionalcorrectionarisesfromthe dently of the tag charge. The subtraction of background det probability that the reconstructed lepton has a wrong is done for each charge individually. charge-flavorcorrelation, the so-called wrong-tag proba- The rejection of events containing π0 or η will fail in bility (ω). The corrected asymmetry is given by: caseswherethedecayisveryasymmetricandthesecond photon has an energy below the threshold, making the 1 reconstruction of the π0 or η impossible. To properly ACP = 1−2ω(AmCPeas−Abkg−Adet). (2) normalize these components, the veto is removed and, foreachcombinationofthe promptphotonwithanother ThecorrectionA accountsforapossibleasymmetryin det photonintheevent,thediphotonmassm iscalculated. theidentificationefficiencybetweenpositiveandnegative γγ A fit to the π0 and η masses is performed to estimate charged leptons (A ) and a possible asymmetry be- LID the number of these mesons in data and MC. The fit is tween the reconstruction of positive and negative tracks performed in eleven meson momentum bins between 1.4 (A ). A is determined using a B → XJ/ψ(ℓ+ℓ−) track LID and 2.6 GeV and the ratio of data to MC yields is used sample, where the selection efficiencies of positively and 5 ×103 son but rather from one of its decay daughters. We 1 find ω = 0.0431 ± 0.0036; this value is estimated Positive sec from MC and the error based on the precision with 0.8 V Negative which the B → DX and D → Xlν branching fractions e G 0.6 are measured. Misidentified hadrons give the smallest 1 0. contribution and consist of π and K mesons faking a s / 0.4 muon and, to a lesser extent, an electron. The corre- e Entri 0.2 swphoenrdeitnhgewfrraocntigo-ntaogfpmriosbidaebnitliitfiyedisheasdtrimonastiesddfertoemrmMinCed, 0 by studying D∗+ → D0(K−π+)π+ decays. After ap- plying the same selection criteria for π and K candi- -0.2 dates as for tag leptons, the fraction of hadrons pass- 2 3 4 E*γ (GeV) ing the selection in the MC is corrected and we obtain ω =0.0069±0.0034. The total wrong-tagprobabil- misID ity value is ω =0.1332±0.0052. FIG. 3: The photon energy spectrum in the CM system af- tersubtractingallthebackground,withverticaldashedlines The asymmetries Adet and Abkg are the dominant showing the signal region. The positive tagged events are uncertainties on ACP and are additive. An additional shown as circles and the negative as squares. Statistical and multiplicative systematic uncertainty arises from the systematic uncertainties are included. wrong-tag probability, leading to a relative uncertainty ∆A /A = 0.01, much less than the additive uncer- CP CP tainties. negatively charged electrons and muons are determined Finally, as some background events remain in the low by performing fits to the invariant dilepton mass mℓℓ energy range after subtraction, we scale the BB¯ compo- for singly- and doubly-identified lepton candidates. The nent to match the data yield below 1.7 GeV and recal- ε+−ε− culate A . The difference between this value and the asymmetry is calculated as: A = . This CP LID ε++ε− nominalistakenasanadditionalsystematicuncertainty. measurement is performed in the full kinematic region, In Table I, the measured and corrected values of A CP in eleven laboratory-frame momentum bins and eight ∗ are summarized for 0.1 GeV steps in the E threshold γ polar-angle bins. The asymmetries for electrons and ∗ from1.7to2.2GeV,withaE upperboundof2.8GeV. γ muons are measured to be A (e) = (0.26 ± 0.14)% LID The statistical and systematic uncertainties are summa- and A (µ)=(−0.03±0.03)%, and averageto A = LID LID rized in Table II. The statistical precisionis improvedin (0.11±0.07)%. The asymmetry A is measured with track comparison to previous measurements [5, 6]; it is, how- partially and fully reconstructed D∗ with D∗ → πD0, ever, the limiting factor in the measurement and is af- D0 → ππK0, K0 → π+π− decays, to be A = S S track fectedbythe sizeofthe continuumsample. As anexam- (−0.01±0.21)%. The total detector-related asymmetry ple, for the 1.7 GeV threshold, the total 4.4% statistical is A =(0.10±0.22)%. det uncertaintyincorporatesa3.0%contributionfromΥ(4S) ∗ We measure Abkg in the low-energy sideband Eγ ≤ data and 3.1% from off-resonance data. The dominant 1.7 GeV. The asymmetries measured in data and MC systematic uncertaintyarisesfromthe asymmetryin the are Abkg(data) = (−0.14 ± 0.78)% and Abkg(MC) = BB¯ background. The asymmetry is consistent with zero (−0.26±0.21)%, which are consistent with zero within across the different photon energy thresholds. uncertainties. TheasymmetryintheBB¯ dataistakenas In conclusion, we have measured the direct CP asym- a correction to ACP. Since this is an asymmetry in the metry ACP(B¯ → Xs+dγ). The measurement is per- BB¯ background,thecorrectionisproportionaltothe ra- formedusing(772±11)×106BB¯ pairsforphotonenergy tio of BB¯ to signalevents in the signalregion,the ratios thresholds between 1.7 and 2.2 GeV. As a nominal re- are taken from MC simulation. sult we choose the 2.1 GeV threshold since it has a low The wrong-tag probability has contributions from uncertainty and keeps a large fraction of signal events: B0B¯0 oscillations (ω ), secondary leptons (ω ) and A (B¯ →X γ)=(2.2±3.9±0.9)%, consistent with osc sec CP s+d misidentified hadrons (ω ) and is given by ω = the SM prediction. This is the first Belle measurement misID ω + ω + ω . The oscillation term is equal to of this asymmetry and the most precise to date. osc sec misID the product of the mixing probability in the B0B¯0 sys- We thank the KEKB group for excellent operation tem χ =0.1875±0.0020 [18], the fraction of neutral B of the accelerator; the KEK cryogenics group for effi- d mesons from the Υ(4S) decay, f = 0.487±0.006 [18], cientsolenoidoperations;andtheKEKcomputergroup, 00 and the fraction of leptons coming directly from a B de- the NII, and PNNL/EMSL for valuable computing and cay, which is estimated to be 91.1% from MC, result- SINET4 network support. We acknowledge support ing in ω = 0.0832± 0.0015. Secondary leptons are fromMEXT,JSPSandNagoya’sTLPRC(Japan);ARC osc true leptons that do not come directly from a B me- (Australia); FWF (Austria); NSFC (China); MSMT 6 TABLE I: CP asymmetry, in percent, for different photon energy thresholds, the E∗ kinematic limit is 2.8 GeV. For the γ measured asymmetry, only the statistical uncertainty is shown; for the corrected asymmetry ACP, statistical and systematic uncertainties are given. The ratio B/S representes the ratio in the number of BB¯ to signal events that is used to scale the asymmetryA . Anadditionalsystematicuncertaintyrelatedtothewrong-tagprobabilityisnotexplicitlylistedbutistaken bkg into account in thetotal uncertainty;its relative value is 1%. Thesystematic contributions are added in quadrature. Eγ∗(thresh.) AmCPeas B/S Abkg Adet MC stats. BB¯ norm. ACP 1.7 GeV 1.3±3.1 3.20 −0.4±2.5 0.1±0.2 ±0.8 ±0.5 2.2±4.3±3.5 1.8 GeV 2.0±3.0 2.41 −0.3±1.9 0.1±0.2 ±0.7 ±0.1 3.0±4.1±2.7 1.9 GeV 0.9±2.9 1.70 −0.2±1.3 0.1±0.2 ±0.6 ±0.3 1.4±4.0±1.9 2.0 GeV 1.6±2.8 1.10 −0.2±0.9 0.1±0.2 ±0.5 ±0.0 2.2±3.8±1.3 2.1 GeV 1.6±2.9 0.65 −0.1±0.5 0.1±0.2 ±0.4 ±0.1 2.2±3.9±0.9 2.2 GeV 1.1±2.9 0.38 −0.1±0.3 0.1±0.2 ±0.3 ±0.2 1.4±3.9±0.6 TABLEII:AbsoluteuncertaintiesinACP,inpercent. Thesystematicuncertaintiesareaddedinquadraturetoyieldthetotal. E∗ thresh. Statistical Total systematic A A MC stat. BB¯ norm. Wrong tag γ det bkg 1.70 GeV 4.26 3.52 0.30 3.40 0.76 0.42 0.02 1.80 GeV 4.13 2.72 0.30 2.56 0.68 0.53 0.05 1.90 GeV 3.96 1.92 0.30 1.81 0.58 0.10 0.02 2.00 GeV 3.84 1.32 0.30 1.17 0.48 0.19 0.04 2.10 GeV 3.91 0.86 0.30 0.70 0.39 0.12 0.04 2.20 GeV 3.89 0.59 0.30 0.41 0.30 0.04 0.03 (Czechia); CZF,DFG,andVS (Germany);DST (India); 93, 031803 (2004). INFN (Italy); MOE, MSIP, NRF, GSDC of KISTI, and [9] J.P.Leesetal.[BaBarCollaboration], Phys.Rev.D90, BK21Plus (Korea); MNiSW and NCN (Poland); MES 092001 (2014). [10] H. Y.Cheng, Phys.Rev.D 51, 6228 (1995). andRFAAE(Russia);ARRS(Slovenia);IKERBASQUE [11] S. Kurokawa and E. Kikutani, Nucl. Instrum. Methods and UPV/EHU (Spain); SNSF (Switzerland); NSC and Phys. Res. Sect. A 499, 1 (2003), and other papers in- MOE (Taiwan); and DOE and NSF (USA). cluded in this Volume; T.Abe et al., Prog. Theor. Exp. Phys. 2013, 03A001 (2013) and following articles up to 03A011. [12] A. Abashian et al., Nucl. Instrum. Meth. A 479, 117 (2002); also see detector section in J. Brodzicka et al., [1] P. Gambino and M. Misiak, Nucl. Phys. B 611, 338 Prog. Theor. Exp. Phys.2012, 04D001 (2012). (2001). [13] A.L.KaganandM.Neubert,Eur.Phys.J.C7,5(1999). [2] M.Benzke,S.J.Lee,M.NeubertandG.Paz,Phys.Rev. [14] P. Koppenburg et al. [Belle Collaboration], Phys. Rev. Lett.106, 141801 (2011). Lett. 93, 061803 (2004). [3] T. Hurth,E. Lunghi and W. Porod, Nucl. Phys. B 704, [15] G. C. Fox and S. Wolfram, Phys. Rev. Lett. 41, 1581 56 (2005). (1978). [4] A. L. Kagan and M. Neubert, Phys. Rev. D 58, 094012 [16] E. Barberio, B. van Eijk and Z. Was, Comput. Phys. (1998). Commun. 66, 115 (1991). [5] T.E.Coanetal.[CLEOCollaboration],Phys.Rev.Lett. [17] K. Nishimura et al. [Belle Collaboration], Phys. Rev. 86, 5661 (2001). Lett. 105, 191803 (2010). [6] J.P.Leesetal.[BaBarCollaboration], Phys.Rev.D86, [18] J.Beringeretal.[ParticleDataGroup],Phys.Rev.D86, 112008 (2012). 010001 (2012) and 2013 partial update for the 2014 edi- [7] A.Limosanietal.[BelleCollaboration], Phys.Rev.Lett. tion. 103, 241801 (2009). [8] S. Nishida et al. [Belle Collaboration], Phys. Rev. Lett. 7