1 QCD Description of Hadronic Tau Decays 1 0 A. Picha 2 aDepartament de F´ısica Teo`rica, IFIC, Univ. Val`encia–CSIC, Apt. 22085, E-46071Val`encia, Spain n a J The QCD analysis of hadronic τ decays is reviewed and a summary of the present phenomenological status is 1 presented. The following topics are discussed: the determination of αs(m2τ)=0.338±0.012 from the inclusive τ 1 hadronicwidth,themeasurement of|Vus|throughtheCabibbo-suppressed decaysoftheτ,andtheextractionof chiral-perturbation-theory couplings from thespectral tau data. ] h p - p 1. THEORETICAL FRAMEWORK invariant mass s of the final-state hadrons: e h The hadronic τ decays turn out to be a beau- m2τ ds s 2 [ R = 12π 1− tiful laboratory for studying strong interaction τ m2 m2 1 Z0 τ (cid:18) τ(cid:19) effects at low energies [1–3]. The τ is the s v × 1+2 ImΠ(1)(s)+ImΠ(0)(s) . (3) 7 only known lepton massive enough to decay into m2 0 hadrons. Its semileptonic decays are then ideally (cid:20)(cid:18) τ(cid:19) (cid:21) The appropriate combinations of correlators are 1 suited to investigate the hadronic weak currents. 2 The inclusive character of the total τ hadronic Π(J)(s) ≡ |V |2 Π(J) (s)+Π(J) (s) 1. width renders possible an accurate calculation of ud ud,V ud,A 0 the ratio [4–8] + |V |2 (cid:16)Π(J) (s)+Π(J) (s)(cid:17). (4) 1 us us,V us,A 1 Γ[τ− →ντhadrons] Thecontributions(cid:16)comingfromthefirstt(cid:17)woterms v: Rτ ≡ Γ[τ− →ντe−ν¯e] = Rτ,V+Rτ,A+Rτ,S. correspond to Rτ,V and Rτ,A respectively, while i R contains the remaining Cabibbo-suppressed X τ,S The theoretical analysis involves the two-point contributions. ar correlationfunctionsforthevector Viµj =ψ¯jγµψi The integrand in Eq. (3) cannot be calculated and axial-vector Aµ = ψ¯ γµγ ψ colour-singlet atpresentfromQCD.Neverthelesstheintegralit- ij j 5 i quark currents (i,j =u,d,s): selfcanbecalculatedsystematicallybyexploiting the analytic propertiesofthe correlatorsΠ(J)(s). Πµν (q)≡i d4xeiqxh0|T(Jµ(x)Jν(0)†)|0i,(1) They are analytic functions of s except along the ij,J ij ij positive real s-axis, where their imaginary parts Z havediscontinuities. R canthen be writtenas a which have the Lorentz decompositions τ contour integral in the complex s-plane running Πµijν,J(q) = −gµνq2+qµqν Π(ij1,)J(q2) counter-clockwise around the circle |s|=m2τ [6]: (cid:0)+qµqνΠ(0) (q2)(cid:1), (2) ds s 2 ij,J R = 6πi 1− τ m2 m2 where the superscript (J = 0,1) denotes the an- I|s|=m2τ τ (cid:18) τ(cid:19) s s gular momentum in the hadronic rest frame. × 1+2 Π(0+1)(s)−2 Π(0)(s) .(5) The imaginary parts of Π(J) (q2) are pro- (cid:20)(cid:18) m2τ(cid:19) m2τ (cid:21) ij,J portional to the spectral functions for hadrons This expression requires the correlators only with the corresponding quantum numbers. The for complex s of order m2, which is signifi- τ hadronic decay rate of the τ can be written as cantly larger than the scale associated with non- an integral of these spectral functions over the perturbativeeffects. UsingtheOperatorProduct 1 2 Expansion (OPE), Π(J)(s) = C(J)/(−s)D/2, firmed by ALEPH [17], CLEO [18] and OPAL D D to evaluate the contour integral, R can be ex- [19]. The most recent analysis gives [20] τ pressed as an expansion in powPers of 1/m2. The τ δ = −0.0059±0.0014. (8) uncertainties associatedwith the use of the OPE NP near the time-like axis are heavily suppressed by The QCD prediction for R is then com- τ,V+A the presence in (5) of a double zero at s=m2. pletelydominatedbyδ ;non-perturbativeeffects τ P In the chiral limit (m = 0), the vector being smaller than the perturbative uncertain- u,d,s and axial-vector currents are conserved. This ties from uncalculated higher-order corrections. implies sΠ(0)(s) = 0. Therefore, only the cor- Assuming lepton universality, the measured val- relator Π(0+1)(s) contributes to Eq. (5). Since ues of the τ lifetime and leptonic branching ra- (1 − x)2(1+ 2x) = 1− 3x2 +2x3 [x ≡ s/m2], tios imply R = 3.6291±0.0086 [21]. Subtract- τ τ Cauchy’s theorem guarantees that, up to tiny ing the Cabibbo-suppressed contribution R = τ,S logarithmic running corrections, the only non- 0.1613 ± 0.0028 [21], one obtains R = τ,V+A perturbative contributions to the circle integra- 3.4678±0.0090. Using |V |=0.97425±0.00022 ud tionin(5)originatefromoperatorsofdimensions [22] and (8), the pure perturbative contribution D = 6 and 8. The usually leading D = 4 opera- to R is determined to be: τ torscanonlycontributetoR withanadditional τ δ =0.1997±0.0035. (9) suppression factor of O(α2), which makes their P s effect negligible [6]. The predicted value of δ turns out to be very P sensitive to α (m2), allowing for an accurate de- s τ α 2. DETERMINATION OF s termination of the fundamental QCD coupling [5,6]. The calculation of the O(α4) contribu- s TheCabibbo-allowedcombinationRτ,V+A can tion [12] has triggered a renewed theoretical in- be written as [6] terest on the α (m2) determination, since it al- s τ lows to push the accuracy to the four-loop level. R = N |V |2S {1+δ +δ }, (6) τ,V+A C ud EW P NP However,asshowninTable1,therecenttheoret- ical analyses slightly disagree on the final result. whereN =3isthenumberofquarkcoloursand C ThedifferencesarelargerthantheclaimedO(α4) SEW = 1.0201±0.0003 contains the electroweak s accuracy and originate in the different inputs or radiative corrections [9–11]. The dominant cor- theoretical procedures which have been adopted. rection (∼ 20%) is the perturbative QCD con- tribution δ , which is already known to O(α4) P s 2.1. Perturbative contribution to R τ [6,12]. Quark mass effects [6,13,14] are tiny In the chiral limit, the result is more con- for the Cabibbo-allowed current and amount to veniently expressed in terms of the logarithmic a negligible correction smaller than 10−4 [6,15]. derivative of the two-point correlation function Non-perturbative contributions are suppressed of the vector (axial) current, Π(s)= 1Π(0+1)(s), by six powers of the τ mass [6] and, therefore, 2 which satisfies an homogeneous renormalization– are very small. Their numerical size has been group equation: determined from the invariant-mass distribution ofthefinalhadronsinτ decay,throughthestudy d 1 αs(−s) n D(s)≡−s Π(s)= K .(10) of weighted integrals [16], ds 4π2 n π n=0 (cid:18) (cid:19) X Rτkl ≡ m2τ ds 1− ms2 k ms2 l ddRsτ , (7) Wallitlhogtahreitchhmoiiccecoofrrreecntoiormnsa,lipzraotpioonrtsiocanlaelµt2o=po−ws- Z0 (cid:18) τ(cid:19) (cid:18) τ(cid:19) ers of log(−s/µ2), have been summed into the whichcanbe calculatedtheoreticallyinthe same running coupling. For three flavours, the known way as R , but are more sensitive to OPE cor- coefficients take the values: K = K = 1; K = τ 0 1 2 rections. The predicted suppression [6] of the 1.63982; K (MS) = 6.37101 and K (MS) = 3 4 non-perturbativecorrectionsto R hasbeencon- 49.07570 [12]. τ 3 Table 1 O(α4) determinations of α (m2). The assumed value of δ is also given (if quoted by the authors). s s τ P Reference Method δ α (m2) α (M2) P s τ s Z Baikov et al. [12] CIPT, FOPT 0.1998±0.0043 0.332±0.016 0.1202±0.0019 Davier et al. [23] CIPT 0.2066±0.0070 0.344±0.009 0.1212±0.0011 Beneke-Jamin [15] BSR + FOPT 0.2042±0.0050 0.316±0.006 0.1180±0.0008 Maltman-Yavin [24] PWM + CIPT — 0.321±0.013 0.1187±0.0016 Menke [25] CIPT, FOPT 0.2042±0.0050 0.342+0.011 0.1213±0.0012 −0.010 Caprini-Fischer [26] BSR + CIPT 0.2042±0.0050 0.320+0.011 — −0.009 Cvetiˇc et al. [27] β + CIPT 0.2040±0.0040 0.341±0.008 0.1211±0.0010 exp Pich [1] CIPT 0.2038±0.0040 0.342±0.012 0.1213±0.0014 Table 2 Exact results for A(n)(α ) (n ≤ 4) at different β-function approximations, and corresponding values of s δ = 4 K A(n)(α ), for a ≡α (m2)/π =0.11. The last row shows the FOPT estimates at O(a4). P n=1 n s τ s τ τ A(1)(α ) A(2)(α ) A(3)(α ) A(4)(α ) δ P s s s s P β =0 0.14828 0.01925 0.00225 0.00024 0.20578 n>1 β =0 0.15103 0.01905 0.00209 0.00020 0.20537 n>2 β =0 0.15093 0.01882 0.00202 0.00019 0.20389 n>3 β =0 0.15058 0.01865 0.00198 0.00018 0.20273 n>4 O(a4) 0.16115 0.02431 0.00290 0.00015 0.22665 τ The perturbative component of R is given by 4 K A(n)(α ), taking a = 0.11. The per- τ n=1 n s τ turbativeconvergenceisverygoodandtheresults δP = KnA(n)(αs), (11) Pare stable under changes of the renormalization n=1 scale. The errorinduced by the truncationof the X where the functions [7] β function at fourth order can be conservatively estimated through the variation of the results at A(n)(α ) = 1 ds αs(−s) n five loops, assuming β5 =±β42/β3 =∓443, i.e. a s 2πi s π geometric growth of the β function. I|s|=m2τ (cid:18) (cid:19) Higher-order contributions to the Adler func- s s3 s4 × 1−2 +2 − (12) tion D(s) will be taken into account adding the m2 m6 m8 (cid:18) τ τ τ(cid:19) fifth-ordertermK A(5)(α )withK =275±400. 5 s 5 arecontourintegralsinthe complexplane,which Moreover, we will include the 5-loop variation only depend on a ≡ α (m2)/π. Using the with changes of the renormalization scale in the τ s τ exact solution (up to unknown β contribu- range µ2/(−s) ∈ [0.5,1.5]. Adopting this very n>4 tions) for α (−s) given by the renormalization- conservativeprocedure,theexperimentalvalueof s group β-function equation, they can be numer- δP given in Eq. (9) implies ically computed with a very high accuracy [7]. α (m2)=0.338±0.012. (13) Table 2 gives the numerical values for A(n)(α ) s τ s (n ≤ 4) obtained at the one-, two-, three- and The result is slightly lower than the one given in four-loop approximations (i.e. β =0, β = Ref. [1], due to the smaller value of δ . n>1 n>2 P 0, β = 0 and β = 0, respectively), to- The strong coupling measured at the τ mass n>3 n>4 gether with the corresponding results for δ = scale is significantly larger than the values ob- P 4 tained at higher energies. From the hadronic de- which is only convergent for a < 0.14. At τ caysoftheZ,onegetsα (M2)=0.1190±0.0027 the four-loop level the radius of convergence is s Z [28], which differs from α (m2) by 18σ. After slightly smaller than the physical value of a . s τ τ evolutionuptothescaleM [29],thestrongcou- Thus, FOPT gives rise to a pathological non- Z pling constant in (13) decreases to convergent series. The long running along the circle makes compulsory to resum the large loga- αs(MZ2) = 0.1209±0.0014, (14) rithms, logn(−s/m2), using the renormalization τ group. This is precisely what CIPT does. in excellent agreement with the direct measure- ments at the Z peak and with a better accuracy. 2.3. Renormalon hypothesis Thecomparisonofthesetwodeterminationsofα s The perturbative expansion of the Adler func- intwoverydifferentenergyregimes,m andM , τ Z tion is expected to be an asymptotic series. If its providesabeautifultestofthe predictedrunning Borel transform, B(t) ≡ K tn/n!, were of the QCD coupling; i.e., a very significant ex- n=0 n+1 well-behaved, one could define D(s) through the perimental verification of asymptotic freedom. P Borel integral 2.2. Fixed-order perturbation theory 1 ∞ The integrals A(n)(α ) can be expanded in D(s)= 1+ dte−πt/αs(s)B(t) . (17) s 4π2 powers of a , A(n)(α ) = an + O(an+1). One (cid:26) Z0 (cid:27) τ s τ τ recovers in this way the naive perturbative ex- However, B(t) has pole singularities at positive pansion [7] (infrared renormalons) and negative (ultravio- let renormalons) integer values of the variable δP = (Kn+gn)anτ ≡ rnanτ . (15) u ≡ −β1t/2, with the exception of u = 1 [31]. nX=1 nX=1 The infrared renormalons at u = +n are related This approximation is known as fixed-order per- to OPE corrections of dimension D = 2n. The turbation theory (FOPT), while the improvedex- renormalon poles closer to the origin dominate pression(11),keepingthenon-expandedvaluesof the large-order behaviour of D(s). A(n)(α ), is usually called contour-improved per- Ithasbeenarguedthat,onceintheasymptotic s turbation theory (CIPT) [7,30]. regime (large n), the renormalonic behaviour As shown in the last row of Table 2, even at of the Kn coefficients could induce cancelations O(a4τ), FOPT gives a rather bad approximation with the running gn corrections, which would be to the integrals A(n)(α ), overestimating δ by missed by CIPT. In that case, FOPT could ap- s P 12% at a = 0.11. The long running of α (−s) proach faster the ‘true’ result provided by the τ s along the circle |s| = m2 generates very large g Borel summation of the full renormalon series τ n coefficients, which depend on K and β (BSR) [15]. This happens actually in the large– m<n m<n [7]: g1 = 0, g2 = 3.56, g3 = 19.99, g4 = 78.00, β1 limit[32,33],whichhoweverdoesnotapproxi- g5 = 307.78. These corrections are much larger mate well the known Kn coefficients. A model of than the original K contributions and lead to higher-order corrections with this behaviour has n valuesofα (m2)smallerthan(13). FOPTsuffers been recently advocated [15]. The model mixes s τ fromalargerenormalization-scaledependence[7] three different types of renormalons (n = −1, 2 and its actual uncertainties are much largerthan and3)plusalinearpolynomial. Itcontains5free usually estimated [25]. parameters which are determined by the known The origin of this bad behaviour can be un- values of K1,2,3,4 and the assumption K5 = 283. derstood analytically at one loop [7]. In FOPT One gets in this way a larger δP, implying a one makes within the contour integral the series smaller value for αs(m2τ). The result looks how- expansion (log(−s/m2)=iφ, φ∈[−π,π]) ever model dependent [34]. τ The implications of a renormalonic behaviour n αs(−s) aτ i have been put on more solid grounds, using an ≈ ≈a β a φ ,(16) τ 1 τ π 1−iβ a φ/2 2 optimal conformal mapping in the Borel plane, 1 τ n (cid:18) (cid:19) X 5 whichachievesthebestasymptoticrateofconver- a different estimate of the non-perturbative con- gence,andproperlyimplementing the CIPTpro- tributions. Unfortunately, Ref. [24] does not cedurewithintheBoreltransform[26]. Assuming quote any explicit values for δ and δ . From NP P thattheknownfourth-orderseriesisalreadygov- the information given in that reference, I deduce erned by the u=−1 and u=2 renormalons,the δ = 0.012±0.018. Although compatible with NP conformal mapping generates a full series expan- (8), the central value is larger and has the op- sion (K = 256, K = 2929 ...) which results, posite sign. This shift implies a smaller δ and, 5 6 P afterBorelsummation,inalargervalueofδ ;i.e. therefore, a slightly smaller strong coupling. P theK termsgiveapositivecontributiontoδ The so-called duality violation effects, i.e. the n>4 P implying a smaller α (m2) [26]. uncertainties associated with the use of the OPE s τ Renormalons provide an interesting guide to to approximate the exact correlator, have been possible higher-order corrections, making appar- also investigated [23,35]. Owing to the presence ent that the associated uncertainties have to be in (5) of a double zero at s = m2, these effects τ carefully estimated. However,one shouldkeep in arequitesuppressedinR . Theyaresmallerthan τ mindthe adoptedassumptions. Infact,thereare the errorsinducedbyδ ,whichareinturnsub- NP no visible signs of renormalonic behaviour in the dominantwithrespecttotheleadingperturbative presently known series: the n = −1 ultraviolet uncertainties. renormalon is expected to dominate the asymp- toticregime,implyinganalternatingseries,while 3. |Vus| DETERMINATION allknownK coefficientshavethesamesign. One n The separate measurement of the |∆S| = 0 could either assume that renormalons only be- and |∆S| = 1 tau decay widths provides a very come relevant at higher orders, for instance at cleandeterminationofV [36,37]. To a firstap- n = 7, and apply the conformal mapping with us proximation the Cabibbo mixing can be directly arbitrary input values for K and K . Differ- 5 6 obtainedfromexperimentalmeasurements,with- entassumptionsaboutthese twounknowncoeffi- out any theoretical input. Neglecting the small cients would result in different central values for α (m2). SU(3)-breaking corrections from the ms − md s τ quark-mass difference, one gets: A different reshuffling of the perturbative se- ries,notrelatedtorenormalons,hasbeenrecently 1/2 R proposed [27]. Instead of the usual expansion in |V |SU(3) = |V | τ,S = 0.210±0.002. us ud powers of the strong coupling, one expands in (cid:18)Rτ,V+A(cid:19) termsoftheβ functionanditsderivatives(β ), exp The new branching ratios measured by BaBar which effectively results in a different estimate of and Belle are all smaller than the previous world higher-order corrections. One gets in this way a averages, which translates into a smaller value weaker dependence on the renormalization scale of R and |V |. For comparison, the previ- and a value of α (m2) similar to the standard τ,S us s τ ous value R = 0.1686± 0.0047 [20] resulted CIPT result. τ,S in |V |SU(3) =0.215±0.003. us This rather remarkable determination is only 2.4. Non-perturbative corrections slightly shifted by the small SU(3)-breaking con- Atthepresentlyachievedprecision,oneshould tributions induced by the strange quark mass. worry about the small non-perturbative correc- These effects can be estimated through a QCD tions. In fact, a proper definition of the infrared analysis of the differences [13,14,36–43] renormalon contributions is linked to the corre- sponding OPE corrections with D = 2n. A re- Rkl Rkl cent re-analysis of the ALEPH data [24], with δRkl ≡ τ,V+A − τ,S . (18) τ |V |2 |V |2 ud us pinched-weight moments of the hadronic distri- bution (PWM) and CIPT, obtains α (m2) = The onlynon-zerocontributionsareproportional s τ 0.321± 0.013. This smaller value originates in to the mass-squared difference m2 − m2 or to s d 6 vacuumexpectationvaluesofSU(3)-breakingop- malized difference between the two sets of mea- erators such as δO ≡ h0|m s¯s − m d¯d|0i ≈ surements is −1.36σ. Thus,the result(20)could 4 s d (−1.4 ± 0.4)· 10−3 GeV4 [13,36]. The dimen- easily fluctuate in the near future. In fact, com- sions of these operators are compensated by cor- bining the measured Cabibbo-suppressed τ dis- responding powers of m2, which implies a strong tribution with electroproduction data, a slightly τ suppression of δRkl [13]: larger value of |V | is obtained [46]. τ us The final error of the V determination from m2(m2) us δRkl ≈ 24S s τ 1−ǫ2 ∆ (α ) τ decay is dominated by the experimental uncer- τ EW (cid:26) m2τ d kl s tainties. IfRτ,S ismeasuredwitha 1%precision, δO (cid:0) (cid:1) the resulting V uncertainty will get reduced to −2π2 4 Q (α ) , (19) us m4 kl s around 0.6%, i.e. ±0.0013, making τ decay the τ (cid:27) best source of information about V . us where ǫ ≡ m /m = 0.053± 0.002 [44]. The d d s An accurate measurement of the invariant- perturbativecorrections∆ (α )andQ (α )are kl s kl s massdistributionofthefinalhadronscouldmake known to O(α3) and O(α2), respectively [13,14]. s s possible a simultaneous determination of V us The J = 0 contribution to ∆ (α ) shows a 00 s and the strange quark mass, through a corre- rather pathologicalbehaviour,with clear signs of latedanalysisofseveralweighteddifferencesδRkl. τ being a non-convergent perturbative series. For- However, the extraction of m suffers from theo- s tunately, the corresponding longitudinal contri- reticaluncertaintiesrelatedtothe convergenceof bution to δR ≡ δR00 can be estimated phe- τ τ the perturbative series ∆ (α ). A better under- kl s nomenologically with a much better accuracy, standing of these corrections is needed. δR |L = 0.1544±0.0037 [36,45], because it is τ dominated by far by the well-known τ → ν π τ 4. CHIRAL SUM RULES and τ → ν K contributions. To estimate the re- τ maining transverse component, one needs an in- When m = 0, the QCD Lagrangian has u,d,s put value forthe strangequarkmass. Takingthe an independent SU(3) flavour invariance for the range ms(mτ) = (100±10)MeV [ms(2GeV) = left and right quark quiralities. The two quirali- (96±10)MeV], which includes the most recent tieshaveexactlythesamestronginteraction,but determinations of ms from QCD sum rules and they are completely decoupled. This chiral in- latticeQCD[45],onegetsfinallyδRτ,th =0.216± variance guarantees that the two-point correla- 0.016 [37], which implies tionfunction ofa left-handedandaright-handed quark currents, Π (s) = Π(0+1)(s)−Π(0+1)(s), 1/2 LR ud,V ud,A Rτ,S vanishes identically to all orders in perturbation |V | = us R|Vτ,uVd+|2A −δRτ,th! tcheievoeryid(etnhteicvaelctpoerratunrdbaaxtiivael-vceocnttorribcuotriroenlas)t.orsTrhee- = 0.2166±0.0019 ±0.0005 . (20) exp th non-zero value of Π (s) originates in the spon- LR A largercentralvalue, |V |=0.2217±0.0032,is taneousbreakingofchiralsymmetrybytheQCD us obtained with the old world average for R . vacuum. At large momenta, the corresponding τ,S Sizeable changes on the experimental determi- OPE only receives contributions from operators nation of R could be expected from the full with dimension d≥6, τ,S analysis of the huge BaBar and Belle data sam- O O ples. In particular, the high-multiplicity decay ΠOPE(s)=− 6 + 8 +··· (21) LR s3 s4 modes are not well known at present. The re- centdecreaseofseveralexperimentaltaubranch- The non-zero up and down quark masses induce ing ratios is also worrisome. As pointed out by tiny corrections with dimensions two and four, the PDG [22], 15 of the 16 branching fractions which are negligible at high energies. measured at the B factories are smaller than the At very low momenta, Chiral Perturbation previous non-B-factory values. The average nor- Theory (χPT) dictates the low-energy expansion 7 ofΠ (s)intermsofthepiondecayconstantand sum rule of Das et al. [53] that determines the LR the χPT couplings L [O(p4)] and C [O(p6)]. pion electromagnetic mass difference. 10 87 Analyticityrelatestheshort-andlong-distance Ref. [52] performs a statistical analysis, scan- regimes through the dispersion relation ning the parameter space (κ,γ,β,s ) and select- z ing those ‘acceptable’ spectral functions which 1 s0 dsw(s)Π (s) = − dsw(s)ρ(s) satisfy the experimental and theoretical con- LR 2πi I|s|=s0 Zsth straints. From a generated initial sample of +2f2w(m2)+Res[w(s)Π (s),s=0] ,(22) 160,000tuples, one finds 1,789acceptable distri- π π LR butions compatiblewithQCDandthe data. The where ρ(s) ≡ π1 ImΠLR(s) and w(s) is an arbi- differencesamongthemdeterminehowmuchfree- traryweightfunctionthatisanalyticinthewhole domisleftforthe behaviourofthe spectralfunc- complex plane except in the origin (where it can tion beyond the kinematical end of the τ data. have poles). The last term in (22) accounts for For each acceptable spectral function one calcu- the possible residue at the origin. lates the parameters L , C , O and O , ob- 10 87 6 8 For s0 ≤ m2τ, the integral along the real axis tained through the dispersion relation (22) with can be evaluated with the measured tau spectral the appropriate weight functions. The resulting functions. Taking w(s)=sn with n≥0, there is statisticaldistributionsdeterminetheirfinallyes- noresidueattheoriginand,withs0 largeenough timated values; the dispersion of the numerical sothattheOPEcanbeappliedintheentirecircle results provides a good quantitative assessment |s|=s0, the OPE coefficients are directly related of the actual uncertainties. to the spectral function integration. With n = 0 The study has been also performed with and 1, there is no OPE contributionin the chiral pinched weight functions of the form w(s) = limitandonegetsthe celebratedfirstandsecond sn(s − s )m (m > 0) that vanish at s = s . z z Weinberg sum rules [47]. For negative values of As expected, these weights are found to mini- the integer n, the OPE does not contribute ei- mize the uncertainties from duality-violation ef- ther while the residues atzero are determined by fects,allowingforamoreprecisedeterminationof theχPTlow-energycouplings,whichcanbethen thehadronicparameters. Onefinallyobtains[52], experimentally determined [48]. Moreover, the absence of perturbative contri- Lr10(Mρ) = −(4.06±0.39)·10−3, butions makes (22) and ideal tool to investigate Cr (M ) = (4.89±0.19)·10−3 GeV−2, 87 ρ possible quark-hadron duality effects, formally O = (−4.3+0.9)·10−3 GeV6, 6 −0.7 defined through [49–51] O = (−7.2+4.2)·10−3 GeV8. (25) 8 −5.3 1 DVw ≡ 2πi dsw(s) ΠLR(s)−ΠOLRPE(s) ThedeterminationofthetwoχPTcouplingsis I|s|=s0 in good agreement with (but more precise than) ∞ (cid:0) (cid:1) recent theoretical calculations, using Resonance = ds w(s) ρ(s). (23) Chiral Effective Theory [54] and large–N tech- Zs0 C niques at the next-to-leading order [55], which ThishasbeenthoroughlystudiedinRef.[52],us- predict [56]: Lr (M ) = −(4.4±0.9)·10−3 and ingforthespectralfunctionbeyonds ∼2.1GeV 10 ρ z Cr (M )=(3.6±1.3)·10−3 GeV−2. Italsoagrees the parametrization [35,49–51] 87 ρ withthepresentlatticeestimatesofLr (M )[57]. 10 ρ ρ(s≥s ) = κ e−γssin(β(s−s )), (24) Duality-violationeffectshaveverylittleimpact z z on the determinationof L and C because the 10 87 and finding the region in the 4-dimensional corresponding sum rules are dominated by the (κ,γ,β,s ) parameter space that is compatible low-energy region where the data sits. Thus, one z with the most recent experimental data [17] and obtains basically the same results with pinched the following theoretical constraints at s → ∞: and non-pinched weight functions. This is no- 0 first and second Weinberg sum sules [47] and the longer true for O and O , which are sensitive to 6 8 8 Ref.@52DH95%L Prades, who sadly passed away recently. Ximo Ref.@52DH68%L has made many relevant contributions to the Almasyetal.,2008 Masjuan&Peris,2007 physics of the tau lepton, some of which have Almasyetal.,2007 been mentioned here. This work has been sup- Bordesetal.,2006 ALEPHColl.,2005 ported by MICINN, Spain (grants FPA2007- Narison,2005 60323 and Consolider-Ingenio 2010 CSD2007- Friotetal.,2004 Zyablyuk,2004 00042, CPAN) by the EU Contract MRTN-CT- Latorre&Rojo,2004 2006-035482 (FLAVIAnet) and by Generalitat Ciriglianoetal.,2003 Bijnensetal.,2001 Valenciana (PROMETEO/2008/069). OPALColl.,1999 Davieretal.,1998 ALEPHColl.,1998 REFERENCES BeforeALEPH(cid:144)OPAL,1992 GeV6 O -12-9 -6 -3 0 6 1000 1. A. Pich, Acta Phys. Polon. Supp. 3 (2010) 165. 2. A. Pich, Nucl. Phys. B (Proc. Suppl.) 186 Ref.@52DH95%L Ref.@52DH68%L (2009) 187; 181-182 (2008) 300; 169 (2007) Almasyetal.,2008 393; Int. J. Mod. Phys. A 21 (2006) 5652. Masjuan&Peris,2007 3. A. Pich, Tau Physics, in Heavy Flavours II, Almasyetal.,2007 Bordesetal.,2006 eds. A.J. Buras and M. Lindner, Advanced ALEPHColl.,2005 Series on Directions in High Energy Physics Narison,2005 15(WorldScientific, Singapore,1998)p.453, Friotetal.,2004 Zyablyuk,2004 arXiv:hep-ph/9704453. Latorre&Rojo,2004 4. E. Braaten, Phys. Rev. Lett. 60 (1988) 1606; Ciriglianoetal.,2003 Bijnensetal.,2001 Phys. Rev. D 39 (1989) 1458. OPALColl.,1999 5. S. Narison and A. Pich, Phys. Lett. B 211 Davieretal.,1998 ALEPHColl.,1998 (1988) 183. GeV8 6. E. Braaten, S. Narison and A. Pich, Nucl. O -20-10 0 10 20 8 1000 Phys. B 373 (1992) 581. 7. F. Le Diberder and A. Pich, Phys. Lett. B 286 (1992) 147. Figure 1. Published results for O and O [52]. 6 8 8. A. Pich, Nucl. Phys. B (Proc. Suppl.) 39B,C (1995) 326. 9. W.J.MarcianoandA.Sirlin,Phys.Rev.Lett. the high-energy behaviour of the spectral func- 61 (1988) 1815. tion; pinched-weightsprovidethenamuchbetter 10. E. Braaten and C.S. Li, Phys. Rev. D 42 accuracy. This could explain the numerical dif- (1990) 3888. ferences among previous estimates [17,19,48,58], 11. J. Erler, Rev. Mex. Phys. 50 (2004) 200. shown in Fig. 1, where duality violation uncer- 12. P.A. Baikov,K.G. ChetyrkinandJ.H. Ku¨hn, tainties were not properly assessed. The results Phys. Rev. Lett. 101 (2008) 012002. (25) fix with accuracythe value of O and deter- 13. A.PichandJ.Prades,JHEP9910(1999)004; 6 mine the sign of O . This information is needed 9806 (1998) 013. 8 to calculate the electromagnetic penguin contri- 14. P.A. Baikov,K.G. ChetyrkinandJ.H. Ku¨hn, bution to the CP-violating ratio ε′ /ε [59]. Phys. Rev. Lett. 95 (2005) 012003. K K 15. M. Beneke and M. Jamin, JHEP 0809(2008) 044. ACKNOWLEDGEMENTS 16. F. Le Diberder and A. Pich, Phys. Lett. B I would like to dedicate this work to the 289 (1992) 165. memory of our friend and collaborator Ximo 17. ALEPHCollaboration,Phys.Rep.421(2005) 9 191;Eur.Phys.J.C4(1998)409;Phys.Lett. varov, Eur. Phys. J. C 20 (2001) 259. B 307 (1993) 209. 41. K. Maltman and C.E. Wolfe, Phys. Lett. B 18. CLEO Collaboration, Phys. Lett. B 356 639 (2006) 283. (1995) 580. 42. J.KamborandK.Maltman,Phys.Rev.D62 19. OPAL Collaboration, Eur. Phys. J. C 7 (2000) 093023;64 (2001) 093014. (1999) 571. 43. K.Maltman, Phys.Rev. D58 (1998)093015. 20. M. Davier, A. Ho¨cker and Z. Zhang, Rev. 44. H. Leutwyler, Phys. Lett. B 378 (1996) 313. Mod. Phys. 78 (2006) 1043. 45. M.Jamin,J.A.OllerandA.Pich,Phys.Rev. 21. Heavy Flavor Averaging Group (HFAG), D 74 (2006) 074009. arXiv:1010.1589[hep-ex]; 46. K.Maltman,arXiv:1011.6391[hep-ph];Phys. http://www.slac.stanford.edu/xorg/hfag/. Lett. B672 (2009) 257. K. Maltman et al., 22. K. Nakamura et al., Review of Particle Nucl. Phys. B (Proc. Suppl.) 189 (2009) 175. Physics, J. Phys. G 37 (2010) 075021. 47. S. Weinberg, Phys. Rev. Lett. 18 (1967) 507. 23. M. Davier et al., Eur. Phys. J. C 56 (2008) 48. M. Davier et al., Phys. Rev. D 58 (1988) 305. 096014. 24. K. Maltman and T. Yavin, Phys. Rev. D 78 49. M. A. Shifman, arXiv:hep-ph/0009131. (2008) 094020. 50. O. Cat`a, M. Golterman, and S. Peris, JHEP 25. S. Menke, arXiv:0904.1796[hep-ph]. 0508 (2005) 076. 26. I. Caprini and J. Fischer, Eur. Phys. J. C 64 51. M. Gonz´alez-Alonso, Val`encia Univ. Master (2009) 35. Thesis (2007). 27. G. Cvetic et al., Phys. Rev. D 82 (2010) 52. M. Gonz´alez-Alonso, A. Pich, and J. Prades, 093007. Phys. Rev. D 82 (2010) 014019; 81 (2010) 28. The ALEPH, CDF, D0, DELPHI, L3, 074007;78 (2008) 116012. OPAL, SLD Collaborations, the LEP and 53. T. Das etal., Phys.Rev. Lett. 18(1967)759. Tevatron Electroweak Working Groups, 54. G.Eckeretal.,Nucl.Phys.B321(1989)311; and the SLD Electroweak and Heavy Phys. Lett. B 223 (1989) 425. Flavour Groups, arXiv:1012.2367 [hep-ex]; 55. A. Pich, arXiv:hep-ph/0205030. http://www.cern.ch/LEPEWWG/. 56. A.Pich,I.RosellandJ.J.Sanz-Cillero,JHEP 29. G. Rodrigo, A. Pich and A. Santamaria, 0807 (2008) 014. Phys. Lett. B 424 (1998) 367. 57. E.Shintanietal.,Phys.Rev.Lett.101(2008) 30. A.A. Pivovarov,Z. Phys. C 53 (1992) 461. 242001. P.A. Boyle et al., Phys. Rev. D 81 31. M. Beneke, Phys. Rept. 317 (1999) 1. (2010) 014504. 32. P. Ball, M. Beneke and V.M. Braun, Nucl. 58. J.Bijnensetal.,JHEP0110(2001)009.K.N. Phys. B 452 (1995) 563. Zyablyuk, Eur. Phys. J. C 38 (2004) 215. 33. M. Neubert, Phys. B 463 (1996) 511. J. Rojo and J.I. Latorre, JHEP 0401 (2004) 34. S. Descotes-Genon and B. Malaescu, 055.S.Narison,Phys.Lett.B624(2005)223. arXiv:1002.2968[hep-ph]. S. Peris et al., Phys. Rev. Lett. 86 (2001) 14. 35. O. Cat`a, M. Golterman and S. Peris, Phys. S. Friot et al., JHEP 0410 (2004) 043. A.A. Rev. D 79 (2009) 053002;77 (2008) 093006. Almasy et al., Phys. Lett. B 650 (2007) 179; 36. E. G´amiz et al., Phys. Rev. Lett. 94 (2005) Eur.Phys.J.C55(2008)237.P.Masjuanand 011803;JHEP 0301 (2003) 060. S. Peris, JHEP 05 (2007) 040. V. Cirigliano 37. E. G´amiz et al., PoS KAON 008 (2007). et al., Phys. Lett. B 555 (2003) 71; Phys. 38. S. Chen et al., Eur. Phys. J. C 22 (2001) 31. Rev. D 68 (2003) 054013. C.A. Dom´ınguez M.Davieretal.,Nucl.Phys.B(Proc.Suppl.) and K. Schilcher, Phys. Lett. B 581 (2004) 98 (2001) 319. 193. J. Bordes et al., JHEP 0602 (2006) 037. 39. K.G. Chetyrkin, J.H. Ku¨hn and A.A. Pivo- 59. M. Gonz´alez-Alonso, A. Pich and J. Prades, varov, Nucl. Phys. B 533 (1998) 473. to appear. 40. J.G. K¨orner, F. Krajewski and A.A. Pivo-