Lattice Operator Product Expansion and the Structure Functions C.-J. David Lin CTS-NTU NTHU, Hsinchu 21/09/06 Based on W. Detmold and C-JDL, Phys Rev D73, 014501 (2006). Outline • Hadronic tensor and the OPE. • The OPE on the lattice. • Extracting moments from lattice data. • Application to the pion distribution amplitude. • Conclusion. Deeply Inelastic Lepton-Hadron Scattering, I A good review: Aneesh Manohar, hep-ph/9204208. k', E' k, E q = k - k' X p Dimensionless hadronic tensor WSµν (p, q) → decomposed into the structure functions F1,2 and g1,2 . p, λ ν µ q q p, λ' Dimensionless hadronic tensor TSµν (p, q) → decomposed into the structure functions F̃1,2 and g̃1,2 . • Optical theorem relates Imag of F̃ , g̃ to F , g. • The structure functions are functions of x = −q 2 /2p · q and −q 2/p2 . Deeply Inelastic Lepton-Hadron Scattering, II • DIS is the study of the regime −q 2/p2 → ∞ at fixed x = −q 2/2p · q. → Asymptotic freedom of QCD. → The Bjorken scaling. • Physical region is 0 ≤ x ≤ 1. • The DIS regime can be shown to be dominated by the structure of the hadron along the light-cone. → Difficult for field theory formulated in Euclidean space. • Can perform an operator product expansion. → A short distance expansion, works at x → ∞. → Extract information, i.e., moments of the structure functions, in this unphysical region. Hadronic tensor and the OPE WSµν (p, q) = | Z d4 z eiq·z hp, S| [J µ (z), J ν (0)] |p, Si {z } Imaginary part of TSµν (p, q) = | Z d4z eiq·z hp, S|T [J µ (z)J ν (0)] |p, Si {z } T [J µ (z)J ν (0)] = the OPE P Ci (z 2, µ2)zµ1 . . . zµn Oiµνµ1...µn (µ) twist = dimension - spin Powers of 1/x in the matrix elements. Lattice calculations • Analytic continuation → Difficult to obtain TSµν directly. • Operator mixing and renormalisation → Difficult for high-spin operators. The OPE on the lattice General features µ ν hp, S|T [J (z)J (0)] |p, Si = | {z Simulation } X | Ci (z 2, µ2 ) zµ1 . . . zµn hp, S|Oiµνµ1 ...µn (µ)|p, Si } | {z Analytical calculation {z Fits • First investigated in kaon physics. C. Dawson et al., 1998. Our proposal • Simulation of P S TSµν → Continuum limit. → No power divergence. • Perform the OPE in Euclidean space • Fit the matrix elements. → No need for analytic continuation. → No need for operator matching. • Not obtaining TSµν in Minkowski space directly. } The OPE on the lattice Specific features • A fictitious “valence” heavy quark Ψ and current µ (z) = Ψ̄(z)γ µ ψ(z) + ψ̄(z)γ µ Ψ(z). JΨ,ψ • Study the Euclidean Compton scattering tensor Z h E i D X µ µν ν 4 iq·z d z e TΨ,ψ = p, S T JΨ,ψ (z)JΨ,ψ (0) p, S S • Sum the target-mass effects. • Compute the twist-two matrix elements. Why a “valence” heavy quark? • Two large scales, q 2 and mΨ . p ΛQCD ≪ mΨ ∼ q 2 ≪ 1 a • Remove many higher-twist contributions. • No all-to-all propagator in the simulation. • The Fourier transform is practical → z4 ∼ 1/mΨ . The OPE on the lattice Some details i(iD /+q/)+mΨ (iD+q)2 +m2Ψ /+q/)+mΨ = − i(iD Q2 +D 2 −m2 Q2 = −q 2, Ψ P∞ n=0 Higher-twist components n −2i q·D Q2 +D 2 −m2Ψ MΨ = mψ + α/2 2 + αM + β Q2 + D2 − m2Ψ = Q̃2 = Q2 − MΨ Ψ µν µν . − TΨ,d No contribution in TΨ,u Removed because Ψ is non-dynamical. Λ2QCD q 2 +m2Ψ Extrating moments from data The Euclidean Compton tensor X {µ p, S ψ̄γ 1 (iDµ2 ) . . . S X S {µν} p, S ψ̄ TΨ,ψ (p, q) = iD i {µ1 ∞ X n=2 even iD µn } . . . iD µn } − tr p, S − tr p, S = Anψ (µ2 ) [pµ1 . . . pµn − tr] = Ânψ (µ2 ) [pµ1 . . . pµn − tr] (2) (3) Anψ (µ2 ) ζ n F Cn(1) (η), Cn−1 (η), Cn−2 (η), n, q 2 , Q̃2 , µ2 ∞ M (mΨ − m) µν X Cˆn Ânψ (µ2 ) ζ n Cn(1) (η) δ −2i 2 Q̃ n=0 even • The Gegenbauer polynomial: target-mass effects. η = √p·q2 p q2 2 is the large scale for the OPE. • Q̃2 ∼ −q 2 − MΨ √22 pq ζ = Q̃2 • Remove Ânψ by choosing µ 6= ν. Extracting moments from data Plots {34} TΨ,ψ (p, q) ∞ X = Anψ (µ2 )f (n) n=2,even p = (0, 0, 0, iM ) q = (0, 0, , p q02 − Q2 , iq0 ) M = 1.2 GeV 0 fHnL -0.1 -0.2 Α = 0.4 GeV Α = 1.0 GeV Α = 1.2 GeV -0.3 -0.4 -0.5 MQ = 3.54 GeV, q0 = 2.76 GeV, Q2 = 1.5 GeV2 0 5 10 n 15 20 M = 1.2 GeV 0 fHnL -0.1 -0.2 -0.3 -0.4 MQ = 2.1 GeV -0.5 q0 = 1.98 GeV, Q2 = -3.85 GeV2 0 5 10 n 15 20 Extracting moments from data Correlators µν µν No contribution for TΨ,u − TΨ,d Additional contractions if light-light current is used. µν µν − TΨ,d Contributes to TΨ,u Application to the Pion Distribution Amplitude ¯ Related to the matrix element hπ|T d(z)γµ γ5u(0) |0i. • A crucial input in B → ππ decays via QCD factorisation. • The OPE leads to the need of the matrix elements µ µ {µ n} 2 1 0 . π(p) ψ̄γ γ5 (iD ) . . . iD • These can be obtained by applying the OPE to the “unphysical” matrix element h i E D µ ν π(p) T VΨ,ψ (z)AΨ,ψ (0) 0 . Conclusion • Extract moments via the OPE on the lattice. • Can be applied to other nucleon structure functions. • Can be applied to the pion distribution amplitude • Numerical work is being carried out.