Introduction
Transverse polarization of hyperons produced in unpolarized collisions have been observed in many high-energy inclusive processes such as p p → Λ ↑ X pp \to\Lambda^\uparrow X pp → Λ ↑ X , e + e − → Λ ↑ X e^+e^- \to\Lambda^\uparrow X e + e − → Λ ↑ X and e p → e Λ ↑ X ep \to e\Lambda^\uparrow X e p → e Λ ↑ X 1 . This phenomenon is an example of transverse single spin asymmetries (SSAs) in which only one particle appearing in the process is transversely polarized. The SSAs associated with the initial state spin such as p ↑ p → h X p^\uparrow p \to hX p ↑ p → h X and e p ↑ → e h X ep^\uparrow\to ehX e p ↑ → e h X (h = π , K , η , … h = \pi, K, \eta, \ldots h = π , K , η , … ) have also been observed. The parton model or perturbative QCD at twist-2 level fails to produce large SSAs [1 ] , and its description in terms of QCD has been a challenge in QCD spin physics. In the collinear factorization of perturbative QCD, the SSAs occur as a twist-3 observable, which reflects multi-parton correlations either in the initial nucleon or in the fragmentation processes [2 , 3 , 4 ] . Derivation of the twist-3 cross sections for SSAs required lots of technical development. By now the collinear twist-3 formalism for all kinds of the twist-3 distribution functions (DFs) and the fragmentation functions (FFs) has been well established in the leading order (LO) with respect to the QCD coupling constant [5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 ] . Those functions are, in general, not independent from each other but obey some constraint relations based on the operator identities. The complete set of those relations have been also derived [21 , 22 ] , which are crucial to obtain the frame-independent expressions for the twist-3 cross sections [21 , 22 , 23 ] . There have been also some attempts to extend the twist-3 calculation to the next-to-leading order level [24 , 25 , 26 ] .
In this paper, we study the transverse polarization of hyperons produced in semi-inclusive deep inelastic scattering (SIDIS), e p → e Λ ↑ X ep \to e\Lambda^\uparrow X e p → e Λ ↑ X , in the LO collinear twist-3 factorization. As for the case of p p → Λ ↑ X pp \to\Lambda^\uparrow X pp → Λ ↑ X [20 , 27 , 28 , 29 , 30 ] , two kinds of the twist-3 cross sections contribute. One is from the twist-3 DF in unpolarized proton combined with the “transversity” FF for the polarized Λ \Lambda Λ . The other one is from the twist-3 polarized FFs for Λ \Lambda Λ combined with the unpolarized parton DFs in the proton. In the latter contribution, both twist-3 quark FFs [30 ] and twist-3 gluon FFs [20 ] contribute due to the chiral-even nature of the FFs and DFs. Here we report the LO twist-3 cross section formula for e p → e Λ ↑ X ep \to e\Lambda^\uparrow X e p → e Λ ↑ X from the twist-3 DFs and the twist-3 quark FFs. This is relevant for the large-P T P_T P T polarized hyperon production in the future Electron-Ion-Collider (EIC) experiment. The contribution from twist-3 purely gluonic FFs will be reported in a separate publication [31 ] .
The remainder of this paper is organized as follows: In section 2, we summarize the twist-3 DFs and FFs relevant to the present study. In section 3, after summarizing the kinematics of e p → e Λ ↑ X ep \to e\Lambda^\uparrow X e p → e Λ ↑ X (see § 3.1 ), we present the cross section from the twist-3 DFs (§ 3.2 ) and from the twist-3 quark FFs (§ 3.3 ). Section 4 is devoted to a brief summary of the present study.
Twist-3 quark DFs and FFs
In this section, we summarize the twist-3 distribution function (DF) and the fragmentation functions (FF), which are necessary to calculate the polarized cross section for e p → e Λ ↑ X ep \to e\Lambda^\uparrow X e p → e Λ ↑ X . For the twist-3 distribution, we need only one function E F ( x 1 , x 2 ) E_F(x_1,x_2) E F ( x 1 , x 2 ) in an unpolarized proton, which is defined from the quark-gluon correlation function as [32 ]
M F α ( x 1 , x 2 ) = ∫ d λ 2 π ∫ d μ 2 π e i λ x 1 e i μ ( x 2 − x 1 ) ⟨ p ∣ ψ ˉ j ( 0 ) [ 0 , μ n ] g F α n ( μ n ) [ μ n , λ n ] ψ i ( λ n ) ∣ p ⟩ = − M N 4 ϵ α β n p ( γ 5 γ β p / ) i j E F ( x 1 , x 2 ) + ⋯ , (1) \begin{aligned}
M_{F}^{\alpha}(x_{1},x_{2})=\int\frac{d\lambda}{2\pi}\int\frac{d\mu}{2\pi}e^{i\lambda x_{1}}e^{i\mu(x_{2}-x_{1})}\langle p|\bar{\psi}_{j}(0)[0,\mu n]gF^{\alpha n}(\mu n)[\mu n,\lambda n]\psi_{i}(\lambda n)|p\rangle\\
&=-\frac{M_{N}}{4}\epsilon^{\alpha\beta np}(\gamma_{5}\gamma_{\beta}p\!\!\!/)_{ij}E_{F}(x_{1},x_{2})+\cdots, \tag*{(1)}
\end{aligned} M F α ( x 1 , x 2 ) = ∫ 2 π d λ ∫ 2 π d μ e iλ x 1 e i μ ( x 2 − x 1 ) ⟨ p ∣ ψ ˉ j ( 0 ) [ 0 , μ n ] g F α n ( μ n ) [ μ n , λn ] ψ i ( λn ) ∣ p ⟩ = − 4 M N ϵ α β n p ( γ 5 γ β p / ) ij E F ( x 1 , x 2 ) + ⋯ , (1)
where M N M_{N} M N is the nucleon mass, ∣ p ⟩ |p\rangle ∣ p ⟩ is the nucleon state with momentum p p p which can be regarded as lightlike, n n n is another lightlike vector satisfying p ⋅ n = 1 p\cdot n=1 p ⋅ n = 1 , and i i i , j j j denotes the spinor indices. We use the convention for the ϵ \epsilon ϵ -tensor as ϵ 0123 = 1 \epsilon^{0123}=1 ϵ 0123 = 1 and the notation ϵ α β n p ≡ ϵ α β μ ν n μ p ν \epsilon^{\alpha\beta np}\equiv\epsilon^{\alpha\beta\mu\nu}n_{\mu}p_{\nu} ϵ α β n p ≡ ϵ α β μν n μ p ν is used. [ μ n , λ n ] = P e x p [ i g ∫ λ μ d τ n ⋅ A ( τ n ) ] [\mu n,\lambda n]=\mathop{\mathrm{P\,exp}}\left[ig\int_{\lambda}^{\mu}d\tau\,n\cdot A(\tau n)\right] [ μ n , λn ] = P exp [ i g ∫ λ μ d τ n ⋅ A ( τ n ) ] is the gauge link operator, which makes the correlation function color gauge invariant. From P P P , T T T -invariance, one has E F ( x 1 , x 2 ) = E F ( x 2 , x 1 ) E_{F}(x_{1},x_{2})=E_{F}(x_{2},x_{1}) E F ( x 1 , x 2 ) = E F ( x 2 , x 1 ) . The support of E F ( x 1 , x 2 ) E_{F}(x_{1},x_{2}) E F ( x 1 , x 2 ) is ∣ x 1 , 2 ∣ < 1 |x_{1,2}|<1 ∣ x 1 , 2 ∣ < 1 and ∣ x 1 − x 2 ∣ < 1 |x_{1}-x_{2}|<1 ∣ x 1 − x 2 ∣ < 1 . The anti-quark or “charge-conjugated” distribution E ˉ F ( x 1 , x 2 ) \bar{E}_{F}(x_{1},x_{2}) E ˉ F ( x 1 , x 2 ) for E F E_{F} E F is defined by ψ → C ψ ˉ T \psi\to C\bar{\psi}^{T} ψ → C ψ ˉ T , ψ ˉ → − ψ T C − 1 \bar{\psi}\to-\psi^{T}C^{-1} ψ ˉ → − ψ T C − 1 (C C C is the charge conjugation matrix) and F α β → − F α β T F_{\alpha\beta}\to-F_{\alpha\beta}^{T} F α β → − F α β T in (1) and it satisfies the relation E ˉ F ( x 1 , x 2 ) = E F ( − x 2 , − x 1 ) \bar{E}_{F}(x_{1},x_{2})=E_{F}(-x_{2},-x_{1}) E ˉ F ( x 1 , x 2 ) = E F ( − x 2 , − x 1 ) .
We need several kinds of FFs, which are summarized below using the notation in [21 ] . The simplest ones are defined from the lightcone correlation functions of quark fields:
Δ i j ( z ) = 1 N ∑ X ∫ d λ 2 π e − i λ / z ⟨ 0 ∣ [ ∞ w , 0 ] ψ i ( 0 ) ∣ h ( P h , S ⊥ ) X ⟩ ⟨ h ( P h , S ⊥ ) X ∣ ψ ˉ j ( λ w ) [ λ w , ∞ w ] ∣ 0 ⟩ = ( γ 5 S / ⊥ P / h z ) i j H 1 ( z ) + M h ϵ α S ⊥ w P h ( γ α ) i j D T ( z ) z + M h ( γ 5 S / ⊥ ) i j G T ( z ) z + ⋯ . (2) \begin{aligned}
\Delta_{ij}(z)=\frac{1}{N}\sum_{X}\int\frac{d\lambda}{2\pi}e^{-i\lambda/z}\langle0|[\infty w,0]\psi_{i}(0)|h(P_{h},S_{\perp})X\rangle\langle h(P_{h},S_{\perp})X|\bar{\psi}_{j}(\lambda w)[\lambda w,\infty w]|0\rangle\\
&=\left(\gamma_{5}S\!\!\!/_{\perp}\frac{P\!\!\!/_{h}}{z}\right)_{ij}H_{1}(z)+M_{h}\epsilon^{\alpha S_{\perp}wP_{h}}(\gamma_{\alpha})_{ij}\frac{D_{T}(z)}{z}+M_{h}(\gamma_{5}S\!\!\!/_{\perp})_{ij}\frac{G_{T}(z)}{z}+\cdots. \tag*{(2)}
\end{aligned} Δ ij ( z ) = N 1 X ∑ ∫ 2 π d λ e − iλ / z ⟨ 0∣ [ ∞ w , 0 ] ψ i ( 0 ) ∣ h ( P h , S ⊥ ) X ⟩ ⟨ h ( P h , S ⊥ ) X ∣ ψ ˉ j ( λ w ) [ λ w , ∞ w ] ∣0 ⟩ = ( γ 5 S / ⊥ z P / h ) ij H 1 ( z ) + M h ϵ α S ⊥ w P h ( γ α ) ij z D T ( z ) + M h ( γ 5 S / ⊥ ) ij z G T ( z ) + ⋯ . (2)
where ∣ h ( P h , S ⊥ ) ⟩ |h(P_{h},S_{\perp})\rangle ∣ h ( P h , S ⊥ )⟩ denotes the hyperon state with mass M h M_{h} M h , momentum P h P_{h} P h and the transverse spin vector S ⊥ S_{\perp} S ⊥ normalized as S ⊥ 2 = − 1 S_{\perp}^{2}=-1 S ⊥ 2 = − 1 . Since we are interested in the twist-3 cross section, we treat P h P_{h} P h as lightlike, and w w w is another lightlike vector satisfying P h ⋅ w = 1 P_{h}\cdot w=1 P h ⋅ w = 1 . H 1 ( z ) H_{1}(z) H 1 ( z ) is the twist-2 transversity FF, D T ( z ) D_{T}(z) D T ( z ) and G T ( z ) G_{T}(z) G T ( z ) are twist-3 and are called intrinsic twist-3 FFs. D T ( z ) D_{T}(z) D T ( z ) is naively T T T -odd, while G T ( z ) G_{T}(z) G T ( z ) is naively T T T -even. In (2), gauge link operator [ λ w , ∞ w ] ≡ P e x p [ i g ∫ ∞ λ d τ w ⋅ A ( τ w ) ] [\lambda w,\infty w]\equiv\mathop{\mathrm{P\,exp}}\left[ig\int_{\infty}^{\lambda}d\tau\,w\cdot A(\tau w)\right] [ λ w , ∞ w ] ≡ P exp [ i g ∫ ∞ λ d τ w ⋅ A ( τ w ) ] is inserted, which makes the correlation function gauge invariant.
The next one is the twist-3 kinematical FFs which are defined as
Δ ∂ i j α ( z ) = 1 N ∑ X ∫ d λ 2 π e − i λ / z ⟨ 0 ∣ [ ∞ w , 0 ] ψ i ( 0 ) ∣ h ( P h , S ⊥ ) X ⟩ ⟨ h ( P h , S ⊥ ) X ∣ ψ ˉ j ( λ w ) [ λ w , ∞ w ] ∣ 0 ⟩ ∂ ← α = − i M h ϵ α S ⊥ w P h ( P / h ) i j D 1 T ⊥ ( 1 ) ( z ) z + i M h S ⊥ α ( γ 5 P / h ) i j G 1 T ⊥ ( 1 ) ( z ) z + ⋯ , (3) \begin{aligned}
\Delta_{\partial ij}^{\alpha}(z)=\frac{1}{N}\sum_{X}\int\frac{d\lambda}{2\pi}e^{-i\lambda/z}\langle0|[\infty w,0]\psi_{i}(0)|h(P_{h},S_{\perp})X\rangle\langle h(P_{h},S_{\perp})X|\bar{\psi}_{j}(\lambda w)[\lambda w,\infty w]|0\rangle\overleftarrow{\partial}^{\alpha} \\
&=-iM_{h}\epsilon^{\alpha S_{\perp}wP_{h}}(P\!\!\!/_{h})_{ij}\frac{D_{1T}^{\perp(1)}(z)}{z}+iM_{h}S_{\perp}^{\alpha}(\gamma_{5}P\!\!\!/_{h})_{ij}\frac{G_{1T}^{\perp(1)}(z)}{z}+\cdots, \tag*{(3)}
\end{aligned} Δ ∂ ij α ( z ) = N 1 X ∑ ∫ 2 π d λ e − iλ / z ⟨ 0∣ [ ∞ w , 0 ] ψ i ( 0 ) ∣ h ( P h , S ⊥ ) X ⟩ ⟨ h ( P h , S ⊥ ) X ∣ ψ ˉ j ( λ w ) [ λ w , ∞ w ] ∣0 ⟩ ∂ α = − i M h ϵ α S ⊥ w P h ( P / h ) ij z D 1 T ⊥ ( 1 ) ( z ) + i M h S ⊥ α ( γ 5 P / h ) ij z G 1 T ⊥ ( 1 ) ( z ) + ⋯ , (3)
where each FF is defined to be real, and they are related to k T 2 / M h 2 k_{T}^{2}/M_{h}^{2} k T 2 / M h 2 -moment of the transverse-momentum-dependent (TMD) FFs [33 ] . G 1 T ⊥ ( 1 ) ( z ) G_{1T}^{\perp(1)}(z) G 1 T ⊥ ( 1 ) ( z ) is naively T T T -even, while D 1 T ⊥ ( 1 ) ( z ) D_{1T}^{\perp(1)}(z) D 1 T ⊥ ( 1 ) ( z ) is naively T T T -odd and contributes to the hyperon polarization.
Next we introduce the twist-3 dynamical FFs which are defined from the three parton correlation function as:
Δ F i j α ( z , z 1 ) = 1 N ∑ X ∫ d λ 2 π ∫ d μ 2 π e − i λ / z 1 e − i μ ( 1 / z − 1 / z 1 ) ⟨ 0 ∣ ψ i ( 0 ) ∣ h ( P h , S ⊥ ) X ⟩ ⟨ h ( P h , S ⊥ ) X ∣ ψ ˉ j ( λ w ) g F α w ( μ w ) ∣ 0 ⟩ = M h ϵ α S ⊥ w P h ( P / h ) i j D ^ F T ∗ ( z , z 1 ) z − i M h S ⊥ α ( γ 5 P / h ) i j G ^ F T ∗ ( z , z 1 ) z + ⋯ , (4) \begin{aligned}
\Delta_{Fij}^{\alpha}(z,z_{1})=\frac{1}{N}\sum_{X}\int\frac{d\lambda}{2\pi}\int\frac{d\mu}{2\pi}e^{-i\lambda/z_{1}}e^{-i\mu(1/z-1/z_{1})}\langle0|\psi_{i}(0)|h(P_{h},S_{\perp})X\rangle\langle h(P_{h},S_{\perp})X|\bar{\psi}_{j}(\lambda w)gF^{\alpha w}(\mu w)|0\rangle\\
&=M_{h}\epsilon^{\alpha S_{\perp}wP_{h}}(P\!\!\!/_{h})_{ij}\frac{\widehat{D}_{FT}^{*}(z,z_{1})}{z}-iM_{h}S_{\perp}^{\alpha}(\gamma_{5}P\!\!\!/_{h})_{ij}\frac{\widehat{G}_{FT}^{*}(z,z_{1})}{z}+\cdots, \tag*{(4)}
\end{aligned} Δ F ij α ( z , z 1 ) = N 1 X ∑ ∫ 2 π d λ ∫ 2 π d μ e − iλ / z 1 e − i μ ( 1/ z − 1/ z 1 ) ⟨ 0∣ ψ i ( 0 ) ∣ h ( P h , S ⊥ ) X ⟩ ⟨ h ( P h , S ⊥ ) X ∣ ψ ˉ j ( λ w ) g F α w ( μ w ) ∣0 ⟩ = M h ϵ α S ⊥ w P h ( P / h ) ij z D F T ∗ ( z , z 1 ) − i M h S ⊥ α ( γ 5 P / h ) ij z G F T ∗ ( z , z 1 ) + ⋯ , (4)
where the gauge link is suppressed for simplicity. The dynamical FFs D ^ F T ( z , z 1 ) \widehat{D}_{FT}(z,z_1) D F T ( z , z 1 ) and G ^ F T ( z , z 1 ) \widehat{G}_{FT}(z,z_1) G F T ( z , z 1 ) are complex functions and their complex conjugates are defined in (4). The real parts of these functions are naively T T T -even, while the imaginary parts are naively T T T -odd and contribute to the hyperon polarization. Replacing g F α w ( μ w ) gF^{\alpha w}(\mu w) g F α w ( μ w ) by the covariant derivative D α ( μ w ) = ∂ α − i g A α ( μ w ) D^\alpha(\mu w)=\partial^\alpha-igA^\alpha(\mu w) D α ( μ w ) = ∂ α − i g A α ( μ w ) , one can define another set of the twist-3 FFs, D ^ D T ( z , z 1 ) \widehat{D}_{DT}(z,z_1) D D T ( z , z 1 ) and G ^ D T ( z , z 1 ) \widehat{G}_{DT}(z,z_1) G D T ( z , z 1 ) , by the same tensor decomposition as above. But they can be related to the above functions [17 , 19 ] :
S D ^ D T ( z , z 1 ) = P 1 1 / z − 1 / z 1 S D ^ F T ( z , z 1 ) − δ ( 1 z − 1 z 1 ) D 1 T ⊥ ( 1 ) ( z ) , (5) \mathcal{S}\widehat{D}_{DT}(z,z_1)=P\frac{1}{1/z-1/z_1}\mathcal{S}\widehat{D}_{FT}(z,z_1)-\delta\left(\frac{1}{z}-\frac{1}{z_1}\right)D_{1T}^{\perp(1)}(z), \tag*{(5)} S D D T ( z , z 1 ) = P 1/ z − 1/ z 1 1 S D F T ( z , z 1 ) − δ ( z 1 − z 1 1 ) D 1 T ⊥ ( 1 ) ( z ) , (5)
S G ^ D T ( z , z 1 ) = P 1 1 / z − 1 / z 1 S G ^ F T ( z , z 1 ) . (6) \mathcal{S}\widehat{G}_{DT}(z,z_1)=P\frac{1}{1/z-1/z_1}\mathcal{S}\widehat{G}_{FT}(z,z_1). \tag*{(6)} S G D T ( z , z 1 ) = P 1/ z − 1/ z 1 1 S G F T ( z , z 1 ) . (6)
Three kinds of twist-3 FFs (2), (3) and (4) are not independent, but are subject to the EOM relation and the Lorentz invariance relations (LIRs). Here we quote those relations from [21 ] . The EOM relation involving the naively T T T -odd FFs is given by
∫ z ∞ d z 1 z 1 2 1 1 / z − 1 / z 1 ( S D ^ F T ( z , z 1 ) − S G ^ F T ( z , z 1 ) ) = D T ( z ) z + D 1 T ⊥ ( 1 ) ( z ) , (7) \int_z^\infty\frac{d z_1}{z_1^2}\frac{1}{1/z-1/z_1}\left(\mathcal{S}\widehat{D}_{FT}(z,z_1)-\mathcal{S}\widehat{G}_{FT}(z,z_1)\right)=\frac{D_T(z)}{z}+D_{1T}^{\perp(1)}(z), \tag*{(7)} ∫ z ∞ z 1 2 d z 1 1/ z − 1/ z 1 1 ( S D F T ( z , z 1 ) − S G F T ( z , z 1 ) ) = z D T ( z ) + D 1 T ⊥ ( 1 ) ( z ) , (7)
and the LIR reads
− 2 z ∫ z ∞ d z 1 z 1 2 S D ^ F T ( z , z 1 ) 1 / z 1 − 1 / z = D T ( z ) z + d ( D 1 T ⊥ ( 1 ) ( z ) / z ) d ( 1 / z ) . (8) -\frac{2}{z}\int_z^\infty\frac{d z_1}{z_1^2}\frac{\mathcal{S}\widehat{D}_{FT}(z,z_1)}{1/z_1-1/z}=\frac{D_T(z)}{z}+\frac{d\left(D_{1T}^{\perp(1)}(z)/z\right)}{d(1/z)}. \tag*{(8)} − z 2 ∫ z ∞ z 1 2 d z 1 1/ z 1 − 1/ z S D F T ( z , z 1 ) = z D T ( z ) + d ( 1/ z ) d ( D 1 T ⊥ ( 1 ) ( z ) / z ) . (8)
It has been shown that these relations are crucial to guarantee the gauge invariance and the frame independence of twist-3 cross sections for various processes [21 , 23 , 30 ] .
In addition to those in (4), there are another type of dynamical FFs defined from the matrix elements like ∼ ⟨ 0 ∣ g F a α w ∣ h X ⟩ ⟨ h X ∣ ψ ˉ t a ψ ∣ 0 ⟩ \sim\langle0|gF^{\alpha w}_a|hX\rangle\langle hX|\bar{\psi}t^a\psi|0\rangle ∼ ⟨ 0∣ g F a α w ∣ h X ⟩ ⟨ h X ∣ ψ ˉ t a ψ ∣0 ⟩ with t a t^a t a the generator of color SU(3). They are, however, related to the purely gluonic twist-3 FFs by the EOM relations and the LIRs as was shown in [22 ] . It’s been also shown that the combination of the contributions from the twist-3 purely gluonic FFs and these dynamical FFs gives the gauge and frame independent cross section for p p → Λ ↑ X pp\to\Lambda^\uparrow X pp → Λ ↑ X thanks to the LIRs and the EOM relations [20 ] . Therefore we will discuss those contributions together in a separate publication [31 ] .
Twist-3 cross section for e p → e Λ ↑ X ep\to e\Lambda^\uparrow X e p → e Λ ↑ X
Kinematics
Here we summarize the kinematics for
e ( ℓ ) + p ( p ) → e ( ℓ ′ ) + Λ ↑ ( P h , S ⊥ ) + X , (9) e(\ell)+p(p)\to e(\ell')+\Lambda^\uparrow(P_h,S_\perp)+X, \tag*{(9)} e ( ℓ ) + p ( p ) → e ( ℓ ′ ) + Λ ↑ ( P h , S ⊥ ) + X , (9)
where ℓ \ell ℓ , ℓ ′ \ell' ℓ ′ , p p p and P h P_h P h are the momenta of each particle, and S ⊥ S_\perp S ⊥ is the transverse spin vector of the produced Λ ↑ \Lambda^\uparrow Λ ↑ . To derive the cross section for this process, we define the following five Lorentz invariants:
S e p = ( p + ℓ ) 2 , x b j = Q 2 2 p ⋅ q , Q 2 = − q 2 = − ( ℓ − ℓ ′ ) 2 , z f = p ⋅ P h p ⋅ q , q T = − q t 2 , (10) \begin{aligned}
S_{ep} &= (p+\ell)^2,\\
x_{bj} &= \frac{Q^2}{2p\cdot q},\\
Q^2 &= -q^2=-(\ell-\ell')^2,\\
z_f &= \frac{p\cdot P_h}{p\cdot q},\\
q_T &= \sqrt{-q_t^2}, \tag*{(10)}
\end{aligned} S e p x bj Q 2 z f q T = ( p + ℓ ) 2 , = 2 p ⋅ q Q 2 , = − q 2 = − ( ℓ − ℓ ′ ) 2 , = p ⋅ q p ⋅ P h , = − q t 2 , (10)
where the space-like four momentum q t q_t q t is defined by
q t μ = q μ − P h ⋅ q p ⋅ P h p μ − p ⋅ q p ⋅ P h P h μ , (11) q_t^\mu=q^\mu-\frac{P_h\cdot q}{p\cdot P_h}p^\mu-\frac{p\cdot q}{p\cdot P_h}P_h^\mu, \tag*{(11)} q t μ = q μ − p ⋅ P h P h ⋅ q p μ − p ⋅ P h p ⋅ q P h μ , (11)
which satisfies q t ⋅ p = q t ⋅ P h = 0 q_t\cdot p=q_t\cdot P_h=0 q t ⋅ p = q t ⋅ P h = 0 .
We work in the hadron frame [34 ] , in which the momenta of the virtual photon and the initial proton are collinear as shown in Figure 1 . In this frame those momenta take
Figure 1. Hadron frame in which q ⃗ \vec{q} q and p ⃗ \vec{p} p are collinear. The angles ϕ \phi ϕ and χ \chi χ are, respectively, the azimuthal angles for the lepton plane and the hadron plane (measured from a certain reference plane), and Φ S \Phi_S Φ S is the azimuthal angle of the transverse spin vector of Λ ↑ \Lambda^\uparrow Λ ↑ , S ⃗ ⊥ \vec{S}_{\perp} S ⊥ , measured from the hadron plane.
q μ = ( q 0 , q ⃗ ) = ( 0 , 0 , 0 , − Q ) , (12) q^\mu=(q^0,\vec{q})=(0,0,0,-Q), \tag*{(12)} q μ = ( q 0 , q ) = ( 0 , 0 , 0 , − Q ) , (12)
p μ = ( Q 2 x b j , 0 , 0 , Q 2 x b j ) . (13) p^\mu=\left(\frac{Q}{2x_{bj}},0,0,\frac{Q}{2x_{bj}}\right). \tag*{(13)} p μ = ( 2 x bj Q , 0 , 0 , 2 x bj Q ) . (13)
We define the azimuthal angles of the hadron plane and the lepton plane as χ \chi χ and ϕ \phi ϕ , respectively.
With these angles, P h μ P_h^\mu P h μ can be written as
P h μ = z f Q 2 ( 1 + q T 2 Q 2 , 2 q T Q cos χ , 2 q T Q sin χ , − 1 + q T 2 Q 2 ) , (14) P_h^\mu= \frac{z_f Q}{2}\left(1+\frac{q_T^2}{Q^2},\frac{2q_T}{Q}\cos\chi,\frac{2q_T}{Q}\sin\chi,-1+\frac{q_T^2}{Q^2}\right), \tag*{(14)} P h μ = 2 z f Q ( 1 + Q 2 q T 2 , Q 2 q T cos χ , Q 2 q T sin χ , − 1 + Q 2 q T 2 ) , (14)
which implies the transverse momentum of Λ ↑ \Lambda^\uparrow Λ ↑ is P h T = z f q T P_{hT}=z_f q_T P h T = z f q T . For this P h μ P_h^\mu P h μ , w μ w^\mu w μ in (2) takes the following form:
w μ = 1 z f Q ( 1 + q T 2 / Q 2 ) ( 1 + q T 2 Q 2 , − 2 q T Q cos χ , − 2 q T Q sin χ , 1 − q T 2 Q 2 ) . (15) w^\mu= \frac{1}{z_f Q(1+q_T^2/Q^2)}\left(1+\frac{q_T^2}{Q^2},-\frac{2q_T}{Q}\cos\chi,-\frac{2q_T}{Q}\sin\chi,1-\frac{q_T^2}{Q^2}\right). \tag*{(15)} w μ = z f Q ( 1 + q T 2 / Q 2 ) 1 ( 1 + Q 2 q T 2 , − Q 2 q T cos χ , − Q 2 q T sin χ , 1 − Q 2 q T 2 ) . (15)
The initial and the scattered lepton momenta are, respectively, given by
ℓ μ = Q 2 ( cosh ψ , sinh ψ cos ϕ , sinh ψ sin ϕ , − 1 ) , (16) \ell^\mu= \frac{Q}{2}(\cosh\psi,\sinh\psi\cos\phi,\sinh\psi\sin\phi,-1), \tag*{(16)} ℓ μ = 2 Q ( cosh ψ , sinh ψ cos ϕ , sinh ψ sin ϕ , − 1 ) , (16)
ℓ ′ μ = Q 2 ( cosh ψ , sinh ψ cos ϕ , sinh ψ sin ϕ , 1 ) , (17) \ell'^\mu= \frac{Q}{2}(\cosh\psi,\sinh\psi\cos\phi,\sinh\psi\sin\phi,1), \tag*{(17)} ℓ ′ μ = 2 Q ( cosh ψ , sinh ψ cos ϕ , sinh ψ sin ϕ , 1 ) , (17)
with
cosh ψ = 2 x b j S e p Q 2 − 1. (18) \cosh\psi= \frac{2x_{bj}S_{ep}}{Q^2}-1. \tag*{(18)} cosh ψ = Q 2 2 x bj S e p − 1. (18)
In order to calculate the cross section, we introduce the following four vectors orthogonal to each other [34 ] ,
T μ = 1 Q ( q μ + 2 x b j p μ ) = ( 1 , 0 , 0 , 0 ) , (19) T^\mu= \frac{1}{Q}(q^\mu+2x_{bj}p^\mu)=(1,0,0,0), \tag*{(19)} T μ = Q 1 ( q μ + 2 x bj p μ ) = ( 1 , 0 , 0 , 0 ) , (19)
X μ = 1 q T { P h μ z f − q μ − ( 1 + q T 2 Q 2 ) x b j p μ } = ( 0 , cos χ , sin χ , 0 ) , (20) X^\mu= \frac{1}{q_T}\left\{\frac{P_h^\mu}{z_f}-q^\mu-\left(1+\frac{q_T^2}{Q^2}\right)x_{bj}p^\mu\right\}=(0,\cos\chi,\sin\chi,0), \tag*{(20)} X μ = q T 1 { z f P h μ − q μ − ( 1 + Q 2 q T 2 ) x bj p μ } = ( 0 , cos χ , sin χ , 0 ) , (20)
Z μ = − q μ Q = ( 0 , 0 , 0 , 1 ) , (21) Z^\mu=-\frac{q^\mu}{Q}=(0,0,0,1), \tag*{(21)} Z μ = − Q q μ = ( 0 , 0 , 0 , 1 ) , (21)
Y μ = ϵ μ ν ρ σ T ν X ρ Z σ = ( 0 , − sin χ , cos χ , 0 ) . (22) Y^\mu=\epsilon^{\mu\nu\rho\sigma}T_\nu X_\rho Z_\sigma=(0,-\sin\chi,\cos\chi,0). \tag*{(22)} Y μ = ϵ μν ρ σ T ν X ρ Z σ = ( 0 , − sin χ , cos χ , 0 ) . (22)
The polar angle θ \theta θ of P ⃗ h \vec{P}_h P h measured from the Z Z Z -axis can be written as
cos θ = P h z ∣ P ⃗ h ∣ = q T 2 − Q 2 q T 2 + Q 2 , (23) \cos\theta=\frac{P_{hz}}{|\vec{P}_h|}=\frac{q_T^2-Q^2}{q_T^2+Q^2}, \tag*{(23)} cos θ = ∣ P h ∣ P h z = q T 2 + Q 2 q T 2 − Q 2 , (23)
sin θ = P h T ∣ P ⃗ h ∣ = 2 q T Q q T 2 + Q 2 , (24) \sin\theta=\frac{P_{hT}}{|\vec{P}_h|}=\frac{2q_TQ}{q_T^2+Q^2}, \tag*{(24)} sin θ = ∣ P h ∣ P h T = q T 2 + Q 2 2 q T Q , (24)
where P h T = P h x 2 + P h y 2 P_{hT}=\sqrt{P_{hx}^2+P_{hy}^2} P h T = P h x 2 + P h y 2 . The transverse spin vector S ⃗ ⊥ \vec{S}_\perp S ⊥ of the hyperon resides in the plane which is orthogonal to P ⃗ h \vec{P}_h P h , and we define the azimuthal angle of S ⃗ ⊥ \vec{S}_\perp S ⊥ measured from the hadron plane around P ⃗ h \vec{P}_h P h as Φ S \Phi_S Φ S (See Figure 1 ). Then S ⊥ μ S_\perp^\mu S ⊥ μ can be written as
S ⊥ μ = cos θ cos Φ S X μ + sin Φ S Y μ − sin θ cos Φ S Z μ . (25) S_\perp^\mu=\cos\theta\cos\Phi_S X^\mu+\sin\Phi_S Y^\mu-\sin\theta\cos\Phi_S Z^\mu. \tag*{(25)} S ⊥ μ = cos θ cos Φ S X μ + sin Φ S Y μ − sin θ cos Φ S Z μ . (25)
The polarized cross section for (9) can be written as
d Δ σ = 1 2 S e p d 3 P ⃗ h ( 2 π ) 3 2 P h 0 d 3 ℓ ⃗ ( 2 π ) 3 2 ℓ 0 e 4 q 4 L μ ν ( ℓ , ℓ ′ ) W μ ν ( p , q , P h ) , (26) d\Delta\sigma=\frac{1}{2S_{ep}}\frac{d^3\vec{P}_h}{(2\pi)^3 2P_h^0}\frac{d^3\vec{\ell}}{(2\pi)^3 2\ell^0}\frac{e^4}{q^4}L^{\mu\nu}(\ell,\ell')W_{\mu\nu}(p,q,P_h), \tag*{(26)} d Δ σ = 2 S e p 1 ( 2 π ) 3 2 P h 0 d 3 P h ( 2 π ) 3 2 ℓ 0 d 3 ℓ q 4 e 4 L μν ( ℓ , ℓ ′ ) W μν ( p , q , P h ) , (26)
where W μ ν W_{\mu\nu} W μν is the hadronic tensor and L μ ν L^{\mu\nu} L μν is the unpolarized leptonic tensor defined by
L μ ν ( ℓ , ℓ ′ ) = 2 ( ℓ μ ℓ ′ ν + ℓ ′ μ ℓ ν ) − Q 2 g μ ν . (27) L^{\mu\nu}(\ell,\ell') = 2(\ell^\mu\ell'^\nu+\ell'^\mu\ell^\nu)-Q^2g^{\mu\nu}. \tag*{(27)} L μν ( ℓ , ℓ ′ ) = 2 ( ℓ μ ℓ ′ ν + ℓ ′ μ ℓ ν ) − Q 2 g μν . (27)
Using the kinematic variables introduced in (10) , the differential cross section can be written as
d 6 Δ σ d x b j d Q 2 d z f d q T 2 d ϕ d χ = α e m 2 z f 128 π 4 S e p 2 x b j 2 Q 2 L μ ν ( ℓ , ℓ ′ ) W μ ν ( p , q , P h ) , (28) \frac{\mathrm{d}^{6}\Delta\sigma}{\mathrm{d}x_{bj}\mathrm{d}Q^{2}\mathrm{d}z_{f}\mathrm{d}q_{T}^{2}\mathrm{d}\phi\mathrm{d}\chi}
=
\frac{\alpha_{\mathrm{em}}^{2}z_{f}}{128\pi^{4}S_{ep}^{2}x_{bj}^{2}Q^{2}}L^{\mu\nu}(\ell,\ell')W_{\mu\nu}(p,q,P_{h}), \tag*{(28)} d x bj d Q 2 d z f d q T 2 d ϕ d χ d 6 Δ σ = 128 π 4 S e p 2 x bj 2 Q 2 α em 2 z f L μν ( ℓ , ℓ ′ ) W μν ( p , q , P h ) , (28)
where α e m = e 2 / ( 4 π ) \alpha_{\mathrm{em}}=e^{2}/(4\pi) α em = e 2 / ( 4 π ) is the fine structure constant in QED. The hadronic tensor W μ ν W_{\mu\nu} W μν satisfies the current conservation q μ W μ ν = q ν W μ ν = 0 q^{\mu}W_{\mu\nu}=q^{\nu}W_{\mu\nu}=0 q μ W μν = q ν W μν = 0 and can be expanded by the following six tensors, V k μ ν \mathcal{V}_{k}^{\mu\nu} V k μν [34 ] :
V 1 μ ν = X μ X ν + Y μ Y ν , V 2 μ ν = g μ ν + Z μ Z ν , V 3 μ ν = T μ X ν + X μ T ν , V 4 μ ν = X μ X ν − Y μ Y ν , V 8 μ ν = T μ Y ν + Y μ T ν , V 9 μ ν = X μ Y ν + Y μ X ν . (29) \begin{aligned}
\mathcal{V}_{1}^{\mu\nu} &= X^{\mu}X^{\nu}+Y^{\mu}Y^{\nu},\\
\mathcal{V}_{2}^{\mu\nu} &= g^{\mu\nu}+Z^{\mu}Z^{\nu},\\
\mathcal{V}_{3}^{\mu\nu} &= T^{\mu}X^{\nu}+X^{\mu}T^{\nu},\\
\mathcal{V}_{4}^{\mu\nu} &= X^{\mu}X^{\nu}-Y^{\mu}Y^{\nu},\\
\mathcal{V}_{8}^{\mu\nu} &= T^{\mu}Y^{\nu}+Y^{\mu}T^{\nu},\\
\mathcal{V}_{9}^{\mu\nu} &= X^{\mu}Y^{\nu}+Y^{\mu}X^{\nu}. \tag*{(29)}
\end{aligned} V 1 μν V 2 μν V 3 μν V 4 μν V 8 μν V 9 μν = X μ X ν + Y μ Y ν , = g μν + Z μ Z ν , = T μ X ν + X μ T ν , = X μ X ν − Y μ Y ν , = T μ Y ν + Y μ T ν , = X μ Y ν + Y μ X ν . (29)
In order to calculate L μ ν W μ ν L^{\mu\nu}W_{\mu\nu} L μν W μν in (26) , we introduce the inverse tensors V ~ k μ ν \widetilde{\mathcal{V}}_{k}^{\mu\nu} V k μν for V k μ ν \mathcal{V}_{k}^{\mu\nu} V k μν :
V ~ 1 μ ν = 1 2 ( 2 T μ T ν + X μ X ν + Y μ Y ν ) , V ~ 2 μ ν = T μ T ν , V ~ 3 μ ν = − 1 2 ( T μ X ν + X μ T ν ) , V ~ 4 μ ν = 1 2 ( X μ X ν − Y μ Y ν ) , V ~ 8 μ ν = − 1 2 ( T μ Y ν + Y μ T ν ) , V ~ 9 μ ν = 1 2 ( X μ Y ν + Y μ X ν ) . (30) \begin{aligned}
\widetilde{\mathcal{V}}_{1}^{\mu\nu} &= \frac{1}{2}(2T^{\mu}T^{\nu}+X^{\mu}X^{\nu}+Y^{\mu}Y^{\nu}),\\
\widetilde{\mathcal{V}}_{2}^{\mu\nu} &= T^{\mu}T^{\nu},\\
\widetilde{\mathcal{V}}_{3}^{\mu\nu} &= -\frac{1}{2}(T^{\mu}X^{\nu}+X^{\mu}T^{\nu}),\\
\widetilde{\mathcal{V}}_{4}^{\mu\nu} &= \frac{1}{2}(X^{\mu}X^{\nu}-Y^{\mu}Y^{\nu}),\\
\widetilde{\mathcal{V}}_{8}^{\mu\nu} &= -\frac{1}{2}(T^{\mu}Y^{\nu}+Y^{\mu}T^{\nu}),\\
\widetilde{\mathcal{V}}_{9}^{\mu\nu} &= \frac{1}{2}(X^{\mu}Y^{\nu}+Y^{\mu}X^{\nu}). \tag*{(30)}
\end{aligned} V 1 μν V 2 μν V 3 μν V 4 μν V 8 μν V 9 μν = 2 1 ( 2 T μ T ν + X μ X ν + Y μ Y ν ) , = T μ T ν , = − 2 1 ( T μ X ν + X μ T ν ) , = 2 1 ( X μ X ν − Y μ Y ν ) , = − 2 1 ( T μ Y ν + Y μ T ν ) , = 2 1 ( X μ Y ν + Y μ X ν ) . (30)
With these V k μ ν \mathcal{V}_{k}^{\mu\nu} V k μν and V ~ k μ ν \widetilde{\mathcal{V}}_{k}^{\mu\nu} V k μν , one obtains L μ ν W μ ν L^{\mu\nu}W_{\mu\nu} L μν W μν as
L μ ν W μ ν = ∑ k = 1 , ⋯ , 9 [ L μ ν V k μ ν ] [ W ρ σ V ~ k ρ σ ] = Q 2 ∑ k = 1 , ⋯ , 9 A k ( ϕ − χ ) [ W ρ σ V ~ k ρ σ ] , (31) L^{\mu\nu}W_{\mu\nu}
=
\sum_{k=1,\cdots,9}
\left[L_{\mu\nu}\mathcal{V}_{k}^{\mu\nu}\right]
\left[W_{\rho\sigma}\widetilde{\mathcal{V}}_{k}^{\rho\sigma}\right]
=
Q^{2}\sum_{k=1,\cdots,9}
A_{k}(\phi-\chi)
\left[W_{\rho\sigma}\widetilde{\mathcal{V}}_{k}^{\rho\sigma}\right], \tag*{(31)} L μν W μν = k = 1 , ⋯ , 9 ∑ [ L μν V k μν ] [ W ρ σ V k ρ σ ] = Q 2 k = 1 , ⋯ , 9 ∑ A k ( ϕ − χ ) [ W ρ σ V k ρ σ ] , (31)
where A k ( φ ) A_{k}(\varphi) A k ( φ ) (φ ≡ ϕ − χ \varphi\equiv\phi-\chi φ ≡ ϕ − χ , k = 1 , ⋯ , 4 , 8 , 9 k=1,\cdots,4,8,9 k = 1 , ⋯ , 4 , 8 , 9 ) is defined by
A k ( φ ) = L μ ν V k μ ν / Q 2 , (32) A_{k}(\varphi)=L_{\mu\nu}\mathcal{V}_{k}^{\mu\nu}/Q^{2}, \tag*{(32)} A k ( φ ) = L μν V k μν / Q 2 , (32)
and they are calculated to be
A 1 ( φ ) = 1 + cosh 2 ψ , A 2 ( φ ) = − 2 , A 3 ( φ ) = − cos φ sinh 2 ψ , A 4 ( φ ) = cos 2 φ sinh 2 ψ , A 8 ( φ ) = − sin φ sinh 2 ψ , A 9 ( φ ) = sin 2 φ sinh 2 ψ . (33) \begin{aligned}
A_{1}(\varphi) &= 1+\cosh^{2}\psi,\\
A_{2}(\varphi) &= -2,\\
A_{3}(\varphi) &= -\cos\varphi\sinh2\psi,\\
A_{4}(\varphi) &= \cos2\varphi\sinh^{2}\psi,\\
A_{8}(\varphi) &= -\sin\varphi\sinh2\psi,\\
A_{9}(\varphi) &= \sin2\varphi\sinh^{2}\psi. \tag*{(33)}
\end{aligned} A 1 ( φ ) A 2 ( φ ) A 3 ( φ ) A 4 ( φ ) A 8 ( φ ) A 9 ( φ ) = 1 + cosh 2 ψ , = − 2 , = − cos φ sinh 2 ψ , = cos 2 φ sinh 2 ψ , = − sin φ sinh 2 ψ , = sin 2 φ sinh 2 ψ . (33)
Corresponding to A k ( ϕ − χ ) A_k(\phi-\chi) A k ( ϕ − χ ) (k = 1 , 2 ⋯ 9 k=1,2\cdots9 k = 1 , 2 ⋯ 9 ), the cross section for (9) consists of five components with different dependences on the azimuthal angle ϕ − χ \phi-\chi ϕ − χ .
Contribution from unpolarized twist-3 distribution and the transversity fragmentation function
In this section we calculate the twist-3 cross section for e p → e Λ ↑ X ep\to e\Lambda^\uparrow X e p → e Λ ↑ X which arises from the twist-3 DF in the nucleon. The method of the calculation is described in detail in [6 ] , which developed the formalism for a similar twist-3 process e p ↑ → e π X ep^\uparrow\to e\pi X e p ↑ → e π X . At leading order with respect to the QCD coupling constant, only the dynamical twist-3 DF in the nucleon E F ( x 1 , x 2 ) E_F(x_1,x_2) E F ( x 1 , x 2 ) defined in (1) contributes together with the transversity FF H 1 ( z ) H_1(z) H 1 ( z ) for Λ ↑ \Lambda^\uparrow Λ ↑ , which is schematically shown in Figure 2 . Since the original calculation in [6 ] overlooked some of the diagrams for the hard part [11 , 35 ] , we shall also include those new diagrams below. The cross section occurs as pole contributions from the hard part, which are classified into the hard pole (HP), soft-gluon-pole (SGP) and soft-fermion-pole (SFP). For the SFP contribution, the new type of diagrams found in [11 ] cancel the contribution from the quark’s SFP function E F ( 0 , x ) E_F(0,x) E F ( 0 , x ) paired with the quark’s transversity FF arising from the SFP diagrams considered in [6 ] . Accordingly, in the present case, the SFP contribution occurs from the left four diagrams in Fig. 1 of [35 ] in which the lower quark line (for anti-quark) crossing the final state cut fragments into Λ ↑ \Lambda^\uparrow Λ ↑ , together with the diagrams obtained by reversing the arrows on the quark lines. Therefore the SFP contribution survives only for the anti-quark’s transversity FF paired with the quark’s SFP DF or the quark’s transversity FF paired with the anti-quark’s SFP DF [35 ] . We expect these contributions involving the FF or SFP functions for an anti-quark should be much smaller compared with those from the SGP and HP ones. Hence we will not consider them below. It was also found in [35 ] that there are some other HP diagrams not considered in [6 ] . We will also include those new types of HP contributions below.
Figure 2. Twist-3 distribution function contribution to e p → e Λ ↑ X ep\to e\Lambda^\uparrow X e p → e Λ ↑ X . The top blob indicates the fragmentation correlator for H 1 ( z ) H_1(z) H 1 ( z ) , the bottom one represents the correlation function for E F ( x 1 , x 2 ) E_F(x_1,x_2) E F ( x 1 , x 2 ) and the middle one corresponds to the partonic hard part.
Using (28) and (31), and factorizing the twist-2 transversity FF H 1 ( z ) H_1(z) H 1 ( z ) from the hadronic tensor, one can write this cross section as
d 6 Δ σ t w 3 − d i s t d x b j d Q 2 d z f d q T 2 d ϕ d χ = z f α e m 2 128 π 4 S e p 2 x b j 2 ∑ k = 1 9 A k ( ϕ − χ ) ∫ d z z 2 H 1 ( z ) [ w μ ν ( p , q , P h / z ) V ~ k μ ν ] , (34) \frac{d^6\Delta\sigma^{\mathrm{tw3-dist}}}{dx_{bj}dQ^2dz_fdq_T^2d\phi d\chi}
=
\frac{z_f\alpha_{\mathrm{em}}^2}{128\pi^4S_{ep}^2x_{bj}^2}
\sum_{k=1}^{9} A_k(\phi-\chi)
\int\frac{dz}{z^2}H_1(z)
\left[w_{\mu\nu}(p,q,P_h/z)\widetilde V_k^{\mu\nu}\right], \tag*{(34)} d x bj d Q 2 d z f d q T 2 d ϕ d χ d 6 Δ σ tw3 − dist = 128 π 4 S e p 2 x bj 2 z f α em 2 k = 1 ∑ 9 A k ( ϕ − χ ) ∫ z 2 d z H 1 ( z ) [ w μν ( p , q , P h / z ) V k μν ] , (34)
where the summation over quark flavors as well as the factor associated with the quark’s fractional electric charge is omitted. Applying the formalism described in [6 ] , we obtain w μ ν w_{\mu\nu} w μν in (34) in terms of the gauge-invariant correlation function M F α ( x 1 , x 2 ) M_F^\alpha(x_1,x_2) M F α ( x 1 , x 2 ) in (1) as
w μ ν ( p , q , P h / z ) = ∫ d x 1 ∫ d x 2 Tr [ i ω β α M F β ( x 1 , x 2 ) ∂ S σ , μ ν H P / S G P ( k 1 , k 2 , q , P h / z ) p σ ∂ k 2 α ] k i = x i p , (35) w_{\mu\nu}(p,q,P_h/z)=\int dx_1\int dx_2\operatorname{Tr}\left[i\omega_\beta^\alpha M_F^\beta(x_1,x_2)\frac{\partial S_{\sigma,\mu\nu}^{\mathrm{HP/SGP}}(k_1,k_2,q,P_h/z)p^\sigma}{\partial k_2^\alpha}\right]_{k_i=x_i p}, \tag*{(35)} w μν ( p , q , P h / z ) = ∫ d x 1 ∫ d x 2 Tr [ i ω β α M F β ( x 1 , x 2 ) ∂ k 2 α ∂ S σ , μν HP/SGP ( k 1 , k 2 , q , P h / z ) p σ ] k i = x i p , (35)
where ω β α = g β α − p α n β \omega_\beta^\alpha=g_\beta^\alpha-p^\alpha n_\beta ω β α = g β α − p α n β and the summation over color indices is implicit. S σ , μ ν H P / S G P S_{\sigma,\mu\nu}^{\mathrm{HP/SGP}} S σ , μν HP/SGP is the partonic hard part corresponding to the original nucleon matrix element M σ ( k 1 , k 2 ) ∼ F T ⟨ p ∣ ψ ˉ g A σ ψ ∣ p ⟩ M^\sigma(k_1,k_2)\sim\mathcal{F}\mathcal{T}\langle p|\bar{\psi}gA^\sigma\psi|p\rangle M σ ( k 1 , k 2 ) ∼ F T ⟨ p ∣ ψ ˉ g A σ ψ ∣ p ⟩ . Although we start from ∫ d 4 k 1 ∫ d 4 k 2 M σ ( k 1 , k 2 ) S σ ( k 1 , k 2 ) \int d^4k_1\int d^4k_2 M^\sigma(k_1,k_2)S_\sigma(k_1,k_2) ∫ d 4 k 1 ∫ d 4 k 2 M σ ( k 1 , k 2 ) S σ ( k 1 , k 2 ) for the cross section, reorganization of the collinear expansion using the Ward identity allows us to convert the gauge field into the field strength in the correlation function and leads to the gauge invariant expression for the twist-3 cross section as shown in (35) (see [6 ] for the details).
The LO diagrams for S σ , μ ν H P S_{\sigma,\mu\nu}^{\mathrm{HP}} S σ , μν HP are given in Fig. 2 of [6 ] and the left two diagrams of Fig. 2 of [35 ] .2 The LO diagrams for S σ , μ ν S G P S_{\sigma,\mu\nu}^{\mathrm{SGP}} S σ , μν SGP are given in Fig. 8 of [6 ] . We remind that the hard part for the HP contribution satisfies the following relation owing to the Ward identity [6 ] ,
∂ S σ , μ ν H P ( k 1 , k 2 , q , P h / z ) p σ ∂ k 2 α ∣ k i = x i p = 1 x 1 − x 2 S α , μ ν H P ( x 1 p , x 2 p , q , P h / z ) , (36) \left.\frac{\partial S_{\sigma,\mu\nu}^{\mathrm{HP}}(k_1,k_2,q,P_h/z)p^\sigma}{\partial k_2^\alpha}\right|_{k_i=x_i p}=\frac{1}{x_1-x_2}S_{\alpha,\mu\nu}^{\mathrm{HP}}(x_1p,x_2p,q,P_h/z), \tag*{(36)} ∂ k 2 α ∂ S σ , μν HP ( k 1 , k 2 , q , P h / z ) p σ k i = x i p = x 1 − x 2 1 S α , μν HP ( x 1 p , x 2 p , q , P h / z ) , (36)
and thus one can obtain the HP contribution without calculating the derivative in (35) . The hard part contains the δ \delta δ -function corresponding to the on-shell condition for the final unobserved parton, which takes
δ ( ( x p + q − P h / z ) 2 ) = 1 Q 2 z ^ δ ( q T 2 Q 2 − ( 1 − 1 x ^ ) ( 1 − 1 z ^ ) ) , (37) \delta\left((xp+q-P_h/z)^2\right)=\frac{1}{Q^2\hat{z}}\delta\left(\frac{q_T^2}{Q^2}-\left(1-\frac{1}{\hat{x}}\right)\left(1-\frac{1}{\hat{z}}\right)\right), \tag*{(37)} δ ( ( x p + q − P h / z ) 2 ) = Q 2 z ^ 1 δ ( Q 2 q T 2 − ( 1 − x ^ 1 ) ( 1 − z ^ 1 ) ) , (37)
where the variables x ^ \hat{x} x ^ and z ^ \hat{z} z ^ are defined as
x ^ = x b j x , z ^ = z f z . (38) \hat{x}=\frac{x_{bj}}{x},\qquad\hat{z}=\frac{z_f}{z}. \tag*{(38)} x ^ = x x bj , z ^ = z z f . (38)
Calculating the LO diagrams for the hard part, we have obtained the HP contribution as
d 6 Δ σ H P d x b j d Q 2 d z f d q T 2 d ϕ d χ = α e m 2 α s 16 π 2 s e p b j 2 Q 2 ( − π M N 4 ) ∫ d z z H 1 ( z ) × ∫ d x x [ 2 1 − x ^ E F ( x b j , x ) ( 4 N C q T − 4 N C Q 2 ( x ^ − 1 ) q T 3 x ^ ) sinh 2 ψ sin { Φ S + 2 ( ϕ − χ ) } + E F ( x b j , x b j − x ) ( − 1 N ) { 8 x ^ z ^ q T ( 1 + cosh 2 ψ ) sin Φ S + 8 ( 1 − x ^ ) Q z ^ q T 2 sinh 2 ψ sin ( Φ S + ϕ − χ ) + 8 ( 1 − x ^ ) 2 Q 2 x ^ z ^ q T 3 sinh 2 ψ sin { Φ S + 2 ( ϕ − χ ) } } ] × δ ( q T 2 Q 2 − ( 1 − 1 x ^ ) ( 1 − 1 z ^ ) ) , (39) \begin{aligned}
\frac{d^6\Delta\sigma^{\mathrm{HP}}}{dx_{bj}dQ^2dz_fdq_T^2d\phi d\chi}
={}&\frac{\alpha_{\mathrm{em}}^2\alpha_s}{16\pi^2s_ep_{bj}^2Q^2}\left(\frac{-\pi M_N}{4}\right)\int\frac{dz}{z}H_1(z)\\
&\times\int\frac{dx}{x}\left[\frac{2}{1-\hat{x}}E_F(x_{bj},x)\left(\frac{4}{N_{C}q_T}-\frac{4N_CQ^2(\hat{x}-1)}{q_T^3\hat{x}}\right)\sinh^2\psi\sin\{\Phi_S+2(\phi-\chi)\}\right.\\
&\quad+E_F(x_{bj},x_{bj}-x)\left(-\frac{1}{N}\right)\left\{\frac{8\hat{x}}{\hat{z}q_T}(1+\cosh^2\psi)\sin\Phi_S\right.\\
&\qquad\left.\left.+\frac{8(1-\hat{x})Q}{\hat{z}q_T^2}\sinh2\psi\sin(\Phi_S+\phi-\chi)+\frac{8(1-\hat{x})^2Q^2}{\hat{x}\hat{z}q_T^3}\sinh^2\psi\sin\{\Phi_S+2(\phi-\chi)\}\right\}\right]\\
&\times\delta\left(\frac{q_T^2}{Q^2}-\left(1-\frac{1}{\hat{x}}\right)\left(1-\frac{1}{\hat{z}}\right)\right), \tag*{(39)}
\end{aligned} d x bj d Q 2 d z f d q T 2 d ϕ d χ d 6 Δ σ HP = 16 π 2 s e p bj 2 Q 2 α em 2 α s ( 4 − π M N ) ∫ z d z H 1 ( z ) × ∫ x d x [ 1 − x ^ 2 E F ( x bj , x ) ( N C q T 4 − q T 3 x ^ 4 N C Q 2 ( x ^ − 1 ) ) sinh 2 ψ sin { Φ S + 2 ( ϕ − χ )} + E F ( x bj , x bj − x ) ( − N 1 ) { z ^ q T 8 x ^ ( 1 + cosh 2 ψ ) sin Φ S + z ^ q T 2 8 ( 1 − x ^ ) Q sinh 2 ψ sin ( Φ S + ϕ − χ ) + x ^ z ^ q T 3 8 ( 1 − x ^ ) 2 Q 2 sinh 2 ψ sin { Φ S + 2 ( ϕ − χ )} } ] × δ ( Q 2 q T 2 − ( 1 − x ^ 1 ) ( 1 − z ^ 1 ) ) , (39)
where α s = g 2 / ( 4 π ) \alpha_s = g^2/(4\pi) α s = g 2 / ( 4 π ) is the strong coupling constant. The E F ( x b j , x ) E_F(x_{bj},x) E F ( x bj , x ) contribution occurs from the diagrams in Fig. 2 of [6 ] and the E F ( x b j , x b j − x ) E_F(x_{bj},x_{bj}-x) E F ( x bj , x bj − x ) contribution is from the left two diagrams of Fig. 2 of [35 ] . We note that k = 3 , 8 k=3,8 k = 3 , 8 and k = 4 , 9 k=4,9 k = 4 , 9 terms can be, respectively, combined into the single sin forms ∼ sin ( Φ S + ϕ − χ ) \sim\sin(\Phi_S+\phi-\chi) ∼ sin ( Φ S + ϕ − χ ) and ∼ sin { Φ S + 2 ( ϕ − χ ) } \sim\sin\{\Phi_S+2(\phi-\chi)\} ∼ sin { Φ S + 2 ( ϕ − χ )} .
For the LO calculation of the SGP contribution we found the method using the master formula [10 ] is convenient. The result reads
d 6 Δ σ S G P d x b j d Q 2 d z f d q T 2 d ϕ d χ = α e m 2 α s 16 π 2 S e p 2 x b j 2 Q 2 ( − π M N 4 ) q T Q ∫ d z z H 1 ( z ) ∫ d x x ( − 1 2 N ) [ − 8 ( 1 + cosh 2 ψ ) sin Φ S { 2 x ^ 1 − z ^ ( x d E F ( x , x ) d x − E F ( x , x ) ) + ( ( 1 + z ^ ) Q 2 z ^ 2 q T 2 + 2 x ^ 1 − x ^ ) E F ( x , x ) } + 8 sinh 2 ψ sin ( Φ S + ϕ − χ ) { 2 x ^ Q ( 1 − z ^ ) q T ( x d E F ( x , x ) d x − E F ( x , x ) ) − Q 2 q T ( Q 2 q T 2 + 1 ) E F ( x , x ) } − 8 sinh 2 ψ sin { Φ S + 2 ( ϕ − χ ) } { 2 x ^ Q 2 ( 1 − z ^ ) q T 2 ( x d E F ( x , x ) d x − E F ( x , x ) ) − ( 3 Q 4 z ^ q T 4 + Q 2 q T 2 ) E F ( x , x ) } ] δ ( q T 2 Q 2 − ( 1 − 1 x ^ ) ( 1 − 1 z ^ ) ) . (40) \frac{d^6\Delta\sigma^{\mathrm{SGP}}}{dx_{bj}dQ^2dz_fdq_T^2d\phi d\chi}
=
\frac{\alpha_{\mathrm{em}}^2\alpha_s}{16\pi^2S_{ep}^2x_{bj}^2Q^2}
\left(\frac{-\pi M_N}{4}\right)\frac{q_T}{Q}\int\frac{dz}{z}H_1(z)\int\frac{dx}{x}\left(\frac{-1}{2N}\right)
\left[
-8(1+\cosh^2\psi)\sin\Phi_S
\left\{
\frac{2\hat{x}}{1-\hat{z}}\left(x\frac{dE_F(x,x)}{dx}-E_F(x,x)\right)
+\left(\frac{(1+\hat{z})Q^2}{\hat{z}^2q_T^2}+\frac{2\hat{x}}{1-\hat{x}}\right)E_F(x,x)
\right\}
+8\sinh2\psi\sin(\Phi_S+\phi-\chi)
\left\{
\frac{2\hat{x}Q}{(1-\hat{z})q_T}\left(x\frac{dE_F(x,x)}{dx}-E_F(x,x)\right)
-\frac{Q}{2q_T}\left(\frac{Q^2}{q_T^2}+1\right)E_F(x,x)
\right\}
-8\sinh^2\psi\sin\{\Phi_S+2(\phi-\chi)\}
\left\{
\frac{2\hat{x}Q^2}{(1-\hat{z})q_T^2}\left(x\frac{dE_F(x,x)}{dx}-E_F(x,x)\right)
-\left(\frac{3Q^4}{\hat{z}q_T^4}+\frac{Q^2}{q_T^2}\right)E_F(x,x)
\right\}
\right]\delta\left(\frac{q_T^2}{Q^2}-\left(1-\frac{1}{\hat{x}}\right)\left(1-\frac{1}{\hat{z}}\right)\right). \tag*{(40)} d x bj d Q 2 d z f d q T 2 d ϕ d χ d 6 Δ σ SGP = 16 π 2 S e p 2 x bj 2 Q 2 α em 2 α s ( 4 − π M N ) Q q T ∫ z d z H 1 ( z ) ∫ x d x ( 2 N − 1 ) [ − 8 ( 1 + cosh 2 ψ ) sin Φ S { 1 − z ^ 2 x ^ ( x d x d E F ( x , x ) − E F ( x , x ) ) + ( z ^ 2 q T 2 ( 1 + z ^ ) Q 2 + 1 − x ^ 2 x ^ ) E F ( x , x ) } + 8 sinh 2 ψ sin ( Φ S + ϕ − χ ) { ( 1 − z ^ ) q T 2 x ^ Q ( x d x d E F ( x , x ) − E F ( x , x ) ) − 2 q T Q ( q T 2 Q 2 + 1 ) E F ( x , x ) } − 8 sinh 2 ψ sin { Φ S + 2 ( ϕ − χ )} { ( 1 − z ^ ) q T 2 2 x ^ Q 2 ( x d x d E F ( x , x ) − E F ( x , x ) ) − ( z ^ q T 4 3 Q 4 + q T 2 Q 2 ) E F ( x , x ) } ] δ ( Q 2 q T 2 − ( 1 − x ^ 1 ) ( 1 − z ^ 1 ) ) . (40)
It turned out that the LO SGP cross sections for k = 3 , 8 k=3,8 k = 3 , 8 and k = 4 , 9 k=4,9 k = 4 , 9 can be also transformed into the single sin forms as in the HP contribution.
To summarize this subsection, the cross section in (34) is given by the sum of (39) and (40) .
Contribution from twist-3 quark fragmentation function
The formalism for calculating the twist-3 quark FF contribution has been developed in [17 ] for the process e p ↑ → e π X ep^\uparrow\to e\pi X e p ↑ → e π X . This contribution is diagrammatically shown in Figure 3 . From eq. (54) of [17 ] , one can write the twist-3 quark FF contribution to (9) as
Figure 3. Twist-3 quark fragmentation function contribution to e p → e Λ ↑ X ep \to e\Lambda^\uparrow X e p → e Λ ↑ X . The top blob represents the fragmentation correlator, the bottom one indicates the unpolarized quark DF, and the middle one corresponds to the partonic hard part. The contribution from the unpolarized gluon DF should also be included.
d 6 Δ σ t w 3 − f r a g d x b j d Q 2 d z f d q T 2 d ϕ d χ = α e m 2 128 π 4 x b j 2 S e p 2 Q 2 z f Q 2 ∑ k = 1 , ⋯ , 4 , 8 , 9 A k ( ϕ − χ ) ∫ d x x f 1 ( x ) [ ∫ d z z 2 Tr [ Δ ( z ) S ρ σ ( z ) ] V ~ k ρ σ + Ω β α ∫ d z z 2 Tr [ Δ ∂ β ( z ) ∂ S ρ σ ( k ) ∂ k α ∣ c . l . ] V ~ k ρ σ \frac{d^6\Delta\sigma^{\mathrm{tw3-frag}}}{dx_{bj}dQ^2dz_fdq_T^2d\phi d\chi}
=
\frac{\alpha_{\mathrm{em}}^2}{128\pi^4x_{bj}^2S_{ep}^2Q^2}z_fQ^2
\sum_{k=1,\cdots,4,8,9}\mathcal{A}_k(\phi-\chi)\int\frac{dx}{x}f_1(x)
\left[
\int\frac{dz}{z^2}\operatorname{Tr}\left[\Delta(z)S_{\rho\sigma}(z)\right]\widetilde{\mathcal{V}}_k^{\rho\sigma}
+\Omega^\alpha_{\ \beta}\int\frac{dz}{z^2}\operatorname{Tr}\left[\Delta^\beta_\partial(z)\left.\frac{\partial S_{\rho\sigma}(k)}{\partial k^\alpha}\right|_{\mathrm{c.l.}}\right]\widetilde{\mathcal{V}}_k^{\rho\sigma}
\right. d x bj d Q 2 d z f d q T 2 d ϕ d χ d 6 Δ σ tw3 − frag = 128 π 4 x bj 2 S e p 2 Q 2 α em 2 z f Q 2 k = 1 , ⋯ , 4 , 8 , 9 ∑ A k ( ϕ − χ ) ∫ x d x f 1 ( x ) [ ∫ z 2 d z Tr [ Δ ( z ) S ρ σ ( z ) ] V k ρ σ + Ω β α ∫ z 2 d z Tr [ Δ ∂ β ( z ) ∂ k α ∂ S ρ σ ( k ) c.l. ] V k ρ σ
+ Ω β α ∫ d z z 2 ∫ d z ′ z ′ 2 P ( 1 1 / z ′ − 1 / z ) { Tr [ Δ F β ( z , z ′ ) S α , ρ σ L ( z ′ , z ) ] + ( ρ ↔ σ ) } V ~ k ρ σ ] , (41) \left.
+\Omega^\alpha_{\ \beta}\int\frac{dz}{z^2}\int\frac{dz'}{z'^2}\operatorname{P}\left(\frac{1}{1/z'-1/z}\right)
\left\{\operatorname{Tr}\left[\Delta_F^\beta(z,z')S^L_{\alpha,\rho\sigma}(z',z)\right]+(\rho\leftrightarrow\sigma)\right\}\widetilde{\mathcal{V}}_k^{\rho\sigma}
\right], \tag*{(41)} + Ω β α ∫ z 2 d z ∫ z ′2 d z ′ P ( 1/ z ′ − 1/ z 1 ) { Tr [ Δ F β ( z , z ′ ) S α , ρ σ L ( z ′ , z ) ] + ( ρ ↔ σ ) } V k ρ σ ] , (41)
where Ω β α = g β α − P α w β \Omega^\alpha_{\ \beta}=g^\alpha_{\ \beta}-P^\alpha w_\beta Ω β α = g β α − P α w β , f 1 ( x ) f_1(x) f 1 ( x ) denotes the twist-2 unpolarized quark or gluon distribution function, and the summation over quark flavors as well as the factor associated with the quark’s fractional electric charge is omitted. In (41) , S ρ σ ( k ) S_{\rho\sigma}(k) S ρ σ ( k ) and S α , ρ σ L ( z ′ , z ) S^L_{\alpha,\rho\sigma}(z',z) S α , ρ σ L ( z ′ , z ) are the partonic hard parts originally associated, respectively, with the fragmentation matrix elements ∑ X ⟨ 0 ∣ ψ ∣ h X ⟩ ⟨ h X ∣ ψ ˉ ∣ 0 ⟩ \sum_X\langle0|\psi|hX\rangle\langle hX|\bar{\psi}|0\rangle ∑ X ⟨ 0∣ ψ ∣ h X ⟩ ⟨ h X ∣ ψ ˉ ∣0 ⟩ and ∑ X ⟨ 0 ∣ ψ ∣ h X ⟩ ⟨ h X ∣ ψ ˉ g A α ∣ 0 ⟩ \sum_X\langle0|\psi|hX\rangle\langle hX|\bar{\psi}gA^\alpha|0\rangle ∑ X ⟨ 0∣ ψ ∣ h X ⟩ ⟨ h X ∣ ψ ˉ g A α ∣0 ⟩ . Inserting (2), (3) and (4) into (41), one can obtain the twist-3 FF contribution to the cross section as3 ^{3} 3
d 6 Δ σ t w 3 − f r a g d x b j d Q 2 d z f d q T 2 d ϕ d χ = α e m 2 α s ( − M h ) 16 π 2 x b j 2 S e p 2 Q 2 ∑ k = 1 9 A k ( ϕ − χ ) S k ∫ d x x f 1 ( x ) ∫ d z z δ ( q T 2 Q 2 − ( 1 − 1 x ^ ) ( 1 − 1 z ^ ) ) × [ D T ( z ) z σ ^ i n t k − { d d ( 1 / z ) D 1 T ⊥ ( 1 ) ( z ) z } { σ ^ k i n 1 k − D 1 T ⊥ ( 1 ) ( z ) σ ^ k i n 2 k } + ∫ d z ′ z ′ 2 P ( 1 1 / z − 1 / z ′ ) { D ^ F T ( z , z ′ ) [ σ ^ D F 3 k − 2 z ( 1 1 / z − 1 / z ′ ) σ ^ D F 4 k + z ′ z σ ^ D F 1 k + 1 z ( 1 z ′ − 1 ( 1 − q T 2 / Q 2 ) z f ) − 1 σ ^ D F 2 k ] + G ^ F T ( z , z ′ ) [ σ ^ G F 3 k − 2 z ( 1 1 / z − 1 / z ′ ) σ ^ G F 4 k + z ′ z σ ^ G F 1 k + 1 z ( 1 z ′ − 1 ( 1 − q T 2 / Q 2 ) z f ) − 1 σ ^ G F 2 k ] } ] , (42) \begin{aligned}
\frac{d^{6}\Delta\sigma^{\mathrm{tw3-frag}}}{dx_{bj}dQ^{2}dz_{f}dq_{T}^{2}d\phi d\chi}
={}&\frac{\alpha_{\mathrm{em}}^{2}\alpha_{s}(-M_{h})}{16\pi^{2}x_{bj}^{2}S_{ep}^{2}Q^{2}}\sum_{k=1}^{9}A_{k}(\phi-\chi)S_{k}\int\frac{dx}{x}f_{1}(x)\int\frac{dz}{z}\delta(\frac{q_{T}^{2}}{Q^{2}}-(1-\frac{1}{\hat{x}})(1-\frac{1}{\hat{z}}))\\
&\times[\frac{D_{T}(z)}{z}\hat{\sigma}_{\mathrm{int}}^{k}-\{\frac{d}{d(1/z)}\frac{D_{1T}^{\perp(1)}(z)}{z}\}\{\hat{\sigma}_{\mathrm{kin1}}^{k}-D_{1T}^{\perp(1)}(z)\hat{\sigma}_{\mathrm{kin2}}^{k}\}\\
&\quad+\int\frac{dz'}{z'^{2}}\mathcal{P}(\frac{1}{1/z-1/z'})\{\widehat{D}_{FT}(z,z')[\hat{\sigma}_{DF3}^{k}-\frac{2}{z}(\frac{1}{1/z-1/z'})\hat{\sigma}_{DF4}^{k}\\
&\qquad+\frac{z'}{z}\hat{\sigma}_{DF1}^{k}+\frac{1}{z}(\frac{1}{z'}-\frac{1}{(1-q_{T}^{2}/Q^{2})z_{f}})^{-1}\hat{\sigma}_{DF2}^{k}]\\
&\qquad+\widehat{G}_{FT}(z,z')[\hat{\sigma}_{GF3}^{k}-\frac{2}{z}(\frac{1}{1/z-1/z'})\hat{\sigma}_{GF4}^{k}+\frac{z'}{z}\hat{\sigma}_{GF1}^{k}\\
&\qquad+\frac{1}{z}(\frac{1}{z'}-\frac{1}{(1-q_{T}^{2}/Q^{2})z_{f}})^{-1}\hat{\sigma}_{GF2}^{k}]\}], \tag*{(42)}
\end{aligned} d x bj d Q 2 d z f d q T 2 d ϕ d χ d 6 Δ σ tw3 − frag = 16 π 2 x bj 2 S e p 2 Q 2 α em 2 α s ( − M h ) k = 1 ∑ 9 A k ( ϕ − χ ) S k ∫ x d x f 1 ( x ) ∫ z d z δ ( Q 2 q T 2 − ( 1 − x ^ 1 ) ( 1 − z ^ 1 )) × [ z D T ( z ) σ ^ int k − { d ( 1/ z ) d z D 1 T ⊥ ( 1 ) ( z ) } { σ ^ kin1 k − D 1 T ⊥ ( 1 ) ( z ) σ ^ kin2 k } + ∫ z ′ 2 d z ′ P ( 1/ z − 1/ z ′ 1 ) { D F T ( z , z ′ ) [ σ ^ D F 3 k − z 2 ( 1/ z − 1/ z ′ 1 ) σ ^ D F 4 k + z z ′ σ ^ D F 1 k + z 1 ( z ′ 1 − ( 1 − q T 2 / Q 2 ) z f 1 ) − 1 σ ^ D F 2 k ] + G F T ( z , z ′ ) [ σ ^ GF 3 k − z 2 ( 1/ z − 1/ z ′ 1 ) σ ^ GF 4 k + z z ′ σ ^ GF 1 k + z 1 ( z ′ 1 − ( 1 − q T 2 / Q 2 ) z f 1 ) − 1 σ ^ GF 2 k ]}] , (42)
where S 1 , 2 , 3 , 4 = sin Φ S S_{1,2,3,4}=\sin\Phi_S S 1 , 2 , 3 , 4 = sin Φ S and S 8 , 9 = cos Φ S S_{8,9}=\cos\Phi_S S 8 , 9 = cos Φ S , and σ ^ i n t k \hat{\sigma}_{\mathrm{int}}^k σ ^ int k , σ ^ k i n 1 k \hat{\sigma}_{\mathrm{kin1}}^k σ ^ kin1 k , ⋯ \cdots ⋯ etc. represent the partonic hard cross sections. In (42) we have explicitly separated the z ′ z' z ′ -dependence of the cross section for the dynamical FFs, and introduced the z ′ z' z ′ -independent partonic hard cross section, σ ^ D F 1 k \hat{\sigma}_{DF1}^k σ ^ D F 1 k , σ ^ D F 2 k \hat{\sigma}_{DF2}^k σ ^ D F 2 k , σ ^ D F 3 k \hat{\sigma}^{k}_{DF3} σ ^ D F 3 k , σ ^ D F 4 k \hat{\sigma}^{k}_{DF4} σ ^ D F 4 k for S ^ D ^ F T ( z , z ′ ) \hat{\mathcal{S}}\hat{D}_{FT}(z,z') S ^ D ^ F T ( z , z ′ ) , and likewise for S ^ G ^ F T ( z , z ′ ) \hat{\mathcal{S}}\hat{G}_{FT}(z,z') S ^ G ^ F T ( z , z ′ ) . From actual calculation, it is easy to find
σ ^ D F 3 k = − σ ^ G F 3 k , σ ^ G F 4 k = 0. (43) \hat{\sigma}^{k}_{DF3}=-\hat{\sigma}^{k}_{GF3},\qquad\hat{\sigma}^{k}_{GF4}=0. \tag*{(43)} σ ^ D F 3 k = − σ ^ GF 3 k , σ ^ GF 4 k = 0. (43)
Thanks to these relations, the EOM relation (7) and the LIR (8) allow one to absorb the contribution from the dynamical FFs with, σ ^ D F 3 k \hat{\sigma}^{k}_{DF3} σ ^ D F 3 k , σ ^ D F 4 k \hat{\sigma}^{k}_{DF4} σ ^ D F 4 k , σ ^ G F 3 k \hat{\sigma}^{k}_{GF3} σ ^ GF 3 k and σ ^ G F 4 k \hat{\sigma}^{k}_{GF4} σ ^ GF 4 k , into those from the intrinsic and the kinematical FFs. Defining the new partonic hard cross sections,
σ ^ T k ≡ σ ^ i n t k + σ ^ D F 3 k + σ ^ D F 4 k , (44) \hat{\sigma}^{k}_{T}\equiv\hat{\sigma}^{k}_{int}+\hat{\sigma}^{k}_{DF3}+\hat{\sigma}^{k}_{DF4}, \tag*{(44)} σ ^ T k ≡ σ ^ in t k + σ ^ D F 3 k + σ ^ D F 4 k , (44)
σ ^ ⊥ D k ≡ σ ^ k i n 1 k − σ ^ D F 4 k , (45) \hat{\sigma}^{k}_{\perp D}\equiv\hat{\sigma}^{k}_{kin1}-\hat{\sigma}^{k}_{DF4}, \tag*{(45)} σ ^ ⊥ D k ≡ σ ^ k in 1 k − σ ^ D F 4 k , (45)
σ ^ ⊥ k ≡ σ ^ k i n 2 k − σ ^ D F 3 k , (46) \hat{\sigma}^{k}_{\perp}\equiv\hat{\sigma}^{k}_{kin2}-\hat{\sigma}^{k}_{DF3}, \tag*{(46)} σ ^ ⊥ k ≡ σ ^ k in 2 k − σ ^ D F 3 k , (46)
one can rewrite (42) in the following form:
d 6 Δ σ t w 3 − f r a g d x b j d Q 2 d z f d q T 2 d ϕ d χ = α e m 2 α s ( − M h ) 16 π 2 x b j 2 S e p 2 Q 2 ∑ k = 1 9 A k ( ϕ − χ ) S k ∫ d x x f 1 ( x ) ∫ d z z δ ( q T 2 Q 2 − ( 1 − 1 x ^ ) ( 1 − 1 z ^ ) ) × [ D T ( z ) z σ ^ T k − { d d ( 1 / z ) D 1 T ⊥ ( 1 ) ( z ) z } σ ^ ⊥ D k − D 1 T ⊥ ( 1 ) ( z ) σ ^ ⊥ k + ∫ d z ′ z ′ 2 P ( 1 1 / z − 1 / z ′ ) { S ^ D ^ F T ( z , z ′ ) [ z ′ z σ ^ D F 1 k + 1 z ( 1 z ′ − 1 ( 1 − q T 2 / Q 2 ) z ^ f ) − 1 σ ^ D F 2 k ] + S ^ G ^ F T ( z , z ′ ) [ z ′ z σ ^ G F 1 k + 1 z ( 1 z ′ − 1 ( 1 − q T 2 / Q 2 ) z ^ f ) − 1 σ ^ G F 2 k ] } ] . (47) \begin{aligned}
\frac{d^{6}\Delta\sigma^{\mathrm{tw3-frag}}}{dx_{bj}dQ^{2}dz_{f}dq_{T}^{2}d\phi d\chi}
={}&\frac{\alpha_{\mathrm{em}}^{2}\alpha_{s}(-M_{h})}{16\pi^{2}x_{bj}^{2}S_{ep}^{2}Q^{2}}\sum_{k=1}^{9}A_{k}(\phi-\chi)S_{k}\int\frac{dx}{x}f_{1}(x)\int\frac{dz}{z}\delta\left(\frac{q_{T}^{2}}{Q^{2}}-\left(1-\frac{1}{\hat{x}}\right)\left(1-\frac{1}{\hat{z}}\right)\right)\\
&\times\left[\frac{D_{T}(z)}{z}\hat{\sigma}^{k}_{T}-\left\{\frac{d}{d(1/z)}\frac{D_{1T}^{\perp(1)}(z)}{z}\right\}\hat{\sigma}^{k}_{\perp D}-D_{1T}^{\perp(1)}(z)\hat{\sigma}^{k}_{\perp}\right.\\
&\left.+\int\frac{dz'}{z'^{2}}\mathcal{P}\left(\frac{1}{1/z-1/z'}\right)\left\{\hat{\mathcal{S}}\hat{D}_{FT}(z,z')\left[\frac{z'}{z}\hat{\sigma}^{k}_{DF1}+\frac{1}{z}\left(\frac{1}{z'}-\frac{1}{(1-q_{T}^{2}/Q^{2})\hat{z}_{f}}\right)^{-1}\hat{\sigma}^{k}_{DF2}\right]\right.\right.\\
&\left.\left.+\hat{\mathcal{S}}\hat{G}_{FT}(z,z')\left[\frac{z'}{z}\hat{\sigma}^{k}_{GF1}+\frac{1}{z}\left(\frac{1}{z'}-\frac{1}{(1-q_{T}^{2}/Q^{2})\hat{z}_{f}}\right)^{-1}\hat{\sigma}^{k}_{GF2}\right]\right\}\right]. \tag*{(47)}
\end{aligned} d x bj d Q 2 d z f d q T 2 d ϕ d χ d 6 Δ σ tw3 − frag = 16 π 2 x bj 2 S e p 2 Q 2 α em 2 α s ( − M h ) k = 1 ∑ 9 A k ( ϕ − χ ) S k ∫ x d x f 1 ( x ) ∫ z d z δ ( Q 2 q T 2 − ( 1 − x ^ 1 ) ( 1 − z ^ 1 ) ) × [ z D T ( z ) σ ^ T k − { d ( 1/ z ) d z D 1 T ⊥ ( 1 ) ( z ) } σ ^ ⊥ D k − D 1 T ⊥ ( 1 ) ( z ) σ ^ ⊥ k + ∫ z ′ 2 d z ′ P ( 1/ z − 1/ z ′ 1 ) { S ^ D ^ F T ( z , z ′ ) [ z z ′ σ ^ D F 1 k + z 1 ( z ′ 1 − ( 1 − q T 2 / Q 2 ) z ^ f 1 ) − 1 σ ^ D F 2 k ] + S ^ G ^ F T ( z , z ′ ) [ z z ′ σ ^ GF 1 k + z 1 ( z ′ 1 − ( 1 − q T 2 / Q 2 ) z ^ f 1 ) − 1 σ ^ GF 2 k ] } ] . (47)
This is the final form we use to present the twist-3 quark FF contribution to e p → e Λ ↑ X ep\to e\Lambda^{\uparrow}X e p → e Λ ↑ X shown in Figure 3 .
In order to present the partonic hard cross sections in (47) , it is convenient to introduce the quark FF contribution to the twist-2 unpolarized cross section for e p → e Λ X ep\to e\Lambda X e p → e Λ X . It reads
d 6 σ u n p o l d x b j d Q 2 d z f d q T 2 d ϕ d χ = α e m 2 α s 16 π 2 x b j 2 S e p 2 Q 2 ∑ k = 1 4 A k ( ϕ − χ ) ∫ d x x f 1 ( x ) ∫ d z z D q ( z ) δ ( q T 2 Q 2 − ( 1 − 1 x ^ ) ( 1 − 1 z ^ ) ) C F σ ^ U k . (48) \frac{d^{6}\sigma^{\mathrm{unpol}}}{dx_{bj}dQ^{2}dz_{f}dq_{T}^{2}d\phi d\chi}
=\frac{\alpha_{\mathrm{em}}^{2}\alpha_{s}}{16\pi^{2}x_{bj}^{2}S_{ep}^{2}Q^{2}}\sum_{k=1}^{4}A_{k}(\phi-\chi)\int\frac{dx}{x}f_{1}(x)\int\frac{dz}{z}D_{q}(z)\delta\left(\frac{q_{T}^{2}}{Q^{2}}-\left(1-\frac{1}{\hat{x}}\right)\left(1-\frac{1}{\hat{z}}\right)\right)C_{F}\hat{\sigma}^{k}_{U}. \tag*{(48)} d x bj d Q 2 d z f d q T 2 d ϕ d χ d 6 σ unpol = 16 π 2 x bj 2 S e p 2 Q 2 α em 2 α s k = 1 ∑ 4 A k ( ϕ − χ ) ∫ x d x f 1 ( x ) ∫ z d z D q ( z ) δ ( Q 2 q T 2 − ( 1 − x ^ 1 ) ( 1 − z ^ 1 ) ) C F σ ^ U k . (48)
where D q ( z ) D_{q}(z) D q ( z ) denotes the twist-2 unpolarized quark FF, and C F σ ^ U k C_{F}\hat{\sigma}^{k}_{U} C F σ ^ U k represents the partonic hard cross sections which depend on whether f 1 ( x ) = q ( x ) f_{1}(x)=q(x) f 1 ( x ) = q ( x ) (quark DF) or f 1 ( x ) = G ( x ) f_{1}(x)=G(x) f 1 ( x ) = G ( x ) (gluon DF). We will see below that some of the hard cross sections in (47) are related to σ ^ U k \hat{\sigma}^{k}_{U} σ ^ U k .
Below we give LO Feynman diagrams for the hard part and the results for the hard cross sections σ ^ T , ⊥ D , ⊥ \hat{\sigma}_{T,\perp D,\perp} σ ^ T , ⊥ D , ⊥ and σ ^ D F 1 , G F 1 , D F 2 , G F 2 \hat{\sigma}_{DF1,GF1,DF2,GF2} σ ^ D F 1 , GF 1 , D F 2 , GF 2 , which are the functions of Q Q Q , q T q_{T} q T , x ^ = x b j / x \hat{x}=x_{bj}/x x ^ = x bj / x and z ^ = z f / z \hat{z}=z_{f}/z z ^ = z f / z .
(I) γ q → q g \gamma q \to qg γ q → q g channel:
The diagrams for the hard part in this channel are shown in Figure 4 . It is convenient to present some of the hard cross sections in (47) in the following form:
Figure 4. LO diagrams in the γ q → q g \gamma q \to qg γ q → q g channel contributing to the hard parts S ρ σ ( k ) S_{\rho\sigma}(k) S ρ σ ( k ) and S α , ρ σ ( z ′ , z ) S_{\alpha,\rho\sigma}(z',z) S α , ρ σ ( z ′ , z ) in (41) . A cross in an oval in each diagram represents the fragmentation insertion. For S α , ρ σ ( z ′ , z ) S_{\alpha,\rho\sigma}(z',z) S α , ρ σ ( z ′ , z ) , an extra gluon line connecting the cross in the oval and each dot in the diagram should be added.
σ ^ T k = C F σ ^ 1 k − 1 N 1 − z ^ q T z ^ σ ^ U k , (49) \hat{\sigma}_{T}^{k}=C_{F}\hat{\sigma}_{1}^{k}-\frac{1}{N}\frac{1-\hat{z}}{q_{T}\hat{z}}\hat{\sigma}_{U}^{k}, \tag*{(49)} σ ^ T k = C F σ ^ 1 k − N 1 q T z ^ 1 − z ^ σ ^ U k , (49)
σ ^ ⊥ D k = 1 N 1 − z ^ q T z ^ σ ^ U k , (50) \hat{\sigma}_{\perp D}^{k}=\frac{1}{N}\frac{1-\hat{z}}{q_{T}\hat{z}}\hat{\sigma}_{U}^{k}, \tag*{(50)} σ ^ ⊥ D k = N 1 q T z ^ 1 − z ^ σ ^ U k , (50)
σ ^ ⊥ k = C F σ ^ 2 k , (51) \hat{\sigma}_{\perp}^{k}=C_{F}\hat{\sigma}_{2}^{k}, \tag*{(51)} σ ^ ⊥ k = C F σ ^ 2 k , (51)
σ ^ D F 2 k = ( 1 − z ^ ) ( C F 1 − x ^ − z ^ − 1 2 N z ^ ) σ ^ 2 k , (52) \hat{\sigma}_{DF2}^{k}=(1-\hat{z})\left(\frac{C_{F}}{1-\hat{x}-\hat{z}}-\frac{1}{2N\hat{z}}\right)\hat{\sigma}_{2}^{k}, \tag*{(52)} σ ^ D F 2 k = ( 1 − z ^ ) ( 1 − x ^ − z ^ C F − 2 N z ^ 1 ) σ ^ 2 k , (52)
σ ^ G F 2 k = ( 1 − z ^ ) ( C F 1 − x ^ − z ^ − 1 2 N z ^ ) σ ^ 3 k , (53) \hat{\sigma}_{GF2}^{k}=(1-\hat{z})\left(\frac{C_{F}}{1-\hat{x}-\hat{z}}-\frac{1}{2N\hat{z}}\right)\hat{\sigma}_{3}^{k}, \tag*{(53)} σ ^ GF 2 k = ( 1 − z ^ ) ( 1 − x ^ − z ^ C F − 2 N z ^ 1 ) σ ^ 3 k , (53)
where σ ^ U k \hat{\sigma}_{U}^{k} σ ^ U k s are the unpolarized partonic cross sections introduced in (48) for the unpolarized quark distribution f 1 ( x ) = q ( x ) f_{1}(x)=q(x) f 1 ( x ) = q ( x ) and σ ^ 1 , 2 , 3 k \hat{\sigma}_{1,2,3}^{k} σ ^ 1 , 2 , 3 k are newly introduced partonic cross sections, and hence specification of σ ^ U k \hat{\sigma}_{U}^{k} σ ^ U k , σ ^ 1 , 2 , 3 k \hat{\sigma}_{1,2,3}^{k} σ ^ 1 , 2 , 3 k , σ ^ D F 1 k \hat{\sigma}_{DF1}^{k} σ ^ D F 1 k and σ ^ G F 1 k \hat{\sigma}_{GF1}^{k} σ ^ GF 1 k determines the cross sections in (47) . We shall now give those in the γ q → q g \gamma q \to qg γ q → q g channel.
(1) Twist-2 unpolarized cross section:
σ ^ U 1 = 2 ( 2 − 2 x ^ + x ^ 2 + z ^ ( − 2 + 8 x ^ − 6 x ^ 2 ) + z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 ) ) ( 1 − z ^ ) ( 1 − x ^ ) , \hat{\sigma}_{U}^{1}=\frac{2(2-2\hat{x}+\hat{x}^{2}+\hat{z}(-2+8\hat{x}-6\hat{x}^{2})+\hat{z}^{2}(1-6\hat{x}+6\hat{x}^{2}))}{(1-\hat{z})(1-\hat{x})}, σ ^ U 1 = ( 1 − z ^ ) ( 1 − x ^ ) 2 ( 2 − 2 x ^ + x ^ 2 + z ^ ( − 2 + 8 x ^ − 6 x ^ 2 ) + z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 )) ,
σ ^ U 2 = 8 x ^ z ^ , \hat{\sigma}_{U}^{2}=8\hat{x}\hat{z}, σ ^ U 2 = 8 x ^ z ^ ,
σ ^ U 3 = 4 Q ( 1 − x ^ + z ^ ( − 1 + 2 x ^ ) ) q T , \hat{\sigma}_{U}^{3}=\frac{4Q(1-\hat{x}+\hat{z}(-1+2\hat{x}))}{q_{T}}, σ ^ U 3 = q T 4 Q ( 1 − x ^ + z ^ ( − 1 + 2 x ^ )) ,
σ ^ U 4 = 4 x ^ z ^ . (54) \hat{\sigma}_{U}^{4}=4\hat{x}\hat{z}. \tag*{(54)} σ ^ U 4 = 4 x ^ z ^ . (54)
σ ^ U 8 , 9 = 0 \hat{\sigma}_{U}^{8,9}=0 σ ^ U 8 , 9 = 0 is understood in (49) and (50) .
(2) k = 1 k=1 k = 1 :
σ ^ 1 1 = 4 ( 1 − 9 x ^ + 14 x ^ 2 − 6 x ^ 3 + 2 z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 ) + z ^ ( − 4 + 23 x ^ − 30 x ^ 2 + 12 x ^ 3 ) ) q T ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) , σ ^ 2 1 = − 4 ( ( − 1 + x ^ ) 3 − 3 z ^ ( − 1 + x ^ ) 2 ( − 1 + 2 x ^ ) + z ^ 3 ( 1 − 6 x ^ + 6 x ^ 2 ) + z ^ 2 ( − 3 + 16 x ^ − 18 x ^ 2 + 6 x ^ 3 ) ) q T ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) , σ ^ 3 1 = − 4 ( ( − 1 + x ^ ) 3 − 3 z ^ ( − 1 + x ^ ) 2 ( − 1 + 2 x ^ ) + z ^ 3 ( 1 − 6 x ^ + 6 x ^ 2 ) + z ^ 2 ( − 3 + 14 x ^ − 18 x ^ 2 + 6 x ^ 3 ) ) q T ( − 1 + x ^ ) ( − 1 + z ^ ) ( − 1 + z ^ + x ^ ) , σ ^ D F 1 1 = 1 N 2 ( − 1 + ( − 1 + z ^ ) x ^ ) q T z ^ ( − 1 + x ^ ) , σ ^ G F 1 1 = − 1 N 2 ( 1 + ( − 1 + z ^ ) x ^ ) q T z ^ ( − 1 + x ^ ) . (55) \begin{aligned}
\hat{\sigma}_{1}^{1} &= \frac{4(1-9\hat{x}+14\hat{x}^{2}-6\hat{x}^{3}+2\hat{z}^{2}(1-6\hat{x}+6\hat{x}^{2})+\hat{z}(-4+23\hat{x}-30\hat{x}^{2}+12\hat{x}^{3}))}{q_{T}(-1+\hat{x})(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{2}^{1} &= -\frac{4((-1+\hat{x})^{3}-3\hat{z}(-1+\hat{x})^{2}(-1+2\hat{x})+\hat{z}^{3}(1-6\hat{x}+6\hat{x}^{2})+\hat{z}^{2}(-3+16\hat{x}-18\hat{x}^{2}+6\hat{x}^{3}))}{q_{T}(-1+\hat{z})(-1+\hat{x})(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{3}^{1} &= -\frac{4((-1+\hat{x})^{3}-3\hat{z}(-1+\hat{x})^{2}(-1+2\hat{x})+\hat{z}^{3}(1-6\hat{x}+6\hat{x}^{2})+\hat{z}^{2}(-3+14\hat{x}-18\hat{x}^{2}+6\hat{x}^{3}))}{q_{T}(-1+\hat{x})(-1+\hat{z})(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{DF1}^{1} &= \frac{1}{N}\frac{2(-1+(-1+\hat{z})\hat{x})}{q_{T}\hat{z}(-1+\hat{x})},\\
\hat{\sigma}_{GF1}^{1} &= -\frac{1}{N}\frac{2(1+(-1+\hat{z})\hat{x})}{q_{T}\hat{z}(-1+\hat{x})}. \tag*{(55)}
\end{aligned} σ ^ 1 1 σ ^ 2 1 σ ^ 3 1 σ ^ D F 1 1 σ ^ GF 1 1 = q T ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) 4 ( 1 − 9 x ^ + 14 x ^ 2 − 6 x ^ 3 + 2 z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 ) + z ^ ( − 4 + 23 x ^ − 30 x ^ 2 + 12 x ^ 3 )) , = − q T ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) 4 (( − 1 + x ^ ) 3 − 3 z ^ ( − 1 + x ^ ) 2 ( − 1 + 2 x ^ ) + z ^ 3 ( 1 − 6 x ^ + 6 x ^ 2 ) + z ^ 2 ( − 3 + 16 x ^ − 18 x ^ 2 + 6 x ^ 3 )) , = − q T ( − 1 + x ^ ) ( − 1 + z ^ ) ( − 1 + z ^ + x ^ ) 4 (( − 1 + x ^ ) 3 − 3 z ^ ( − 1 + x ^ ) 2 ( − 1 + 2 x ^ ) + z ^ 3 ( 1 − 6 x ^ + 6 x ^ 2 ) + z ^ 2 ( − 3 + 14 x ^ − 18 x ^ 2 + 6 x ^ 3 )) , = N 1 q T z ^ ( − 1 + x ^ ) 2 ( − 1 + ( − 1 + z ^ ) x ^ ) , = − N 1 q T z ^ ( − 1 + x ^ ) 2 ( 1 + ( − 1 + z ^ ) x ^ ) . (55)
(3) k = 2 k=2 k = 2 :
σ ^ 1 2 = 16 ( − 1 + 2 z ^ ) x ^ q T , σ ^ 2 2 = − 16 x ^ z ^ q T , σ ^ 3 2 = − 16 z ^ x ^ q T , σ ^ D F 1 2 = 0 , σ ^ G F 1 2 = 0. (56) \begin{aligned}
\hat{\sigma}_{1}^{2} &= \frac{16(-1+2\hat{z})\hat{x}}{q_{T}},\\
\hat{\sigma}_{2}^{2} &= -\frac{16\hat{x}\hat{z}}{q_{T}},\\
\hat{\sigma}_{3}^{2} &= -\frac{16\hat{z}\hat{x}}{q_{T}},\\
\hat{\sigma}_{DF1}^{2} &= 0,\\
\hat{\sigma}_{GF1}^{2} &= 0. \tag*{(56)}
\end{aligned} σ ^ 1 2 σ ^ 2 2 σ ^ 3 2 σ ^ D F 1 2 σ ^ GF 1 2 = q T 16 ( − 1 + 2 z ^ ) x ^ , = − q T 16 x ^ z ^ , = − q T 16 z ^ x ^ , = 0 , = 0. (56)
(4) k = 3 k=3 k = 3 :
σ ^ 1 3 = 4 x ^ ( ( − 1 + x ^ ) 2 + z ^ 3 ( − 4 + 8 x ^ ) + 2 z ^ 2 ( 5 − 10 x ^ + 4 x ^ 2 ) − 2 z ^ ( 3 − 7 x ^ + 4 x ^ 2 ) ) Q ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) , σ ^ 2 3 = − 4 x ^ z ^ ( − 2 ( − 1 + x ^ ) 2 + z ^ 2 ( − 2 + 4 x ^ ) + z ^ ( 5 − 8 x ^ + 4 x ^ 2 ) ) Q ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) , σ ^ 3 3 = − 4 Q ( − 2 ( − 1 + x ^ ) 2 + z ^ 2 ( − 2 + 4 x ^ ) + z ^ ( 3 − 8 x ^ + 4 x ^ 2 ) ) q T 2 ( − 1 + z ^ + x ^ ) , σ ^ D F 1 3 = 1 N 2 x ^ Q ( − 1 + x ^ ) , σ ^ G F 1 3 = − 1 N 2 x ^ Q ( − 1 + x ^ ) . (57) \begin{aligned}
\hat{\sigma}_{1}^{3} &= \frac{4\hat{x}((-1+\hat{x})^{2}+\hat{z}^{3}(-4+8\hat{x})+2\hat{z}^{2}(5-10\hat{x}+4\hat{x}^{2})-2\hat{z}(3-7\hat{x}+4\hat{x}^{2}))}{Q(-1+\hat{z})(-1+\hat{x})(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{2}^{3} &= -\frac{4\hat{x}\hat{z}(-2(-1+\hat{x})^{2}+\hat{z}^{2}(-2+4\hat{x})+\hat{z}(5-8\hat{x}+4\hat{x}^{2}))}{Q(-1+\hat{z})(-1+\hat{x})(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{3}^{3} &= -\frac{4Q(-2(-1+\hat{x})^{2}+\hat{z}^{2}(-2+4\hat{x})+\hat{z}(3-8\hat{x}+4\hat{x}^{2}))}{q_{T}^{2}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{DF1}^{3} &= \frac{1}{N}\frac{2\hat{x}}{Q(-1+\hat{x})},\\
\hat{\sigma}_{GF1}^{3} &= -\frac{1}{N}\frac{2\hat{x}}{Q(-1+\hat{x})}. \tag*{(57)}
\end{aligned} σ ^ 1 3 σ ^ 2 3 σ ^ 3 3 σ ^ D F 1 3 σ ^ GF 1 3 = Q ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) 4 x ^ (( − 1 + x ^ ) 2 + z ^ 3 ( − 4 + 8 x ^ ) + 2 z ^ 2 ( 5 − 10 x ^ + 4 x ^ 2 ) − 2 z ^ ( 3 − 7 x ^ + 4 x ^ 2 )) , = − Q ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) 4 x ^ z ^ ( − 2 ( − 1 + x ^ ) 2 + z ^ 2 ( − 2 + 4 x ^ ) + z ^ ( 5 − 8 x ^ + 4 x ^ 2 )) , = − q T 2 ( − 1 + z ^ + x ^ ) 4 Q ( − 2 ( − 1 + x ^ ) 2 + z ^ 2 ( − 2 + 4 x ^ ) + z ^ ( 3 − 8 x ^ + 4 x ^ 2 )) , = N 1 Q ( − 1 + x ^ ) 2 x ^ , = − N 1 Q ( − 1 + x ^ ) 2 x ^ . (57)
(5) k = 4 k=4 k = 4 :
σ ^ 1 4 = 4 4 x ^ z ^ 2 − 2 ( − 1 + x ^ ) x + z ^ ( 1 − 6 x ^ + 4 x ^ 2 ) q T ( − 1 + z ^ + x ^ ) , σ ^ 2 4 = − 4 z ^ ( 1 + 2 ( − 1 + z ^ ) x ^ + 2 x ^ 2 ) q T ( − 1 + z ^ + x ^ ) , σ ^ 3 4 = − 4 z ^ ( − 1 + 2 ( − 1 + z ^ ) x ^ + 2 x ^ 2 ) q T ( − 1 + z ^ + x ^ ) , σ ^ D F 1 4 = 1 N 2 q T , σ ^ G F 1 4 = − 1 N 2 q T . (58) \begin{aligned}
\hat{\sigma}_1^4 &= 4\frac{4\hat{x}\hat{z}^2-2(-1+\hat{x})x+\hat{z}(1-6\hat{x}+4\hat{x}^2)}{q_T(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_2^4 &= -\frac{4\hat{z}(1+2(-1+\hat{z})\hat{x}+2\hat{x}^2)}{q_T(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_3^4 &= -\frac{4\hat{z}(-1+2(-1+\hat{z})\hat{x}+2\hat{x}^2)}{q_T(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{DF1}^4 &= \frac{1}{N}\frac{2}{q_T},\\
\hat{\sigma}_{GF1}^4 &= -\frac{1}{N}\frac{2}{q_T}. \tag*{(58)}
\end{aligned} σ ^ 1 4 σ ^ 2 4 σ ^ 3 4 σ ^ D F 1 4 σ ^ GF 1 4 = 4 q T ( − 1 + z ^ + x ^ ) 4 x ^ z ^ 2 − 2 ( − 1 + x ^ ) x + z ^ ( 1 − 6 x ^ + 4 x ^ 2 ) , = − q T ( − 1 + z ^ + x ^ ) 4 z ^ ( 1 + 2 ( − 1 + z ^ ) x ^ + 2 x ^ 2 ) , = − q T ( − 1 + z ^ + x ^ ) 4 z ^ ( − 1 + 2 ( − 1 + z ^ ) x ^ + 2 x ^ 2 ) , = N 1 q T 2 , = − N 1 q T 2 . (58)
(6) k = 8 k=8 k = 8 :
σ ^ 1 8 = 4 x ^ ( − ( − 1 + x ^ ) 2 + 2 z ^ ( − 1 + x ^ ) 2 + 2 x ^ z ^ 2 ) Q ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) , σ ^ 2 8 = − 4 x ^ z ^ 2 Q ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) , σ ^ 3 8 = 4 z ^ Q q T 2 ( − 1 + z ^ + x ^ ) , σ ^ D F 1 8 = − 1 N 2 x ^ Q ( 1 − x ^ ) , σ ^ G F 1 8 = 1 N 2 x ^ Q ( 1 − x ^ ) . (59) \begin{aligned}
\hat{\sigma}_1^8 &= \frac{4\hat{x}(-(-1+\hat{x})^2+2\hat{z}(-1+\hat{x})^2+2\hat{x}\hat{z}^2)}{Q(-1+\hat{z})(-1+\hat{x})(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_2^8 &= -\frac{4\hat{x}\hat{z}^2}{Q(-1+\hat{z})(-1+\hat{x})(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_3^8 &= \frac{4\hat{z}Q}{q_T^2(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{DF1}^8 &= -\frac{1}{N}\frac{2\hat{x}}{Q(1-\hat{x})},\\
\hat{\sigma}_{GF1}^8 &= \frac{1}{N}\frac{2\hat{x}}{Q(1-\hat{x})}. \tag*{(59)}
\end{aligned} σ ^ 1 8 σ ^ 2 8 σ ^ 3 8 σ ^ D F 1 8 σ ^ GF 1 8 = Q ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) 4 x ^ ( − ( − 1 + x ^ ) 2 + 2 z ^ ( − 1 + x ^ ) 2 + 2 x ^ z ^ 2 ) , = − Q ( − 1 + z ^ ) ( − 1 + x ^ ) ( − 1 + z ^ + x ^ ) 4 x ^ z ^ 2 , = q T 2 ( − 1 + z ^ + x ^ ) 4 z ^ Q , = − N 1 Q ( 1 − x ^ ) 2 x ^ , = N 1 Q ( 1 − x ^ ) 2 x ^ . (59)
(7) k = 9 k=9 k = 9 :
σ 1 9 = 4 { z ^ + 2 x ^ z ^ + 2 ( − 1 + x ^ ) x ^ } q T ( − 1 + z ^ + x ^ ) , σ 2 9 = − 4 z ^ q T ( − 1 + z ^ + x ^ ) , σ 3 9 = 4 z ^ q T ( − 1 + z ^ + x ^ ) , σ D F 1 9 = 1 N 2 q T , σ G F 1 9 = − 1 N 2 q T . (60) \begin{aligned}
\sigma_1^9 &= \frac{4\{\hat{z}+2\hat{x}\hat{z}+2(-1+\hat{x})\hat{x}\}}{q_T(-1+\hat{z}+\hat{x})},\\
\sigma_2^9 &= -\frac{4\hat{z}}{q_T(-1+\hat{z}+\hat{x})},\\
\sigma_3^9 &= \frac{4\hat{z}}{q_T(-1+\hat{z}+\hat{x})},\\
\sigma_{DF1}^9 &= \frac{1}{N}\frac{2}{q_T},\\
\sigma_{GF1}^9 &= -\frac{1}{N}\frac{2}{q_T}. \tag*{(60)}
\end{aligned} σ 1 9 σ 2 9 σ 3 9 σ D F 1 9 σ GF 1 9 = q T ( − 1 + z ^ + x ^ ) 4 { z ^ + 2 x ^ z ^ + 2 ( − 1 + x ^ ) x ^ } , = − q T ( − 1 + z ^ + x ^ ) 4 z ^ , = q T ( − 1 + z ^ + x ^ ) 4 z ^ , = N 1 q T 2 , = − N 1 q T 2 . (60)
(II) γ g → q q ˉ \gamma g \to q\bar{q} γ g → q q ˉ channel:
Diagrams for the hard part in this channel are shown in Figure 5 . It is convenient to present some of the hard cross sections in (47) in the following form:
Figure 5. LO diagrams in the γ g → q q ˉ \gamma g \to q\bar{q} γ g → q q ˉ channel contributing to the hard parts S ρ σ ( k ) S_{\rho\sigma}(k) S ρ σ ( k ) and S α , ρ σ L ( z ′ , z ) S^{L}_{\alpha,\rho\sigma}(z',z) S α , ρ σ L ( z ′ , z ) in (41) . A cross in an oval in each diagram represents the fragmentation insertion. For S α , ρ σ L ( z ′ , z ) S^{L}_{\alpha,\rho\sigma}(z',z) S α , ρ σ L ( z ′ , z ) , an extra gluon line connecting the cross in the oval and each dot in the diagram should be added.
σ ^ T k = σ ^ 1 k − 1 N 2 − 1 1 − z ^ q T z ^ σ ^ U k , (61) \hat{\sigma}^{k}_{T}=\hat{\sigma}^{k}_{1}-\frac{1}{N^{2}-1}\frac{1-\hat{z}}{q_{T}\hat{z}}\hat{\sigma}^{k}_{U}, \tag*{(61)} σ ^ T k = σ ^ 1 k − N 2 − 1 1 q T z ^ 1 − z ^ σ ^ U k , (61)
σ ^ L D k = N 2 C F 1 − z ^ q T z ^ σ ^ U k , (62) \hat{\sigma}^{k}_{LD}=\frac{N}{2C_{F}}\frac{1-\hat{z}}{q_{T}\hat{z}}\hat{\sigma}^{k}_{U}, \tag*{(62)} σ ^ L D k = 2 C F N q T z ^ 1 − z ^ σ ^ U k , (62)
σ ^ D F 2 k = 2 x ^ 1 − x ^ − z ^ ( 1 − x ^ 1 − x ^ − z ^ + 1 N 2 − 1 ) σ ^ 2 k , (63) \hat{\sigma}^{k}_{DF2}=\frac{2\hat{x}}{1-\hat{x}-\hat{z}}\left(\frac{1-\hat{x}}{1-\hat{x}-\hat{z}}+\frac{1}{N^{2}-1}\right)\hat{\sigma}^{k}_{2}, \tag*{(63)} σ ^ D F 2 k = 1 − x ^ − z ^ 2 x ^ ( 1 − x ^ − z ^ 1 − x ^ + N 2 − 1 1 ) σ ^ 2 k , (63)
σ ^ G F 2 k = 2 x ^ 1 − x ^ − z ^ ( 1 − x ^ 1 − x ^ − z ^ + 1 N 2 − 1 ) σ ^ 3 k , (64) \hat{\sigma}^{k}_{GF2}=\frac{2\hat{x}}{1-\hat{x}-\hat{z}}\left(\frac{1-\hat{x}}{1-\hat{x}-\hat{z}}+\frac{1}{N^{2}-1}\right)\hat{\sigma}^{k}_{3}, \tag*{(64)} σ ^ GF 2 k = 1 − x ^ − z ^ 2 x ^ ( 1 − x ^ − z ^ 1 − x ^ + N 2 − 1 1 ) σ ^ 3 k , (64)
where σ ^ U k \hat{\sigma}^{k}_{U} σ ^ U k ’s are the unpolarized partonic cross sections in (48) for the unpolarized gluon distribution f 1 ( x ) = G ( x ) f_{1}(x)=G(x) f 1 ( x ) = G ( x ) , and σ ^ 1 , 2 , 3 k \hat{\sigma}^{k}_{1,2,3} σ ^ 1 , 2 , 3 k are newly introduced here. We will give σ ^ T k \hat{\sigma}^{k}_{T} σ ^ T k , σ ^ 1 , 2 , 3 k \hat{\sigma}^{k}_{1,2,3} σ ^ 1 , 2 , 3 k , σ ^ L k \hat{\sigma}^{k}_{L} σ ^ L k , σ ^ D F 1 k \hat{\sigma}^{k}_{DF1} σ ^ D F 1 k and σ ^ G F 1 k \hat{\sigma}^{k}_{GF1} σ ^ GF 1 k below.
(1) Twist-2 unpolarized cross sections:
σ ^ U 1 = 2 1 − 2 x ^ + 2 x ^ 2 − 2 z ^ ( 1 − 6 x ^ + 6 x ^ 2 ) + 2 z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 ) z ^ ( 1 − z ^ ) , \hat{\sigma}^{1}_{U}=2\frac{1-2\hat{x}+2\hat{x}^{2}-2\hat{z}(1-6\hat{x}+6\hat{x}^{2})+2\hat{z}^{2}(1-6\hat{x}+6\hat{x}^{2})}{\hat{z}(1-\hat{z})}, σ ^ U 1 = 2 z ^ ( 1 − z ^ ) 1 − 2 x ^ + 2 x ^ 2 − 2 z ^ ( 1 − 6 x ^ + 6 x ^ 2 ) + 2 z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 ) ,
σ ^ U 2 = 16 x ^ ( 1 − x ^ ) , \hat{\sigma}^{2}_{U}=16\hat{x}(1-\hat{x}), σ ^ U 2 = 16 x ^ ( 1 − x ^ ) ,
σ ^ U 3 = 8 x ^ q T ( 1 − 2 z ^ ) ( 1 − 2 x ^ ) Q ( 1 − z ^ ) , σ ^ U 4 = 8 x ^ ( 1 − x ^ ) , (65) \hat{\sigma}^{3}_{U}=\frac{8\hat{x}q_{T}(1-2\hat{z})(1-2\hat{x})}{Q(1-\hat{z})}, \qquad
\hat{\sigma}^{4}_{U}=8\hat{x}(1-\hat{x}), \tag*{(65)} σ ^ U 3 = Q ( 1 − z ^ ) 8 x ^ q T ( 1 − 2 z ^ ) ( 1 − 2 x ^ ) , σ ^ U 4 = 8 x ^ ( 1 − x ^ ) , (65)
where δ ^ U 8 , 9 = 0 \hat{\delta}^{8,9}_{U}=0 δ ^ U 8 , 9 = 0 is understood in (61) and (62) .
(2) k = 1 k=1 k = 1 :
σ ^ 1 1 = − 2 ( − 1 + x ^ ) { − 2 + 5 x ^ − 3 x ^ 2 + z ^ 2 ( − 1 + 6 x ^ 2 ) + z ^ ( 4 − 11 x ^ + 6 x ^ 2 ) } q T z ^ 2 ( − 1 + z ^ + x ^ ) , σ ^ ⊥ 1 = 2 q T ( − 1 + z ^ ) z ^ 2 ( − 1 + z ^ + x ^ ) { x ^ − x ^ 3 + 2 z ^ 4 ( 1 − 6 x ^ + 6 x ^ 2 ) + z ^ 2 ( 4 − 30 x ^ + 55 x ^ 2 − 24 x ^ 3 ) + z ^ ( − 1 + 5 x ^ − 15 x ^ 2 + 9 x ^ 3 ) + z ^ 3 ( − 5 + 37 x ^ − 54 x ^ 2 + 18 x ^ 3 ) } , σ ^ D F 1 1 = − 1 N 2 − 1 2 ( 3 z ^ ( 1 − 2 x ^ ) x ^ + x ^ ( 1 + x ^ ) + z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 ) ) q T z ^ 2 , σ ^ G F 1 1 = − 1 N 2 − 1 2 ( z ^ ( 5 − 6 x ^ ) x ^ + ( − 1 + x ^ ) x ^ + z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 ) ) q T z ^ 2 , σ ^ 2 1 = 1 q T x ^ z ^ 2 { ( − 1 + x ^ ) 2 ( − 1 + 2 x ^ ) − 3 z ^ ( − 1 + x ^ ) 2 ( − 1 + 4 x ^ ) + z ^ 3 ( 1 − 6 x ^ + 6 x ^ 2 ) + 3 z ^ 2 ( − 1 + 7 x ^ − 10 x ^ 2 + 4 x ^ 3 ) } , σ ^ 3 1 = − ( − 1 + x ^ ) 2 + 3 z ^ ( − 1 + x ^ ) 2 + z ^ 2 ( − 3 + 9 x ^ − 6 x ^ 2 ) + z ^ 3 ( 1 − 6 x ^ + 6 x ^ 2 ) q T x ^ z ^ 2 . (66) \begin{aligned}
\hat{\sigma}^{1}_{1}&=-\frac{2(-1+\hat{x})\{-2+5\hat{x}-3\hat{x}^{2}+\hat{z}^{2}(-1+6\hat{x}^{2})+\hat{z}(4-11\hat{x}+6\hat{x}^{2})\}}{q_{T}\hat{z}^{2}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}^{1}_{\perp}&=\frac{2}{q_{T}(-1+\hat{z})\hat{z}^{2}(-1+\hat{z}+\hat{x})}\{\hat{x}-\hat{x}^{3}+2\hat{z}^{4}(1-6\hat{x}+6\hat{x}^{2})+\hat{z}^{2}(4-30\hat{x}+55\hat{x}^{2}-24\hat{x}^{3})\\
&\qquad+\hat{z}(-1+5\hat{x}-15\hat{x}^{2}+9\hat{x}^{3})+\hat{z}^{3}(-5+37\hat{x}-54\hat{x}^{2}+18\hat{x}^{3})\},\\
\hat{\sigma}^{1}_{DF1}&=-\frac{1}{N^{2}-1}\frac{2(3\hat{z}(1-2\hat{x})\hat{x}+\hat{x}(1+\hat{x})+\hat{z}^{2}(1-6\hat{x}+6\hat{x}^{2}))}{q_{T}\hat{z}^{2}},\\
\hat{\sigma}^{1}_{GF1}&=-\frac{1}{N^{2}-1}\frac{2(\hat{z}(5-6\hat{x})\hat{x}+(-1+\hat{x})\hat{x}+\hat{z}^{2}(1-6\hat{x}+6\hat{x}^{2}))}{q_{T}\hat{z}^{2}},\\
\hat{\sigma}^{1}_{2}&=\frac{1}{q_{T}\hat{x}\hat{z}^{2}}\{(-1+\hat{x})^{2}(-1+2\hat{x})-3\hat{z}(-1+\hat{x})^{2}(-1+4\hat{x})\\
&\qquad+\hat{z}^{3}(1-6\hat{x}+6\hat{x}^{2})+3\hat{z}^{2}(-1+7\hat{x}-10\hat{x}^{2}+4\hat{x}^{3})\},\\
\hat{\sigma}^{1}_{3}&=\frac{-(-1+\hat{x})^{2}+3\hat{z}(-1+\hat{x})^{2}+\hat{z}^{2}(-3+9\hat{x}-6\hat{x}^{2})+\hat{z}^{3}(1-6\hat{x}+6\hat{x}^{2})}{q_{T}\hat{x}\hat{z}^{2}}. \tag*{(66)}
\end{aligned} σ ^ 1 1 σ ^ ⊥ 1 σ ^ D F 1 1 σ ^ GF 1 1 σ ^ 2 1 σ ^ 3 1 = − q T z ^ 2 ( − 1 + z ^ + x ^ ) 2 ( − 1 + x ^ ) { − 2 + 5 x ^ − 3 x ^ 2 + z ^ 2 ( − 1 + 6 x ^ 2 ) + z ^ ( 4 − 11 x ^ + 6 x ^ 2 )} , = q T ( − 1 + z ^ ) z ^ 2 ( − 1 + z ^ + x ^ ) 2 { x ^ − x ^ 3 + 2 z ^ 4 ( 1 − 6 x ^ + 6 x ^ 2 ) + z ^ 2 ( 4 − 30 x ^ + 55 x ^ 2 − 24 x ^ 3 ) + z ^ ( − 1 + 5 x ^ − 15 x ^ 2 + 9 x ^ 3 ) + z ^ 3 ( − 5 + 37 x ^ − 54 x ^ 2 + 18 x ^ 3 )} , = − N 2 − 1 1 q T z ^ 2 2 ( 3 z ^ ( 1 − 2 x ^ ) x ^ + x ^ ( 1 + x ^ ) + z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 )) , = − N 2 − 1 1 q T z ^ 2 2 ( z ^ ( 5 − 6 x ^ ) x ^ + ( − 1 + x ^ ) x ^ + z ^ 2 ( 1 − 6 x ^ + 6 x ^ 2 )) , = q T x ^ z ^ 2 1 {( − 1 + x ^ ) 2 ( − 1 + 2 x ^ ) − 3 z ^ ( − 1 + x ^ ) 2 ( − 1 + 4 x ^ ) + z ^ 3 ( 1 − 6 x ^ + 6 x ^ 2 ) + 3 z ^ 2 ( − 1 + 7 x ^ − 10 x ^ 2 + 4 x ^ 3 )} , = q T x ^ z ^ 2 − ( − 1 + x ^ ) 2 + 3 z ^ ( − 1 + x ^ ) 2 + z ^ 2 ( − 3 + 9 x ^ − 6 x ^ 2 ) + z ^ 3 ( 1 − 6 x ^ + 6 x ^ 2 ) . (66)
(3) k = 2 k=2 k = 2 :
σ ^ 1 2 = − 8 ( − 1 + x ^ ) x ^ ( − 1 + x ^ + z ^ x ^ ) q T z ^ ( − 1 + z ^ + x ^ ) , σ ^ ⊥ 2 = 8 ( − 1 + x ^ ) x ^ ( 1 + 2 z ^ 2 − x ^ + z ^ ( − 4 + 3 x ^ ) ) q T z ^ ( − 1 + z ^ + x ^ ) , σ ^ D F 1 2 = − 1 N 2 − 1 8 q T x ^ 2 Q 2 , σ ^ G F 1 2 = − 1 N 2 − 1 8 q T x ^ 2 Q 2 , σ ^ 2 2 = 4 q T x ^ ( − 2 + z ^ + 2 x ^ ) Q 2 , σ ^ 3 2 = 4 q T x ^ z ^ Q 2 . (67) \begin{aligned}
\hat{\sigma}^{2}_{1}&=-\frac{8(-1+\hat{x})\hat{x}(-1+\hat{x}+\hat{z}\hat{x})}{q_{T}\hat{z}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}^{2}_{\perp}&=\frac{8(-1+\hat{x})\hat{x}(1+2\hat{z}^{2}-\hat{x}+\hat{z}(-4+3\hat{x}))}{q_{T}\hat{z}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}^{2}_{DF1}&=-\frac{1}{N^{2}-1}\frac{8q_{T}\hat{x}^{2}}{Q^{2}},\\
\hat{\sigma}^{2}_{GF1}&=-\frac{1}{N^{2}-1}\frac{8q_{T}\hat{x}^{2}}{Q^{2}},\\
\hat{\sigma}^{2}_{2}&=\frac{4q_{T}\hat{x}(-2+\hat{z}+2\hat{x})}{Q^{2}},\\
\hat{\sigma}^{2}_{3}&=\frac{4q_{T}\hat{x}\hat{z}}{Q^{2}}. \tag*{(67)}
\end{aligned} σ ^ 1 2 σ ^ ⊥ 2 σ ^ D F 1 2 σ ^ GF 1 2 σ ^ 2 2 σ ^ 3 2 = − q T z ^ ( − 1 + z ^ + x ^ ) 8 ( − 1 + x ^ ) x ^ ( − 1 + x ^ + z ^ x ^ ) , = q T z ^ ( − 1 + z ^ + x ^ ) 8 ( − 1 + x ^ ) x ^ ( 1 + 2 z ^ 2 − x ^ + z ^ ( − 4 + 3 x ^ )) , = − N 2 − 1 1 Q 2 8 q T x ^ 2 , = − N 2 − 1 1 Q 2 8 q T x ^ 2 , = Q 2 4 q T x ^ ( − 2 + z ^ + 2 x ^ ) , = Q 2 4 q T x ^ z ^ . (67)
(4) k = 3 k=3 k = 3 :
σ ^ 1 3 = − 2 x ^ ( − 4 ( − 1 + x ^ ) 2 + 2 z ^ ( 3 − 4 x ^ + x ^ 2 ) + z ^ 2 ( − 1 − 2 x ^ + 4 x ^ 2 ) ) Q ( − 1 + z ^ ) z ^ ( − 1 + z ^ + x ^ ) , σ ^ ⊥ 3 = 2 x ^ ( − 1 − x ^ + 2 x ^ 2 + z ^ 3 ( − 4 + 8 x ^ ) + z ^ ( − 3 + 16 x ^ − 10 x ^ 2 ) + z ^ 2 ( 9 − 26 x ^ + 12 x ^ 2 ) ) Q ( − 1 + z ^ ) z ^ ( − 1 + z ^ + x ^ ) , σ ^ D F 1 3 = − 1 N 2 − 1 2 x ^ ( − 1 − 2 x ^ + z ^ ( − 2 + 4 x ^ ) ) Q z ^ , σ ^ G F 1 3 = − 1 N 2 − 1 2 ( − 1 + 2 z ^ ) x ^ ( − 1 + 2 x ^ ) Q z ^ , σ ^ 2 3 = − 3 + 7 x ^ − 4 x ^ 2 + z ^ 2 ( − 2 + 4 x ^ ) + 2 z ^ ( 3 − 7 x ^ + 4 x ^ 2 ) Q z ^ , σ ^ 3 3 = − 1 − 2 z ^ ( − 1 + x ^ ) + x ^ + z ^ 2 ( − 2 + 4 x ^ ) Q z ^ . (68) \begin{aligned}
\hat{\sigma}_1^3 &= -\frac{2\hat{x}\left(-4(-1+\hat{x})^2+2\hat{z}(3-4\hat{x}+\hat{x}^2)+\hat{z}^2(-1-2\hat{x}+4\hat{x}^2)\right)}{Q(-1+\hat{z})\hat{z}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{\perp}^3 &= \frac{2\hat{x}(-1-\hat{x}+2\hat{x}^2+\hat{z}^3(-4+8\hat{x})+\hat{z}(-3+16\hat{x}-10\hat{x}^2)+\hat{z}^2(9-26\hat{x}+12\hat{x}^2))}{Q(-1+\hat{z})\hat{z}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{DF1}^3 &= -\frac{1}{N^2-1}\frac{2\hat{x}(-1-2\hat{x}+\hat{z}(-2+4\hat{x}))}{Q\hat{z}},\\
\hat{\sigma}_{GF1}^3 &= -\frac{1}{N^2-1}\frac{2(-1+2\hat{z})\hat{x}(-1+2\hat{x})}{Q\hat{z}},\\
\hat{\sigma}_2^3 &= \frac{-3+7\hat{x}-4\hat{x}^2+\hat{z}^2(-2+4\hat{x})+2\hat{z}(3-7\hat{x}+4\hat{x}^2)}{Q\hat{z}},\\
\hat{\sigma}_3^3 &= \frac{-1-2\hat{z}(-1+\hat{x})+\hat{x}+\hat{z}^2(-2+4\hat{x})}{Q\hat{z}}. \tag*{(68)}
\end{aligned} σ ^ 1 3 σ ^ ⊥ 3 σ ^ D F 1 3 σ ^ GF 1 3 σ ^ 2 3 σ ^ 3 3 = − Q ( − 1 + z ^ ) z ^ ( − 1 + z ^ + x ^ ) 2 x ^ ( − 4 ( − 1 + x ^ ) 2 + 2 z ^ ( 3 − 4 x ^ + x ^ 2 ) + z ^ 2 ( − 1 − 2 x ^ + 4 x ^ 2 ) ) , = Q ( − 1 + z ^ ) z ^ ( − 1 + z ^ + x ^ ) 2 x ^ ( − 1 − x ^ + 2 x ^ 2 + z ^ 3 ( − 4 + 8 x ^ ) + z ^ ( − 3 + 16 x ^ − 10 x ^ 2 ) + z ^ 2 ( 9 − 26 x ^ + 12 x ^ 2 )) , = − N 2 − 1 1 Q z ^ 2 x ^ ( − 1 − 2 x ^ + z ^ ( − 2 + 4 x ^ )) , = − N 2 − 1 1 Q z ^ 2 ( − 1 + 2 z ^ ) x ^ ( − 1 + 2 x ^ ) , = Q z ^ − 3 + 7 x ^ − 4 x ^ 2 + z ^ 2 ( − 2 + 4 x ^ ) + 2 z ^ ( 3 − 7 x ^ + 4 x ^ 2 ) , = Q z ^ − 1 − 2 z ^ ( − 1 + x ^ ) + x ^ + z ^ 2 ( − 2 + 4 x ^ ) . (68)
(5) k = 4 k=4 k = 4 :
σ ^ 1 4 = − 2 ( − 1 + x ^ ) ( 1 − 3 x ^ + 2 ( 1 + z ^ ) x ^ 2 ) q T z ^ ( − 1 + z ^ + x ^ ) , σ ^ ⊥ 4 = 2 ( − 1 + x ^ ) ( 1 + ( 1 − 8 z ^ + 4 z ^ 2 ) x + ( − 2 + 6 z ^ ) x ^ 2 ) q T z ^ ( − 1 + z ^ + x ^ ) , σ ^ D F 1 4 = − 1 N 2 − 1 2 ( − 1 + x ^ ) ( − 1 + 2 ( − 1 + z ^ ) x ^ ) q T z ^ , σ ^ G F 1 4 = − 1 N 2 − 1 2 ( − 1 + x ^ ) ( 1 + 2 ( − 1 + z ^ ) x ^ ) q T z ^ , σ ^ 2 4 = q T ( 1 + 2 ( − 2 + z ^ ) x ^ + 4 x ^ 2 ) Q 2 , σ ^ 3 4 = q T ( − 1 + 2 x ^ z ^ ) Q 2 . (69) \begin{aligned}
\hat{\sigma}_1^4 &= -\frac{2(-1+\hat{x})(1-3\hat{x}+2(1+\hat{z})\hat{x}^2)}{q_T\hat{z}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{\perp}^4 &= \frac{2(-1+\hat{x})(1+(1-8\hat{z}+4\hat{z}^2)x+(-2+6\hat{z})\hat{x}^2)}{q_T\hat{z}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{DF1}^4 &= -\frac{1}{N^2-1}\frac{2(-1+\hat{x})(-1+2(-1+\hat{z})\hat{x})}{q_T\hat{z}},\\
\hat{\sigma}_{GF1}^4 &= -\frac{1}{N^2-1}\frac{2(-1+\hat{x})(1+2(-1+\hat{z})\hat{x})}{q_T\hat{z}},\\
\hat{\sigma}_2^4 &= \frac{q_T(1+2(-2+\hat{z})\hat{x}+4\hat{x}^2)}{Q^2},\\
\hat{\sigma}_3^4 &= \frac{q_T(-1+2\hat{x}\hat{z})}{Q^2}. \tag*{(69)}
\end{aligned} σ ^ 1 4 σ ^ ⊥ 4 σ ^ D F 1 4 σ ^ GF 1 4 σ ^ 2 4 σ ^ 3 4 = − q T z ^ ( − 1 + z ^ + x ^ ) 2 ( − 1 + x ^ ) ( 1 − 3 x ^ + 2 ( 1 + z ^ ) x ^ 2 ) , = q T z ^ ( − 1 + z ^ + x ^ ) 2 ( − 1 + x ^ ) ( 1 + ( 1 − 8 z ^ + 4 z ^ 2 ) x + ( − 2 + 6 z ^ ) x ^ 2 ) , = − N 2 − 1 1 q T z ^ 2 ( − 1 + x ^ ) ( − 1 + 2 ( − 1 + z ^ ) x ^ ) , = − N 2 − 1 1 q T z ^ 2 ( − 1 + x ^ ) ( 1 + 2 ( − 1 + z ^ ) x ^ ) , = Q 2 q T ( 1 + 2 ( − 2 + z ^ ) x ^ + 4 x ^ 2 ) , = Q 2 q T ( − 1 + 2 x ^ z ^ ) . (69)
(6) k = 8 k=8 k = 8 :
σ ^ 1 8 = − 2 x ^ ( − 2 ( − 1 + x ^ ) 2 + 4 z ^ ( − 1 + x ^ ) 2 + z ^ 2 ( − 1 + 2 x ^ ) ) Q ( − 1 + z ^ ) z ^ ( − 1 + z ^ + x ^ ) \hat{\sigma}_1^8=-\frac{2\hat{x}\left(-2(-1+\hat{x})^2+4\hat{z}(-1+\hat{x})^2+\hat{z}^2(-1+2\hat{x})\right)}{Q(-1+\hat{z})\hat{z}(-1+\hat{z}+\hat{x})} σ ^ 1 8 = − Q ( − 1 + z ^ ) z ^ ( − 1 + z ^ + x ^ ) 2 x ^ ( − 2 ( − 1 + x ^ ) 2 + 4 z ^ ( − 1 + x ^ ) 2 + z ^ 2 ( − 1 + 2 x ^ ) )
σ ^ ⊥ 8 = − 2 x ^ ( 1 − z ^ − x ^ + z ^ 2 ( − 1 + 2 x ^ ) ) Q ( − 1 + z ^ ) z ^ ( − 1 + z ^ + x ^ ) , \hat{\sigma}_{\perp}^8=-\frac{2\hat{x}\left(1-\hat{z}-\hat{x}+\hat{z}^2(-1+2\hat{x})\right)}{Q(-1+\hat{z})\hat{z}(-1+\hat{z}+\hat{x})}, σ ^ ⊥ 8 = − Q ( − 1 + z ^ ) z ^ ( − 1 + z ^ + x ^ ) 2 x ^ ( 1 − z ^ − x ^ + z ^ 2 ( − 1 + 2 x ^ ) ) ,
σ ^ D F 1 8 = 1 N 2 − 1 x ^ 2 Q z ^ , \hat{\sigma}_{DF1}^8=\frac{1}{N^2-1}\frac{\hat{x}}{2Q\hat{z}}, σ ^ D F 1 8 = N 2 − 1 1 2 Q z ^ x ^ ,
σ ^ G F 1 8 = − 1 N 2 − 1 x ^ 2 Q z ^ , σ ^ 2 8 = − ( − 1 + 2 z ^ ) ( − 1 + x ^ ) 2 Q z ^ , σ ^ 3 8 = ( − 1 + 2 z ^ ) ( − 1 + x ^ ) Q z ^ . (70) \begin{aligned}
\hat{\sigma}_{GF1}^{8} &= -\frac{1}{N^{2}-1}\frac{\hat{x}}{2Q\hat{z}},\\
\hat{\sigma}_{2}^{8} &= -\frac{(-1+2\hat{z})(-1+\hat{x})^{2}}{Q\hat{z}},\\
\hat{\sigma}_{3}^{8} &= \frac{(-1+2\hat{z})(-1+\hat{x})}{Q\hat{z}}. \tag*{(70)}
\end{aligned} σ ^ GF 1 8 σ ^ 2 8 σ ^ 3 8 = − N 2 − 1 1 2 Q z ^ x ^ , = − Q z ^ ( − 1 + 2 z ^ ) ( − 1 + x ^ ) 2 , = Q z ^ ( − 1 + 2 z ^ ) ( − 1 + x ^ ) . (70)
(7) k = 9 k=9 k = 9 :
σ ^ 1 9 = − 2 ( − 1 + x ^ ) ( 1 + ( − 5 + 2 z ^ ) x ^ + 4 x ^ 2 ) q T z ^ ( − 1 + z ^ + x ^ ) , σ ^ ⊥ 9 = − 2 ( − 1 + x ^ ) ( − 1 + x ^ + 2 z ^ x ^ ) q T z ^ ( − 1 + z ^ + x ^ ) , σ ^ D F 1 9 = 1 N 2 − 1 q T x ^ 2 Q 2 ( − 1 + z ^ ) , σ ^ G F 1 9 = − 1 N 2 − 1 q T x ^ 2 Q 2 ( − 1 + z ^ ) , σ ^ 2 9 = − q T ( − 1 + 2 x ^ ) Q 2 , σ ^ 3 9 = q T ( − 1 + 2 x ^ ) Q 2 . (71) \begin{aligned}
\hat{\sigma}_{1}^{9} &= -\frac{2(-1+\hat{x})(1+(-5+2\hat{z})\hat{x}+4\hat{x}^{2})}{q_{T}\hat{z}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{\perp}^{9} &= -\frac{2(-1+\hat{x})(-1+\hat{x}+2\hat{z}\hat{x})}{q_{T}\hat{z}(-1+\hat{z}+\hat{x})},\\
\hat{\sigma}_{DF1}^{9} &= \frac{1}{N^{2}-1}\frac{q_{T}\hat{x}}{2Q^{2}(-1+\hat{z})},\\
\hat{\sigma}_{GF1}^{9} &= -\frac{1}{N^{2}-1}\frac{q_{T}\hat{x}}{2Q^{2}(-1+\hat{z})},\\
\hat{\sigma}_{2}^{9} &= -\frac{q_{T}(-1+2\hat{x})}{Q^{2}},\\
\hat{\sigma}_{3}^{9} &= \frac{q_{T}(-1+2\hat{x})}{Q^{2}}. \tag*{(71)}
\end{aligned} σ ^ 1 9 σ ^ ⊥ 9 σ ^ D F 1 9 σ ^ GF 1 9 σ ^ 2 9 σ ^ 3 9 = − q T z ^ ( − 1 + z ^ + x ^ ) 2 ( − 1 + x ^ ) ( 1 + ( − 5 + 2 z ^ ) x ^ + 4 x ^ 2 ) , = − q T z ^ ( − 1 + z ^ + x ^ ) 2 ( − 1 + x ^ ) ( − 1 + x ^ + 2 z ^ x ^ ) , = N 2 − 1 1 2 Q 2 ( − 1 + z ^ ) q T x ^ , = − N 2 − 1 1 2 Q 2 ( − 1 + z ^ ) q T x ^ , = − Q 2 q T ( − 1 + 2 x ^ ) , = Q 2 q T ( − 1 + 2 x ^ ) . (71)
This completes the specification of all partonic hard cross sections in (47) and (48) .
Summary
In this paper, we have studied the twist-3 cross section for the transversely polarized hyperon production in SIDIS e p → e Λ ↑ X ep \to e\Lambda^{\uparrow}X e p → e Λ ↑ X in the framework of collinear factorization. The cross section consists of five structure functions with different dependences on the azimuthal angles. We have presented the LO cross section which occurs from the twist-3 DF in the initial proton combined with the transversity FF for Λ ↑ \Lambda^{\uparrow} Λ ↑ and the twist-3 quark FFs for Λ ↑ \Lambda^{\uparrow} Λ ↑ combined with the unpolarized DF in the proton for all five structure functions. The derived cross section is relevant for the large-P T P_{T} P T Λ ↑ \Lambda^{\uparrow} Λ ↑ production in the future EIC experiment. For completeness the contribution from the twist-3 purely gluon FFs as well as another q ˉ q g \bar{q}qg q ˉ q g -FF of the type ∼ ⟨ 0 ∣ g F α w ∣ h X ⟩ ⟨ h X ∣ ψ ˉ ψ ∣ 0 ⟩ \sim\langle0|gF^{\alpha w}|hX\rangle\langle hX|\bar{\psi}\psi|0\rangle ∼ ⟨ 0∣ g F α w ∣ h X ⟩ ⟨ h X ∣ ψ ˉ ψ ∣0 ⟩ needs to be included (as discussed at the end of § 2 ), which will be reported in a separate paper [31 ] .
Acknowledgments
This work has been supported by the Grant-in-Aid for Scientific Research from the Japanese Society of Promotion of Science under Contract Nos. 19K03843 (Y.K.) and 18J11148 (K.Y.), National Natural Science Foundation in China under grant No. 11950410495, Guangdong Natural Science Foundation under No. 2020A1515010794 and research startup funding at South China Normal University.