-
Additional testing and development of the standard model (SM) and exploration of evidence for the existence of new physics (NP) beyond the SM are the most important research projects in the field of particle physics. The
B/D to light meson exclusive decay processes provide a good platform for these major research projects. The physical quantities of interest in these processes, such as the decay amplitude, decay width, and decay branching ratio, can usually be expressed in the form of a convolution of the distribution amplitudes (DAs) of the final state light mesons. Therefore, light meson DAs, particularly leading-twist DAs, are the key parameters for predicting these exclusive decay processes, and they also constitute the main error sources. Their accuracy is directly related to the theoretical prediction accuracy of such processes. The discussion on meson DAs began in 1980, when Lepage and Brodsky used collinear factorization to study the large momentum transition process [1]. In the following 40 years, the pionic leading-twist DAϕ2;π(x,μ) has attracted considerable attention. However, a completely consistent understanding of the pionic leading-twist DA behavior is still lacking.Generally, the pionic leading-twist DA
ϕ2;π(x,μ) describes the momentum fraction distributions of partons in a pion in a particular Fock state, which is a universal nonperturbative object, and cannot be obtained directly through experiments. In principle, nonperturbative QCD methods should be used to study the pionic leading-twist DA. However, owing to the difficulty associated with nonperturbative QCD,ϕ2;π(x,μ) is usually studied by combining nonperturbative QCD such as QCD sum rules and the lattice QCD theory (LQCD) with the phenomenological model [2, 3] or by combining the phenomenological model with relevant experimental data [4–7]. Evidently, the information regarding the pionic leading-twist DA obtained from the nonperturbative methods and related experiments must be expressed by the phenomenological model. The relevant literature highlights several phenomenological models commonly used to describe the pionic leading-twist DA behavior; these models include the truncation form of the Gegenbauer polynomial series based on{C3/2n} -basis (TF model), the light-cone harmonic oscillator model (LCHO model) [2–10] based on the Brodsky-Huang-Lepage (BHL) description [11], the form obtained from Dyson-Schwinger equations (DSE model) [12], the model derived from light-front holographic AdS/QCD (AdS/QCD model) [13, 14] , and the power-law parametrization form (PLP model) [15]. In addition, the pionic leading-twist DA has been studied using the light-front quark model [16, 17], the light-front constituent quark model [18], the nonlocal chiral-quark model (NLChQM) from the instanten vacuum [19], and the Nambu-Jona-Lasinio model [20–22] or by considering the infinite-momentum limit for the quasidistribution amplitude within NLChQM and LQCD based on the large-momentum effective theory [23].As a first-principles method, however, the computation of LQCD is usually limited to the first few moments of
ϕ2;π(x,μ) [15, 24–33]. Owing to the systematic error arising from the truncation of the high-dimensional condensation terms and the approximation of the spectral density based on the quark-hadron duality, research on the pionic leading-twist DA using QCD sum rules is also limited to the lowest moments [34–44]. However, one can also abstract information on the pionic twist-2 DA via exclusive processes involving pions, such as the pion-photon transition form factor [45–52], pion electromagnetic form factor [53–56], andB/D→πlν semileptonic decays [57–59]. The information regarding the pionic leading-twist DA obtained from nonperturbative methods and relevant experimental data is usually incomplete or indirect. Therefore, the choice of the phenomenological model is particularly important to accurately describe the behavior of the pionic leading-twist DA.Although the calculation of moments using nonperturbative methods is usually limited to the lowest orders, several studies that attempt to calculate adequate moments to obtain more complete information on the pionic leading-twist DA have been reported in the literature. For example, in the aforementioned DSE model, the first 50 x-moments are calculated by analyzing the corresponding gap and Bethe-Salpeter equations [12]. Very recently, to resolve the long-term limitation on the applications of the QCD sum rules, Li proposed a new dispersion relation for the moments of the pionic leading-twist DA [60] and obtained the first 18 Gegenbauer moments by solving the dispersion relation as an inverse problem in the inverse matrix method [61]. In Ref. [3], we studied the pionic leading-twist DA using QCD sum rules within the framework of the background field theory [62]. We clarified the fact that the sum rule of the zeroth ξ-moment of the pionic leading-twist DA cannot be normalized in the entire Borel parameter region, and we proposed a new sum rule formula for the nth ξ-moments. This new formula can reduce the systematic error of the sum rules of the ξ-moments, and it enables us to calculate higher-order moments. Following this, we calculated the values of the ξ-moments up to the 10th order and fitted those values with the LCHO model via the least squares method to determine the behavior of DA
ϕ2;π(x,μ) . This method avoids the unreliability problem encountered in calculating the higher-order Gegenbauer moments of the DA by using the QCD sum rules. More ξ-moments create stronger constraints on the DA behavior. The research idea proposed in Ref. [3] improves the prediction ability of the QCD sum rules in the study of meson DAs and has been used to study the kaon leading-twist DA [63] and axial-vectora1(1260) -meson longitudinal leading-twist DA [64]. Inspired by the results reported in Ref. [3], we attempt to calculate the values of more ξ-moments with orders higher than 10 and analyze several commonly used and relatively simple models from the literature, such as the TF model, LCHO model, DSE model, AdS/QCD model, and PLP model, using the least squares method.The remainder of this paper is organized as follows. In Sec. II, we provide a brief overview of several pionic leading-twist DA models to be analyzed. In Sec. III, we present the sum rules of the nth ξ-moment obtained in Ref. [3] for subsequent discussions. A numerical analysis is described in Sec. IV, and Sec. V provides a summary.
-
TF model By solving the renormalization group equation, the pionic leading-twist DA at a scale μ can be expanded into a Gegenbauer polynomial series [65, 66]. Its truncation form truncated from the Nth term may be the most common phenomenological model, and it is described as
ϕTF2;π(x,μ)=6x(1−x)[1+∑Nn=2a2;πn(μ)C3/2n(2x−1)],
(1) where
C3/2n(2x−1) is the nth order Gegenbauer polynomial, and the coefficienta2;πn(μ) is the corresponding Gegenbauer moment. Owing to isospin symmetry, the odd Gegenbauer moments vanish, and only the even Gegenbauer moments are retained. The Gegenbauer momentsa2;πn(μ) can be calculated directly using nonperturbative LQCD [15, 24–33] or QCD sum rules [34–44]. However, owing to the limitations of LQCD and QCD sum rules and the fact that the stability of the Gegenbauer moments calculated based on these two nonperturbative methods decreases sharply with increasing order n [3], the ability of the TF model in describing the dependence on the parton momentum fraction x is difficult to improve. Therefore, the following two forms for the TF model have been usually adopted in the literature①:ϕTF,I2;π(x,μ)=6x(1−x)[1+a2;π2(μ)C3/22(2x−1)],
(2) and
ϕTF,II2;π(x,μ)=6x(1−x)[1+a2;π2(μ)C3/22(2x−1)+a2;π4(μ)C3/24(2x−1)].
(3) LCHO model By using the approximate bound-state solution of a hadron in terms of the quark model as the starting point, BHL suggest that the hadronic wave function (WF) can be obtained by connecting the equal-time WF in the rest frame with the WF in the infinite-momentum frame. This is the so-called BHL description [11]. Based on the BHL description, the LCHO model for the pionic leading-twist DA is established and improved [2–10], and this can be expressed as [3]
ϕLCHO2;π(x,μ0)=√3A2;πˆmqβ2;π2π3/2fπ√x(1−x)φ2;π(x)×{Erf[√ˆm2q+μ208β22;πx(1−x)]−Erf[√ˆm2q8β22;πx(1−x)]},
(4) with the longitudinal distribution function
φ2;π(x)=[x(1−x)]α2;π[1+B2;π2C3/22(2x−1)],
(5) where
ˆmq is the constituent u or d quark mass, andfπ is the pion decay constant. Several mass schemes forˆmq have been proposed in the literature. In Ref. [3], however, we find that by fitting the values of the ξ-moments with the LCHO model (4), the goodness of fit increases asˆmq decreases. Thus, we considerˆmq=150 MeV in this paper. In addition,A2;π is the normalization constant,β2;π is a harmonious parameter, andErf(x)=2∫x0e−t2dx/√π is the error function. The LCHO model is based on four model parameters:A2;π ,β2;π ,α2;π , andB2;π2 . Usually, two constraints are imposed: the DA normalization condition resulting from the processπ→μν , and the sum rule derived from theπ0→γγ decay amplitude [3]. Only two model parameters are independent, which can be constrained by the ξ-moments [3] or experimental data [4–7].DSE model In Ref. [12], the authors suggest a new form of the pionic leading-twist DA by combining the truncation of the series of the Gegenbauer polynomials with order α,
Cαn(2x−1) , and[x(1−x)]α− withα−=α−1/2 . Compared with the TF model based on the{C3/2n} -basis, its advantage is that one can accelerate the convergence of the procedure by optimizing α, reducing the non-zero coefficients, and reducing the introduced spurious oscillations. The parameters of the DSE model are determined by using the first50 x-moments,⟨xn⟩=∫10dxxnϕ2;π(x,μ) , obtained by analyzing the corresponding gap and Bethe-Salpeter equations [12]. However, the results depend on the kernels. Evident differences are noted between the behaviors with the rainbow-ladder and dynamical-chiral-symmetry-breaking-improved kernels [12]. Based on Ref. [12], the DSE model for the pionic leading-twist DA is described asϕDSE2;π(x)=N[x(1−x)]αDSE−[1+aDSE2CαDSE2(2x−1)],
(6) where
αDSE−=αDSE−1/2 , andN is the normalization constant. With the normalization condition of the pionic leading-twist DA,N=4αDSEΓ(αDSE+1)/[√πΓ(αDSE+1/2)] .AdS/QCD model The light-front (LF) holographic AdS/QCD was developed more than 10 years ago [67–71], and it is a semiclassical first approximation to strongly coupled QCD. In this approach, a holographic duality exists between the hadron dynamics in physical four-dimensional spacetime at a fixed LF time
τ=x+=x0+x3 and the dynamics of the gravitational theory in five-dimensional anti-de Sitter (AdS) space. Based on the LF holographic AdS/QCD, Ref. [13] suggests two forms for the pionic leading-twist DA, which are described asϕAdS,I2;π(x,μ)=√3πfπ∫|k⊥|<μd2k⊥(2π)2N1(xˉx)1/2׈mqψAdS2;π(x,k⊥),
(7) ϕAdS,II2;π(x,μ)=√3πfπ∫|k⊥|<μd2k⊥(2π)2N2(xˉx)1/2×(ˆmq+xˉx˜mπ)ψAdS2;π(x,k⊥),
(8) with the radial WF
ψAdS2;π(x,k⊥)=4π√λ1√xˉxe−k2⊥2λxˉxe−ˆm2q2λxˉx.
(9) Here,
√λ is the mass scale parameter;k⊥ is the transverse momentum;N1 andN2 are the normalization constants and can be obtained with the normalization condition forϕAdS,I2;π(x,μ) andϕAdS,II2;π(x,μ) , respectively; and˜mπ=ˆm2q+k2⊥x(1−x) is the invariant mass of theqˉq pair in the pseudoscalar meson [14]. The difference betweenϕAdS,I2;π(x,μ) andϕAdS,II2;π(x,μ) can be attributed to the fact that the latter is constructed by further considering the impact of the Dirac structure such asp̸γ5 on the spin WF. By substituting Eq. (9) into Eqs. (7) and (8) and after integrating over the transverse momentumk⊥ , the explicit forms ofϕAdS,I2;π(x,μ) andϕAdS,II2;π(x,μ) can be obtained asϕAdS,I2;π(x,μ)=2√3N1√λπfπˆmqexp(−ˆm2q2λxˉx)×[1−exp(−μ22λxˉx)],
(10) ϕAdS,II2;π(x,μ)=2√3N2√λπfπ(xˉx)1/2{exp(−ˆm2q+μ22λxˉx)×[ˆmqexp(μ22λxˉx)(1+(xˉx)−1/2)−ˆmq(xˉx)−1/2−√ˆm2q+μ2]+√π2√λ(xˉx)1/2[Erf(√ˆm2q+μ22λxˉx)−Erf(√ˆm2q2λxˉx)]}.
(11) As pointed out in Refs. [13, 72], the light-quark mass
ˆmq in the AdS/QCD model is the effective quark mass obtained from the reduction of higher Fock states as functionals of the valence states. Hence, we do not fix it as in the LCHO model. Here,ˆmq and√λ should be the undetermined parameters of the AdS/QCD model, and these can be extracted from observable measurements [13, 72, 73].PLP model By analyzing 35 different Coordinated Lattice Simulation ensembles with
Nf=2+1 flavors of dynamical Wilson-Clover fermions, Ref. [15] presents the lattice determination for the first Gegenbauer moments of the pionic leading-twist DA and suggests a model for the pion leading-twist DA based on a simple power-law parametrization:ϕPLP2;π(x)=Γ[2α+2]Γ[α+1]2xα(1−x)α.
(12) Evidently, the PLP model
ϕPLP2;π(x) satisfies the normalization condition owing to the coefficientΓ[2α+2]/Γ[α+1]2 . Recently, the PLP model was also used in Ref. [60] to fit the first 18 Gegenbauer moments by solving the dispersion relation as an inverse problem in the inverse matrix method [61] within the framework of the QCD sum rule method. -
Considering the fact that the sum rule of the zeroth order ξ-moment of the pionic leading-twist DA cannot be normalized in the entire Borel parameter region, we suggested the following form for the nth order ξ-moment in Ref. [3]:
⟨ξn⟩2;π=⟨ξn⟩2;π⟨ξ0⟩2;π√⟨ξ0⟩22;π.
(13) Here, the numerator is obtained based on the following sum rules:
⟨ξn⟩2;π⟨ξ0⟩2;πf2πM2em2π/M2=34π21(n+1)(n+3)(1−e−sπ/M2)+(md+mu)⟨ˉqq⟩(M2)2+⟨αsG2⟩(M2)21+nθ(n−2)12π(n+1)−(md+mu)⟨gsˉqσTGq⟩(M2)38n+118+⟨gsˉqq⟩2(M2)34(2n+1)81−⟨g3sfG3⟩(M2)3nθ(n−2)48π2+⟨g2sˉqq⟩2(M2)32+κ2486π2×{−2(51n+25)(−lnM2μ2)+3(17n+35)+θ(n−2)[2n(−lnM2μ2)+49n2+100n+56n−25(2n+1)[ψ(n+12)−ψ(n2)+ln4]]}. (14) The denominator in Eq. (13) can be obtained by substituting
n=0 in Eq. (14), i.e.,⟨ξ0⟩22;πf2πM2em2π/M2=14π2(1−e−sπ/M2)+(md+mu)⟨ˉqq⟩(M2)2+⟨αsG2⟩(M2)2112π−118(md+mu)⟨gsˉqσTGq⟩(M2)3+481⟨gsˉqq⟩2(M2)3+⟨g2sˉqq⟩2(M2)32+κ2486π2×[−50(−lnM2μ2)+105].
(15) In Eqs. (14) and (15),
mπ is the mass of a pion, M is the Borel parameter,mu(d) is the currentu(d) quark mass,sπ≃1.05 GeV2 [3] is the continuum threshold, andψ(x) is the digamma function. In addition,⟨ˉqq⟩ ,⟨αsG2⟩ ,⟨gsˉqσTGq⟩ , and⟨g3sfG3⟩ are the double-quark condensate, double-gluon condensate, quark-gluon mixed condensate, and triple-gluon condensate, respectively;⟨gsˉqq⟩2 and⟨g2sˉqq⟩2 are two four-quark condensates; and κ is the ratio of the double s quark condensate and doubleu/d quark condensate.As discussed in Ref. [3], the sum rules in Eq. (13) can provide more accurate values for the ξ-moments, owing to the elimination of some systematic errors caused by the continuum state, the absence of high-dimensional condensates, and various input parameters. In particular, we obtain appropriate Borel windows for the first five nonzero ξ-moments of the pionic leading-twist DA by constraining the dimension-six contribution for all
⟨ξn⟩2;π to be no more than5 % and the continuum contribution of⟨ξn⟩2;π to be less than(30,35,40,40,40) % forn=(2,4,6,8,10) [3]. This, in turn, implies that the systematic error caused by the missing higher-dimension condensates and the continuum and excited states under the quark-hadron duality approximation does not increase with the increasing order n. Following this, the sum rules (13) indeed alleviate the limitation of the system error on the prediction ability of the QCD sum rule method for higher-order ξ-moments. Inspired by this, we calculate the ξ-moments⟨ξn⟩2;π(n=12,14,16,18,20) in the next section. Based on the criteria set to obtain the allowable Borel windows for these ξ-moments, one can find that the values of the ξ-moments⟨ξn⟩2;π(n=12,14,16,18,20) obtained in the next section are credible. -
The first five nonzero ξ-moments of the pionic leading-twist DA have been calculated using the sum rules in Eq. (13) in Ref. [3], and these can be summarized as
⟨ξ2⟩2;π=0.271±0.013,⟨ξ4⟩2;π=0.138±0.010,⟨ξ6⟩2;π=0.087±0.006,⟨ξ8⟩2;π=0.064±0.007,⟨ξ10⟩2;π=0.050±0.006
(16) on the scale of
μ=1 GeV .To calculate the values of the ξ-moments
⟨ξn⟩2;π(n=12,14,16,18,20) , we consider the same inputs as those in Ref. [3], i.e.,mπ=139.57039±0.00017MeV,fπ=130.2±1.2 MeV,mu=2.16+0.49−0.26 MeV,md=4.67+0.48−0.17 MeV,⟨ˉqq⟩=(−289.14+9.34−4.47)3 MeV3,⟨gSˉqσTGq⟩=(−1.934+0.188−0.103)×10−2 GeV5,⟨gsˉqq⟩2=(2.082+0.734−0.697)×10−3 GeV6,⟨g2sˉqq⟩2=(7.420+2.614−2.483)×10−3 GeV6,⟨αsG2⟩=0.038±0.011 GeV4,⟨g3sfG3⟩≃0.045 GeV6,κ=0.74±0.03,
(17) on the scale of
μ=2 GeV . For the scale evolutions of the above inputs, readers can refer to Ref. [3].Substituting the inputs presented in Eq. (17) into the sum rules in Eq. (13), the ξ-moments
⟨ξn⟩2;π(n= 12, 14, 16, 18, 20) and the corresponding contributions from the continuum states and dimension-six condensates versus the Borel parameterM2 can be obtained. To obtain the allowable Borel windows for⟨ξn⟩2;π(n= 12, 14, 16, 18, 20), we require the continuum state contribution and the dimension-six condensate contribution to be less than40 % and5 % for the ξ-moments, respectively. The obtained Borel windows and the corresponding values of⟨ξn⟩2;π(n= 12, 14, 16, 18, 20) are listed in Table 1. Visually, we present the pionic leading-twist DA moments⟨ξn⟩2;π withn= (12, 14, 16, 18, 20) versus the Borel parameterM2 in Fig. 1. In the calculations displayed in Table 1 and Fig. 1, all input parameters are set to their central values.n M2 ⟨ξn⟩2;π 12 [5.356,6.082] [0.0333,0.0308] 14 [6.137,6.888] [0.0271,0.0252] 16 [6.921,7.721] [0.0227,0.0211] 18 [7.706,8.578] [0.0194,0.0180] 20 [8.493,9.456] [0.0169,0.0156] Table 1. Determined Borel windows and corresponding pionic leading-twist DA ξ-moments
⟨ξn⟩2;π withn= (12, 14, 16, 18, 20); here, all input parameters are set to their central values.Figure 1. (color online) Pionic leading-twist DA moments
⟨ξn⟩2;π withn=(12,14,16,18,20) versus the Borel parameterM2 ; here, all input parameters are set to their central values. The shaded bands indicate the Borel windows forn=(12,14,16,18,20) .After considering all uncertainty sources, the values of the pionic leading-twist DA ξ-moments
⟨ξn⟩2;π|μ withn=(12,14,16,18,20) can be obtained as⟨ξ12⟩2;π=0.0408+0.0050−0.0049,⟨ξ14⟩2;π=0.0346+0.0045−0.0045,⟨ξ16⟩2;π=0.0301+0.0042−0.0041,⟨ξ18⟩2;π=0.0266+0.0039−0.0038,⟨ξ20⟩2;π=0.0239+0.0037−0.0036
(18) on the scale of
μ=1 GeV .Next, we can fit the values of the ξ-moments listed in Eqs. (16) and (18) with the models introduced in Sec. II by adopting the least squares method. For the specific fitting procedure, readers can refer to Refs. [3, 63]. Based on the brief review on pionic leading-twist DA models such as the TF model, LCHO model, DSE model, AdS/QCD model, and PLP model presented in Sec. II, we first specify the fitting parameters
θ as follows:θ=(a2;π2) ,(a2;π2,a2;π4) ,(α2;π,B2;π2) ,(αDSE,aDSE2) ,(√λ,ˆmq) ,(√λ,ˆmq) , and(α) forϕTF,I2;π ,ϕTF,II2;π ,ϕLCHO2;π ,ϕDSE2;π ,ϕAdS,I2;π ,ϕAdS,II2;π , andϕPLP2;π , respectively.To analyze the constraint strength of the moments on the model, we use the pionic leading-twist DA models introduced in Sec. II to fit the values of the first nonzero three ξ-moments, five ξ-moments, seven ξ-moments, nine ξ-moments, and ten ξ-moments. That is, we set the values of the ξ-moments listed in Eqs. (16) and (18) to the form of the following five groups as the fitting samples:
NS:3,{⟨ξ2⟩2;π,⟨ξ4⟩2;π,⟨ξ6⟩2;π};NS:5,{⟨ξ2⟩2;π,⟨ξ4⟩2;π,⟨ξ6⟩2;π,⟨ξ8⟩2;π,⟨ξ10⟩2;π};NS:7,{⟨ξ2⟩2;π,⟨ξ4⟩2;π,⟨ξ6⟩2;π,⟨ξ8⟩2;π,⟨ξ10⟩2;π,⟨ξ12⟩2;π,⟨ξ14⟩2;π};NS:9,{⟨ξ2⟩2;π,⟨ξ4⟩2;π,⟨ξ6⟩2;π,⟨ξ8⟩2;π,⟨ξ10⟩2;π,⟨ξ12⟩2;π,⟨ξ14⟩2;π,⟨ξ16⟩2;π,⟨ξ18⟩2;π};NS:10,{⟨ξ2⟩2;π,⟨ξ4⟩2;π,⟨ξ6⟩2;π,⟨ξ8⟩2;π,⟨ξ10⟩2;π,⟨ξ12⟩2;π,⟨ξ14⟩2;π,⟨ξ16⟩2;π,⟨ξ18⟩2;π,⟨ξ20⟩2;π}, (19) where
NS is an abbreviation denoting the number of samples.The fitting results are summarized in Table 2, and the corresponding model curves are displayed in Fig. 2. In the fitting process, the errors of the ξ-moments
− arising from error sources such as the hadron parameters, light quark masses, and vacuum condensates− are assumed to be variances. One can discuss the uncertainties of the fitted model parameters by setting an appropriate threshold forPχ2 , as in Ref. [63]. However, to better analyze the constraint imposed by different numbers of ξ-moments on the different pion DA models, only the optimal fitted results are provided here. From Table 2 and Fig. 2, one can observe the following:Models NS 3 5 7 9 10 ϕTF,I2;π a2;π2 0.225 0.247 0.268 0.285 0.293 χ2min 0.565723 4.14428 11.4704 21.9983 28.1611 Pχ2min 0.753624 0.386832 0.0748825 0.00491901 0.000896493 ϕTF,II2;π a2;π2 0.205 0.197 0.187 0.177 0.172 a2;π4 0.060 0.098 0.131 0.158 0.170 χ2min 0.0162808 0.480741 1.6772 3.78499 5.17049 Pχ2min 0.898467 0.923102 0.891759 0.804182 0.739208 ϕLCHO2;π A2;π 5.61391 6.6104 8.78724 8.78198 8.90258 α2;π −0.789 −0.684 −0.504 −0.504 −0.493 B2;π2 −0.163 −0.155 −0.136 −0.135 −0.133 β2;π 1.56878 1.60137 1.61753 1.62474 1.62855 χ2min 0.0522982 0.188917 0.478592 1.0968 1.53822 Pχ2min 0.819111 0.979358 0.992886 0.993112 0.992055 ϕDSE2;π αDSE 1.130 0.948 0.829 0.738 0.703 aDSE2 0.129 0.048 −0.035 −0.124 −0.166 χ2min 0.00559501 0.174471 0.422942 0.694404 0.818127 Pχ2min 0.940374 0.981061 0.994674 0.998378 0.999157 ϕAdS,I2;π N1 6.24998 5.58365 7.40491 4.99094 8.06215 √λ 0.319 0.344 0.305 0.379 0.301 ˆmq 0.076 0.077 0.064 0.075 0.058 χ2min 0.0395365 1.13085 3.82526 7.94241 10.3942 Pχ2min 0.84239 0.769632 0.574839 0.3377 0.238443 ϕAdS,II2;π N2 0.443594 0.405197 0.399993 0.395313 0.392851 √λ 1.342 1.212 1.248 1.286 1.300 ˆmq 0.042 0.084 0.082 0.080 0.080 χ2min 0.00512763 0.256752 1.20083 2.9544 4.08025 Pχ2min 0.942914 0.967946 0.944797 0.889188 0.849812 ϕPLP2;π α 0.380 0.379 0.370 0.359 0.354 χ2min 0.199311 0.219225 0.452207 1.04863 1.46392 Pχ2min 0.905149 0.994414 0.998372 0.997922 0.997406 Table 2. Fitting results for different DA models obtained by fitting the values of the first several nonzero pionic leading-twist DA ξ-moments listed in Eqs. (16) and (18) via the least squares method.
Figure 2. (color online) Curves of pionic leading-twist DA models such as the TF model, LCHO model, DSE model, AdS/QCD model, and PLP model corresponding to the fitting results presented in Table 2.
● At
NS=10 ,Pχ2min forϕTF,I2;π andϕAdS,I2;π is far less than1 ,Pχ2min≃0.74 forϕTF,II2;π , andPχ2min≃0.85 forϕAdS,II2;π ; however,Pχ2min>0.99 and is very close to1 forϕLCHO2;π ,ϕDSE2;π , andϕPLP2;π . This indicates that the results of the LCHO model, DSE model, and PLP model are very consistent with our values for the first ten nonzero ξ-moments listed in Eqs. (16) and (18) obtained based on the QCD sum rules.● The goodness of fits for the models
ϕTF,I2;π andϕAdS,I2;π decrease with increasing NS, whereas the goodness of fits for the modelsϕLCHO2;π ,ϕDSE2;π , andϕPLP2;π are very close to each other for different NS. This indicates that more numerous moments impose stronger constraints on the DA behavior. Thus, adequate moments can easily help us eliminate inappropriate models and select an appropriate model to better describe the correct DA behavior.● The goodness of fit for model
ϕAdS,II2;π is better than that forϕAdS,I2;π underNS=3,5,7,9,10 . This indicates thatϕAdS,II2;π based on a consideration of the influence of the Dirac structurep̸γ5 on the WF is more consistent with our sum rule results for ξ-moments. However, by fitting the first ten nonzero ξ-moments withϕAdS,II2;π , we obtain the following model parameters:√λ=1.300 GeV andˆmq=80 MeV . The obtained value of√λ is much greater than that reported in the literature. The obtained value ofˆmq is consistent with the value obtained by fitting the pion and kaon decay constants in Ref. [13]; however, it is larger than the value obtained by fitting the Regge trajectories of pseudoscalar mesons in Ref. [72].● The curves of
ϕTF,I2;π andϕTF,II2;π with different NS are significantly different from each other. This implies that modelsϕTF,I2;π andϕTF,II2;π are far from being enough to describe the behavior of the pionic leading-twist DA. Therefore, the DAs obtained by calculating the second and/or fourth Gegenbauer moments using LQCD or QCD sum rules and substituting them intoϕTF,I2;π andϕTF,II2;π in Eqs. (2) and (3) are also far from being adequate to describe the behavior of the pionic leading-twist DA. However, the goodness of fit ofϕTF,II2;π is better than that ofϕTF,I2;π with the same NS. This indicates that by increasing the number of terms in the TF, i.e., N in Eq. (1), the ability of the TF model to describe the behavior of the pionic leading-twist DA can be improved.● The curves of
ϕLCHO2;π withNS=3,5,7,9,10 almost coincide with each other; the same results are observed forϕAdS,I2;π ,ϕAdS,II2;π , andϕPLP2;π . This indicates that the LCHO model, AdS/QCD model, and PLP model present strong prediction abilities for the behavior of the pionic leading-twist DA. The three models obtained by using the first few Gegenbauer moments or ξ -moments to constrain the model parameters are adequate to describe the behavior of the pionic leading-twist DA. However, from the goodness of fits forϕAdS,I2;π andϕAdS,II2;π summarized in Table 2, one can find that these two models are not very consistent with our QCD sum rule results. -
In this paper, we calculate the pionic leading-twist DA ξ-moments
⟨ξn⟩2;π(n=12,14,16,18,20) using the sum rule formula for the nth ξ-moment, Eq. (13), suggested in our previous study [3]. In this calculation, the contributions from the continuum state and dimension-six condensate are less than40 % and5 % for the obtained five ξ-moments. Owing to the form of the sum rules in Eq. (13), the limitation of the system error, which results from the missing higher-dimension condensates and the continuum and excited states under the quark-hadron duality approximation, on the prediction ability of the QCD sum rule method for higher-order ξ-moments is alleviated. The values of the ξ-moments⟨ξn⟩2;π(n=12,14,16,18,20) are hence reliable and are listed in Eq. (18). By combining the values of⟨ξn⟩2;π(n=2,4,6,8,10) calculated in our previous study [3] and considering the values of the first ten nonzero ξ-moments as samples, we analyze several common and relatively simple models of the pionic leading-twist DA, including the TF model, LCHO model, DSE model, AdS/QCD model, and PLP model, based on the least squares method. Consequently, we find that the TF model is not sufficient to describe the behavior of the pionic leading-twist DA. By contrast, the LCHO model, AdS/QCD model, and PLP model have strong prediction abilities for the pionic leading-twist DA. The results of the LCHO model, DSE model, and PLP model are extremely consistent with our values for the first ten nonzero ξ-moments listed in Eqs. (16) and (18) obtained based on the QCD sum rules.The relevant literature contains several calculation approaches for the ξ-moments and Gegenbauer moments of the pionic leading-twist DA. However, most of the associated studies only focus on the lowest order. Our numerical analysis reveals that only the lowest order moment is far from being enough to determine the overall behavior of the DA. Therefore, we believe that if higher-order and more accurate ξ-moments can be calculated based on new methods, we may be able to judge various phenomenological models more strictly and estimate the behavior of the pionic leading-twist DA more accurately.
-
We are grateful to Professor Qin Chang for helpful discussions and valuable suggestions.
Constraints of ξ-moments computed using QCD sum rules on piondistribution amplitude models
- Received Date: 2022-09-08
- Available Online: 2023-01-15
Abstract: To date, the behavior of the pionic leading-twist distribution amplitude (DA)