betaFeedback

WorksPhysics and Astronomy

Hybrid nature of the abnormal solutions of the Bethe-Salpeter equation in the Wick-Cutkosky model

Abstract

In the Wick-Cutkosky model, where two scalar massive constituents interact by means of the exchange of a scalar massless particle, the Bethe-Salpeter equation has solutions of two types, called "normal" and "abnormal". In the non-relativistic limit, the normal solutions correspond to the usual Coulomb spectrum, whereas the abnormal ones do not have non-relativistic counterparts – they are absent in the Schrödinger equation framework. We have studied, in the formalism of the light-front dynamics, the Fock-space content of the abnormal solutions. It turns out that, in contrast to the normal ones, the abnormal states are dominated by the massless exchange particles (by 90 % or more), what provides a natural explanation of their decoupling from the two-body Schrödinger equation. Assuming that one of the massive constituents is charged, we have calculated the electromagnetic elastic form factors of the normal and abnormal states, as well as the transition form factors. The results on form factors confirm the many-body nature of the abnormal states, as found from the Fock-space analysis. The abnormal solutions have thus properties similar to those of hybrid states, made here essentially of two massive constituents and several or many massless exchange particles. They could also be interpreted as the Abelian scalar analogs of the QCD hybrid states. The question of the validity of the ladder approximation of the model is also examined.

Introduction

In their pioneering papers [1, 2], Wick and Cutkosky (W-C) have found the solutions of the Bethe-Salpeter (BS) equation [3] for two scalar particles interacting by the exchange of a massless scalar particle. In addition to the states which, in the non-relativistic limit, reproduce the spectrum of the Schrödinger equation with the Coulomb potential, there was found another set of solutions which do not have any non-relativistic counterparts. These solutions were called “abnormal”. Their discovery triggered the discussion as to whether they do indicate a mathematical inconsistency of the W-C model or of the BS equation, or whether they represent new physical systems, whose existence does not contradict any physical principles, although they are not covered by the Schrödinger equation. In the latter case, they might provide examples of relativistic systems which could exist in nature, but which would not be described by continuous extensions of non-relativistic quantum mechanics. A thorough discussion of this issue can be found in Ref. [4] (sections 6 and 8).

One would hope that a complementary lighting to the above questioning might come from experimental data. Unfortunately the conditions of the emergence of abnormal states are not easy to realize. Considering the W-C model as a simplified model of QED, abnormal states would appear as highly excited states for values of the fine structure constant α\alpha above 0.5÷10.5 \div1. Such values might be reached with the aid of heavy ions and dedicated electron-ion scattering experiments might be envisaged. However, to have a clear experimental distinction of abnormal bound states from the ionization threshold, one actually would need to increase the values of α\alpha up to 4÷54 \div5, which then further reduce the probability of an experimental success. Another possibility Keywords Bethe-Salpeter equation · Wick-Cutkosky model · Abnormal solutions · Hybrid states is an analogy of the model with hadron dynamics, where hadrons mutually interact by means of the exchange of light particles, like the pions, and where the coupling constants might lie in the range of values needed for the existence of abnormal states. However, here, the exchanged particles being massive, drastic changes occur with respect to the massless case: the interaction forces become of short-range and one realizes that abnormal states are produced only with very small mass values of the exchanged particle, much smaller than the pion mass. The framework of Quantum Chromodynamics, where quarks mutually interact by means of exchange of massless gauge particles, the gluons, with sufficiently strong forces, might provide another domain to search for possible evidences of abnormal solutions.

In the absence of any direct experimental indication about the existence or nonexistence of abnormal states, one is entitled to explore all possible theoretical paths that might provide complementary information about their properties. From this point of view, we have found that an analysis of the Fock-space content of the abnormal, as well as normal, states would be of great help. The BS amplitude allows one to extract the wave function related to the two-body sector of the Fock space [5]. Its norm, which is positive and bounded by 1, is then interpreted as the weight of that sector in the whole Fock space.

Complementary information to the above analysis comes from the knowledge of the electromagnetic form factors, assuming that one of the massive constituents of the bound state is charged. Their asymptotic behavior qualitatively probes the compositeness of the states: a rapid decrease would be the signature of a many-body structure [6, 7, 8].

Our calculations, as well as the results of ref. [9], show that, in the window of allowed coupling constants, the normal solutions are essentially dominated by the two-body sector of the Fock space. We will show in the present work that, on the contrary, the abnormal solutions have a two-body contribution that vanishes in the limit of zero binding energies and remains small (less than 10%) in all its domain of existence. They are therefore dominated by the many-body sectors, composed of the two massive constituents and of several massless exchange particles. This feature explains why the abnormal solutions disappear from the spectrum in the non-relativistic limit, the latter being formulated in the two-body sector alone, while the other sectors, containing massless particles, are by essence relativistic.

The asymptotic behaviors of the form factors also corroborate the above conclusions. The form factors of the abnormal solutions asymptotically decrease, for spacelike momenta, faster, by factors of the order of 10310^3,

than those of the normal solutions. Also, the transition form factors between normal and abnormal solutions display global suppressions, by factors of 5 ÷ 10, with respect to the normal-normal or abnormal-abnormal transition form factors, signalling a different nature of the normal and abnormal solutions.

These results suggest that the abnormal solutions might correspond to states called “hybrids” in the literature. In the present model, they are dominated by Fock space sectors containing two massive constituents and several or many massless constituents, corresponding to the exchanged-particle fields. They could be considered as the Abelian scalar analogs of the QCD hybrids, which, in the mesonic sector, are dominated by their coupling to the set of fields made of a quark, an antiquark and one or several gluon fields.

Finally, the question of the validity of the ladder approximation of the model, because of the necessity of having large values of the coupling constant to create abnormal states, still remains an open issue.

The plan of the paper is the following. § 2 is devoted to an introductory definition of the BS amplitude and of the Fock-space sectors. In § 3, the properties of the solutions of the W-C model are displayed and some solutions are found numerically. In § 4 the elastic and transition electromagnetic form factors are expressed through the BS amplitudes and are calculated numerically, with special emphasis put on their asymptotic behavior. Concluding remarks follow in § 5. Three appendices give technical details about some of the formulas used in the main text. Preliminary results of the present study were presented in Ref. [10].

Fock space sectors

The BS amplitude, satisfying the BS equation, is defined as

Φ(x1,x2;p)≡⟨0∣T[ϕ1(x1)ϕ2(x2)]∣p⟩,(1)\Phi(x_1,x_2;p) \equiv\langle0|T[\phi_1(x_1)\phi_2(x_2)]|p\rangle, \tag*{(1)}

where ϕa(x)\phi_a(x) (a=1,2a=1,2) are Heisenberg field operators, TT means time ordering, ∣p⟩|p\rangle is the state vector of the bound system and ⟨0∣\langle0| is the vacuum state vector. Since the amplitude Φ(x1,x2;p)\Phi(x_1,x_2;p) depends on two 4D variables, x1x_1 and x2x_2, it is usually called “two-body” BS amplitude, though this terminology, to some extent, is misleading. Asking the questions “what is the content of a system?” or “is it two-body or many-body?” requires that we analyze the state vector ∣p⟩|p\rangle of this system, entering in the matrix element (1), by decomposing it onto the states ∣n⟩|n\rangle with definite numbers nn of particles (the Fock sector decomposition), schematically:

∣p⟩=∑n=2∞ψn∣n⟩,(2)|p\rangle= \sum_{n=2}^{\infty} \psi_n |n\rangle, \tag*{(2)}

and studying the contributions of the two-body component ψ2\psi_2, the three-body component ψ3\psi_3, etc., in the full normalization integral. The answer to the above questions depends on which component (or sum of components) is dominant. It should be mentioned that the state vector ∣p⟩|p\rangle is usually defined on a tt-constant plane in the 4D space. There are however some advantages to choose the so called light-front plane t+z=0t+z=0 (or light-front plane of general orientation, see [5]). In this case, the corresponding Fock components ψn\psi_n are called the light-front wave functions. The two-body light-front wave function ψ2\psi_2 is related to the BS amplitude (1) by eq. (A.2) from Appendix § 6.

In the W-C model, in the ladder approximation, the Fock decomposition can contain two constituent (massive) particles and any number of exchange (massless) particles. The state ∣2⟩|2\rangle (two-body sector) contains two constituents only, the state ∣3⟩|3\rangle (three-body sector) contains two constituents and one exchange particle, the state ∣n⟩|n\rangle (nn-body sector) contains two constituents and (n−2)(n-2) exchange particles, etc. Assuming that the state vectors ∣n⟩|n\rangle (n=2,3,…n=2,3,\ldots) are normalized to unity, the state vector ∣p⟩|p\rangle is then normalized as

⟨p∣p⟩=∑n=2∞Nn=1,(3)\langle p|p\rangle= \sum_{n=2}^{\infty} N_n = 1, \tag*{(3)}

where, schematically, Nn=∫∣ψn∣2…N_n=\int|\psi_n|^2\ldots is the contribution of the nn-body Fock sector (see eq. (A.1) for the exact definition of N2N_2). In practice, knowing the BS amplitude Φ(x1,x2;p)\Phi(x_1,x_2;p) we are able to find N2N_2 only. The calculation of N2N_2 from the BS amplitude is presented in Appendix § 6. If it is dominant, this would mean that the contribution of the other sectors, containing exchange particles, is small. The limiting case, when it is enough to keep the two-body state only (the case corresponding to N2=1N_2=1), whereas the states containing exchange particles can all be omitted, is realized in non-relativistic systems. On the contrary, when the two-body contribution N2N_2 is small, the system is dominated by two constituents with an indefinite number of exchange massless particles, whose contribution ∑n=3∞Nn\sum_{n=3}^{\infty}N_n is close to 1. For the normal solutions of the W-C model (in the equal-mass case), the above analysis has been made in Ref. [9]. It was found that for small binding energies the two-body (constituent) sector dominates, as expected. When the binding energy increases (i.e., the total mass MM decreases), the two-body contribution N2N_2 decreases in parallel. However, it still dominates and as M→0M\to0, it tends, in this model, to 64%. That is, the sectors ∣n⟩|n\rangle with n≥3n\geq3 contribute in total to 36% of the total normalization of the normal state vector. In the present paper, we will carry out the same analysis for the abnormal states.

Wick-Cutkosky solutions

The BS equation [3] for the amplitude (1) containing two spinless fields, restricted to the equal-mass case m1=m2=mm_1=m_2=m, reads, in momentum space,

Φ(k,p)=i2[(p2+k)2−m2+iϵ][(p2−k)2−m2+iϵ]×∫d4k′(2π)4iK(k,k′,p)Φ(k′,p),(4)\Phi(k,p)=\frac{i^2}{\left[\left(\frac{p}{2}+k\right)^2-m^2+i\epsilon\right]\left[\left(\frac{p}{2}-k\right)^2-m^2+i\epsilon\right]}\times\int\frac{d^4k'}{(2\pi)^4}iK(k,k',p)\Phi(k',p), \tag*{(4)}

where pp and kk are the total and relative four-momenta, respectively, and KK is the interaction kernel. The bound state mass squared is M2=p2M^2=p^2. For nonconfining interactions, the mass MM is smaller than 2m2m, allowing the introduction of the binding energy BB (defined positive) through the relation M2=(2m−B)2M^2=(2m-B)^2. In the ladder approximation of the kernel, represented by the exchange of a scalar particle with mass μ\mu, the kernel has the form

iK(k,k′,p)=i(−ig)2(k−k′)2−μ2+iϵ,(5)iK(k,k',p)=\frac{i(-ig)^2}{(k-k')^2-\mu^2+i\epsilon}, \tag*{(5)}

leading to an attractive interaction and the possible emergence of bound states.

General properties of the solutions

The W-C model corresponds to the case μ=0\mu=0 in Eq. (5). In the non-relativistic limit, this model leads to the well-known Coulomb bound state spectrum. Cutkosky showed that in the relativistic case, the BS amplitude, henceforth limited to SS-wave states, characterized by a principal quantum number n=1,2,…n=1,2,\ldots, can be represented in terms of nn functions {gnν}ν=0,1,…,n−1\{g^\nu_n\}_{\nu=0,1,\ldots,n-1}, depending on a single scalar argument z∈[−1,+1]z\in[-1,+1], as

Φn(k,p)=−iNtot∑ν=0n−1∫−11gnν(z) dz m2(n−ν)+1[m2−14M2−k2−p⋅kz−iϵ]2+n−ν,n=1,2,…(6)\Phi_n(k,p)=-\frac{i}{\sqrt{N_{\mathrm{tot}}}}\sum_{\nu=0}^{n-1} \int_{-1}^{1}g^\nu_n(z)\,dz\, \frac{m^{2(n-\nu)+1}}{\left[m^2-\frac{1}{4}M^2-k^2-p\cdot k z-i\epsilon\right]^{2+n-\nu}},\quad n=1,2,\ldots \tag*{(6)}

NtotN_{\mathrm{tot}} is a dimensionless normalization factor, determined in Appendix C, ensuring the condition Fel(0)=1F_{\mathrm{el}}(0)=1 for the elastic form factor. The factor m2(n−ν)+1m^{2(n-\nu)+1} in the numerator is introduced to deal with dimensionless gnν(z)g^\nu_n(z) functions.

By inserting (6) in the BS equation (4), Cutkosky obtained Eq. (14) of [2] a system of homogeneous coupled integral equations for the functions gnνg_n^\nu. For S-waves, it reads:1

gnν(z)=λ2∑ν′=0ν(n−ν+1)(n−ν)(n−ν′+1)(n−ν′)∫−11dt∫01dx x(1−x)n−ν−1∫−1+1dz′ δ[z−xt−(1−x)z′][1−η2(1−z2)]ν−ν′+1gnν′(z′),(7)g_n^\nu(z)=\frac{\lambda}{2}\sum_{\nu'=0}^{\nu}\frac{(n-\nu+1)(n-\nu)}{(n-\nu'+1)(n-\nu')}\int_{-1}^{1}dt\int_{0}^{1}dx\,x(1-x)^{n-\nu-1}\int_{-1}^{+1}dz'\,\frac{\delta[z-xt-(1-x)z']}{[1-\eta^2(1-z^2)]^{\nu-\nu'+1}}g_n^{\nu'}(z'), \tag*{(7)}

where λ\lambda is related to the coupling constant g2g^2 of the interaction kernel (5) by λ=g216π2m2\lambda=\frac{g^2}{16\pi^2m^2}, and the total mass square M2M^2, eigenvalue of the system (7), appears through the parameter

η2=M24m2.\eta^2=\frac{M^2}{4m^2}.

Integrating (7), first with respect to tt through the δ\delta function, taking into account the bounds to be satisfied by tt and xx, and distinguishing the two cases, z>z′z>z' and z′>zz'>z, one obtains the set of equations

gnν(z)=λ2∑ν′=0νcnνν′∫−1+1dz′ [R(z,z′)]n−ν[Q(z′)]ν−ν′+1gnν′(z′),n=1,2,…,ν=0,1,…n−1,(8)g_n^\nu(z)=\frac{\lambda}{2}\sum_{\nu'=0}^{\nu}c_n^{\nu\nu'}\int_{-1}^{+1}dz'\,\frac{[R(z,z')]^{n-\nu}}{[Q(z')]^{\nu-\nu'+1}}g_n^{\nu'}(z'), \qquad n=1,2,\ldots,\quad\nu=0,1,\ldots n-1, \tag*{(8)}

where we have introduced

cnνν′=n−ν+1(n−ν′+1)(n−ν′),c_n^{\nu\nu'}=\frac{n-\nu+1}{(n-\nu'+1)(n-\nu')},
Q(z)=1−η2(1−z2),(9)Q(z)=1-\eta^2(1-z^2), \tag*{(9)}

and

R(z,z′)={1−z1−z′,for z′<z,1+z1+z′,for z′>z.(10)R(z,z')=\begin{cases}\frac{1-z}{1-z'},&\text{for }z'<z,\\[4pt]\frac{1+z}{1+z'},&\text{for }z'>z.\end{cases} \tag*{(10)}

By expanding (8), one is left with a n×nn\times n triangular system of one-dimensional integral equations of the form

gn0(z)=λ2[cn00∫−1+1dz′ [R(z,z′)]nQ(z′)gn0(z′)],(11)g_n^0(z)=\frac{\lambda}{2}\left[c_n^{00}\int_{-1}^{+1}dz'\,\frac{[R(z,z')]^n}{Q(z')}g_n^0(z')\right], \tag*{(11)}
gn1(z)=λ2[cn10∫−1+1dz′ [R(z,z′)]n−1[Q(z′)]2gn0(z′)+cn11∫−1+1dz′ [R(z,z′)]n−1Q(z′)gn1(z′)],g_n^1(z)=\frac{\lambda}{2}\left[c_n^{10}\int_{-1}^{+1}dz'\,\frac{[R(z,z')]^{n-1}}{[Q(z')]^2}g_n^0(z')+c_n^{11}\int_{-1}^{+1}dz'\,\frac{[R(z,z')]^{n-1}}{Q(z')}g_n^1(z')\right],
gn2(z)=λ2[cn20∫−1+1dz′ [R(z,z′)]n−2[Q(z′)]3gn0(z′)+cn21∫−1+1dz′ [R(z,z′)]n−2[Q(z′)]2gn1(z′)+cn22∫−1+1dz′ [R(z,z′)]n−2Q(z′)gn2(z′)],g_n^2(z)=\frac{\lambda}{2}\left[c_n^{20}\int_{-1}^{+1}dz'\,\frac{[R(z,z')]^{n-2}}{[Q(z')]^3}g_n^0(z')+c_n^{21}\int_{-1}^{+1}dz'\,\frac{[R(z,z')]^{n-2}}{[Q(z')]^2}g_n^1(z')+c_n^{22}\int_{-1}^{+1}dz'\,\frac{[R(z,z')]^{n-2}}{Q(z')}g_n^2(z')\right],

…………………\ldots\ldots\ldots\ldots\ldots\ldots\ldots,

gnn−1(z)=λ2[cnn−1,0∫−1+1dz′ R(z,z′)[Q(z′)]ngn0(z′)+…+cnn−1,n−1∫−1+1dz′ R(z,z′)Q(z′)gnn−1(z′)].g_n^{n-1}(z)=\frac{\lambda}{2}\left[c_n^{n-1,0}\int_{-1}^{+1}dz'\,\frac{R(z,z')}{[Q(z')]^n}g_n^0(z')+\ldots+c_n^{n-1,n-1}\int_{-1}^{+1}dz'\,\frac{R(z,z')}{Q(z')}g_n^{n-1}(z')\right].

Remarkably, the function gn0g_n^0, which allows the calculation of the energy spectrum via the M2M^2-dependence of QQ Eq. (9), is totally decoupled from the rest of the system. It fulfills the single equation (11), that we will hereafter write in terms of the fine structure coupling constant α\alpha, usual in the Coulomb problems:

gn0(z)=α2πn∫−11dz′ [R(z,z′)]nQ(z′)gn0(z′),(12)g_n^0(z)=\frac{\alpha}{2\pi n}\int_{-1}^{1}dz'\,\frac{[R(z,z')]^n}{Q(z')}g_n^0(z'), \tag*{(12)}
α=πλ=g216πm2.\alpha=\pi\lambda=\frac{g^2}{16\pi m^2}.

The remaining equations allow the determination of gnk>0g_n^{k>0} – and so of the BS amplitude (6) – by solving an inhomogeneous problem with an inhomogeneous term given by gn0g_n^0. Notice that it is a quite unusual situation in Quantum Mechanics that a part of the total system wave function, which, as we will see in what follows is far from being dominant, determines the full spectrum of the system.

Although the results presented here are limited to SS-wave only, it is worth noticing that for l≠0l\ne0 the corresponding spherical function YlmY_{lm} would appear as a prefactor in (6). The angular momentum ll would enter in the system of equations (8) and (11), but it turns out to be absent in the first equation (11) determining the spectrum. As a consequence, the BS amplitude would depend on ll, while the spectrum would remain ll-degenerate.

In view of its numerical solution, it is interesting to write (12) in a differential form:

gn0′′(z)+2(n−1)z(1−z2)−1gn0′(z)−n(n−1)(1−z2)−1gn0(z)+απ(1−z2)Q(z)gn0(z)=0,(13)g_n^{0\prime\prime}(z)+2(n-1)z(1-z^2)^{-1}g_n^{0\prime}(z)-n(n-1)(1-z^2)^{-1}g_n^0(z)+\frac{\alpha}{\pi(1-z^2)Q(z)}g_n^0(z)=0, \tag*{(13)}

with the boundary conditions gn0(±1)=0g_n^0(\pm1)=0.

For a fixed nn, Eq. (13) has an infinite number of solutions, labeled by an additional quantum number κ=0,1,2,…\kappa=0,1,2,\ldots, which also labels the corresponding discrete spectrum of mass squared eigenvalues Mnκ2M_{n\kappa}^2. We will use the notation gnκ0g_{n\kappa}^0 to identify a particular solution. The function gnκ0g_{n\kappa}^0 has κ\kappa nodes within the interval ]−1,+1[]-1,+1[ and a well-defined parity given by κ\kappa [2]:

gnκ0(−z)=(−1)κgnκ0(z)g_{n\kappa}^0(-z)=(-1)^\kappa g_{n\kappa}^0(z)

The parity is also preserved inside the ensemble {gnκν}\{g_{n\kappa}^{\nu}\} when varying ν=0,1,…\nu= 0,1,\ldots and this entails, through Eq. (6), that for even (odd) values of κ\kappa, the BS amplitude Φ(k,p)\Phi(k,p) is an even (odd) function of the relative energy k0k_0 in the c.m. frame.

The mass squared M2M^2 of the ground state g100g_{10}^{0} as function of the coupling constant α\alpha is shown in Figure 1. Its value vanishes for α=2π\alpha= 2\pi. In the range α∈[0,2π]\alpha\in[0,2\pi], M2≥0M^2 \ge0, the total mass of the system MM is well defined as well as its binding energy B=2m−M>0B = 2m-M > 0. This determines the domain where this model is physically consistent with a well-defined ground state. It is worth mentioning, however, that the solutions of the BS equation, as well as its spectral parameter M2M^2, can be analytically continued for α>2π\alpha> 2\pi without encountering any kind of singularity. This is illustrated with the dashed line in the lower right corner of the figure. All excited states lie above the M2(α)M^2(\alpha) curve and thus can have a well-defined MM even in the unphysical region.

Figure 1. (Color online) Dependence of the squared mass of the ground state (n=1n = 1, κ=0\kappa= 0) on the coupling constant α\alpha. Beyond the critical value α=2π\alpha= 2\pi, the solutions are smootly continued without any singularity but having negative values of M2M^2. The physical region, where the system has well-defined ground state mass and binding energy, is thus limited to α∈[0,2π]\alpha\in[0,2\pi].

Among the infinity of solutions existing for a given nn, the one with κ=0\kappa= 0 coincides, in the limit of small binding energies B/m≪1B/m \ll1, with the solution of the non-relativistic Coulomb problem with main quantum number nn. This solution is called, following the original works of Wick and Cutkosky [1, 2], “normal”. Indeed, these authors, analyzing in this limit the system of equations for the functions gnκ(z)g_n^\kappa(z) determining the BS amplitude (6), reproduced, for κ=0\kappa= 0, the Coulomb spectrum, i.e. the Balmer series

Bn=mα24n2.(14)B_n = \frac{m\alpha^2}{4n^2}. \tag*{(14)}

This result corresponds to the Schrödinger equation with the potential V(r)=−αrV(r) = -\frac{\alpha}{r}. The relativistic perturbative correction to the binding energy (14) was found in [11]; the binding energy, incorporating it, reads

Bn=mα24n2[1−4απln⁡(1α)+O(α)].(15)B_n = \frac{m\alpha^2}{4n^2}\left[1-\frac{4\alpha}{\pi}\ln\left(\frac{1}{\alpha}\right)+O(\alpha)\right]. \tag*{(15)}

On the contrary, the solutions corresponding to non-zero values of κ\kappa (κ=1,2,…\kappa= 1,2,\ldots), have a spectrum totally decoupled from the non relativistic one. They are genuinely of relativistic nature, without non-relativistic counterparts, and were named ”abnormal” by Wick.

These different behaviours are illustrated in Figure 2 where we have displayed the dependence of the coupling constant λ=απ\lambda= \frac{\alpha}{\pi} on the binding energy BB for the lowest solutions of the W-C model. Upper panel contains only the n=1n=1 states with κ=0,1,2,3,4\kappa= 0,1,2,3,4. The curve corresponding to κ=0\kappa= 0 (black solid line) is tangent to the non-relativistic (NR) one (black dashed line) from which it departs logarithmically, as it is visible, starting at B≈0.001B \approx0.001. The perturbative results, provided by Eq. (15), are indistinguishable from the exact ones in the considered energy range. Those corresponding to κ>0\kappa> 0 (colored solid lines) do not have any non-relativistic counterparts. Lower panel represents the spectrum for n≥1n \ge1 states and different values of κ=0,1,2,3\kappa= 0,1,2,3. The horizontal line λ=2\lambda= 2 (α=2π\alpha= 2\pi) indicates the maximal value of the coupling constant ensuring a well-defined ground state.

Spectrum of the W-C model as a function of the coupling constant

Figure 2. (Color online) Spectrum of the W-C model as a function of the coupling constant λ(B)\lambda(B) in the low energy limit. Upper panel corresponds to n=1n=1 states with different values of κ=0,1,…\kappa= 0, 1, \ldots. The case κ=0\kappa= 0 is compared to the non relativistic solution. Horizontal lines correspond respectively to the maximal values of the coupling constant for a well-defined ground state (λ=2\lambda= 2) and to the minimal value for which the abnormal solutions exist (λ=1/4\lambda= 1/4). Lower panel contains the full spectrum for n≤6n \le6 and κ≤3\kappa\le3 to make explicit the different crossings. Notice that the unphysical solutions with odd κ\kappa (giving no contribution to the SS-matrix) are naturally inserted in the spectrum.

The normal and abnormal solutions have also different domains of existence with respect to the coupling constant α\alpha. As a mathematical solution of the BS equation (4), the normal solutions exist for any (positive) values of α\alpha, although, as we have already discussed, they have a clear physical meaning only in the range α∈[0,2π]\alpha\in[0,2\pi], where M2≥0M^2 \ge0. However, the very existence of the abnormal solutions (all of them) requires a coupling constant greater than some critical value, α≥π4\alpha\ge\frac{\pi}{4} (λ≥1/4\lambda\ge1/4). This can be clearly seen from the results of Figure 2, where all the abnormal states (color line) were found above the horizontal λ=1/4\lambda= 1/4 line.

Wick and Cutkosky [1, 2] found the following approximate analytic expression for the abnormal spectrum near the continuum threshold (B→0B \to0):

Mnκ2≃4m2(1−e−(κ−1)π/α/π−1/4),κ=2,3,…(16)M_{n\kappa}^{2} \simeq4m^{2}\left(1-e^{-(\kappa-1)\pi/\sqrt{\alpha/\pi-1/4}}\right), \quad\kappa= 2,3,\ldots \tag*{(16)}

where the condition α>π/4\alpha> \pi/4 is explicitly obtained. At B/m≪1B/m \ll1 this spectrum vs. κ\kappa does not depend on nn. It also indicates that to be able to distinguish abnormal states from the continuum threshold on experimental grounds, the coupling constant should be increased at least up to values of α≈4÷5\alpha\approx4 \div5; otherwise, for values of α\alpha very close to π/4\pi/4, the exponential in Eq. (16) is nearly zero and the discrete spectrum becomes hardly distinguishable from the continuum.

It is worth noting that the existence of a lower bound of the coupling constant for the abnormal solutions is reminiscent of the massive-exchange case, i.e. μ≠0\mu\ne0 in the kernel (5), which was considered with some detail in [12] both in the BS and the Light-Front Dynamics frameworks. The Bμ(α)B_\mu(\alpha) dependences (Figure 5 and Figure 7 of [12]) are similar to those displayed in Figure 2, what suggest the possibility to associate a mass with the abnormal states. However, essential differences in the number of bound states for a given μ\mu remain: infinite in the W-C model and (at most) finite in the massive case.

Figure 5. Components g200g_{20}^0 and g201g_{20}^1 of the normal state No. 2 from Table 1.

Figure 7. (Color online) N2N_{2}-dependence on the binding energy for the n=1n=1 and n=2n=2 states: normal (upper panel) and abnormal (lower panel).

In summary, the range of the coupling constants to be considered in this model is 0<α<2π0 < \alpha< 2\pi (0<λ<20 < \lambda< 2) for the normal states, i.e. κ=0\kappa= 0, and π4<α<2π\frac{\pi}{4} < \alpha< 2\pi (1/4<λ<21/4 < \lambda< 2) for the abnormal ones (κ>0\kappa> 0). On another hand, according to Refs. [13, 14], the abnormal solutions with odd values of κ\kappa do not contribute to the SS-matrix and therefore only those with even κ\kappa can have a physical meaning. In the subsequent part of this work, we will concentrate on the latter case. Furthermore we will restrict ourselves, to the l=0l = 0 states with n=1n = 1 and n=2n = 2.

For n=1n = 1 states, the sum (6) is reduced to a single term involving only the function g10g^0_1 satisfying the homogeneous integral equation

g10(z)=α2π∫−11R(z,z′)Q(z′)g10(z′) dz′,(17)g^0_1(z) = \frac{\alpha}{2\pi}\int_{-1}^{1}\frac{R(z,z')}{Q(z')}g^0_1(z')\,\mathrm{d}z', \tag*{(17)}

or, equivalently, in its differential form

g1′′0(z)+απ1(1−z2)Q(z)g10(z)=0,(18)g^{\prime\prime}_1{}^0(z) + \frac{\alpha}{\pi}\frac{1}{(1-z^2)Q(z)}g^0_1(z) = 0, \tag*{(18)}

with the boundary conditions g10(±1)=0g^0_1(\pm1) = 0. The corresponding BS amplitude is expressed in terms of g10g^0_1 as

Φ1(k,p)=∫−11−im3g10(z) dz[m2−14M2−k2−p⋅kz−iϵ]3.(19)\Phi_1(k,p) = \int_{-1}^{1}\frac{-\mathrm{i}m^3 g^0_1(z)\,\mathrm{d}z}{[m^2-\frac{1}{4}M^2-k^2-p\mathbin{\cdot}kz-\mathrm{i}\epsilon]^3}. \tag*{(19)}

For n=2n = 2 states, the sum (6) involves two functions g20g^0_2 and g21g^1_2. The function g20g^0_2 satisfies the homogeneous integral equation:

g20(z)=α4π∫−11R2(z,z′)Q(z′)g20(z′) dz′,(20)g^0_2(z) = \frac{\alpha}{4\pi}\int_{-1}^{1}\frac{R^2(z,z')}{Q(z')}g^0_2(z')\,\mathrm{d}z', \tag*{(20)}

and in differential form

g2′′0(z)+2z1−z2g2′0(z)−21−z2g20(z)+απ1(1−z2)Q(z)g20(z)=0,(21)g^{\prime\prime}_2{}^0(z) + \frac{2z}{1-z^2}g^{\prime}_2{}^0(z) - \frac{2}{1-z^2}g^0_2(z) + \frac{\alpha}{\pi}\frac{1}{(1-z^2)Q(z)}g^0_2(z) = 0, \tag*{(21)}

with the boundary conditions g20(±1)=0g^0_2(\pm1) = 0, while g21g^1_2 is determined from g20g^0_2 through the integral equation

g21(z)=α6π∫−11R(z,z′)[Q(z′)]2g20(z′) dz′+α2π∫−11R(z,z′)Q(z′)g21(z′) dz′,(22)g^1_2(z) = \frac{\alpha}{6\pi}\int_{-1}^{1}\frac{R(z,z')}{[Q(z')]^2}g^0_2(z')\,\mathrm{d}z' + \frac{\alpha}{2\pi}\int_{-1}^{1}\frac{R(z,z')}{Q(z')}g^1_2(z')\,\mathrm{d}z', \tag*{(22)}

which can also be rewritten in the form of an inhomogeneous differential equation:

g2′′1(z)+απ1(1−z2)Q(z)g21(z)=−α3π1(1−z2)[Q(z)]2g20(z).(23)g_2^{\prime\prime1}(z)+\frac{\alpha}{\pi}\frac{1}{(1-z^2)Q(z)}g_2^1(z)=-\frac{\alpha}{3\pi}\frac{1}{(1-z^2)[Q(z)]^2}g_2^0(z). \tag*{(23)}

The BS amplitude (6) is now expressed in terms of two functions g21g_2^1 and g20g_2^0:

Φ2(k,p)=∫−11−im3g21(z)dz[m2−14M2−k2−p⋅kz−iϵ]3+∫−11−im5g20(z)dz[m2−14M2−k2−p⋅kz−iϵ]4.(24)\Phi_2(k,p)=\int_{-1}^{1}\frac{-im^3g_2^1(z)dz}{[m^2-\frac{1}{4}M^2-k^2-p\cdot kz-i\epsilon]^3}+\int_{-1}^{1}\frac{-im^5g_2^0(z)dz}{[m^2-\frac{1}{4}M^2-k^2-p\cdot kz-i\epsilon]^4}. \tag*{(24)}

Numerical solutions for some selected states gnκνg_{n\kappa}^\nu

We present, in this subsection, the numerical results concerning the first states of the W-C spectrum. We fix hereafter m=1m=1 and the coupling constant to the value α=5\alpha=5. We will consider along the work an ensemble of states with n=1,2n=1,2 and κ=0,2,4\kappa=0,2,4 that, for the sake of simplicity in notation, will be numbered with No. 1-6 in the Tables 1 and 2.

The binding energies for the lowest n=1,2n=1,2 normal (κ=0\kappa=0) and abnormal (κ=2\kappa=2) states gnκνg_{n\kappa}^\nu are presented in Table 1. All ν=0\nu=0 components are arbitrarily normalized to gnκ0(0)=1g_{n\kappa}^0(0)=1. The corresponding solutions for the n=1n=1 states – g100g_{10}^0 and g120g_{12}^0 – are displayed in Figure 3 and Figure 4. They have comparable sizes and their nodal structure is determined by κ\kappa only.

No.nnκ\kappaBBN2N_2
1100.9992590.65
2200.2084100.61
3123.51169⋅10−33.51169\cdot10^{-3}0.094
4221.12118⋅10−31.12118\cdot10^{-3}0.077

Table 1. Binding energy BB and two-body norm (N2N_2) of the low-lying normal (κ=0\kappa=0) and abnormal (κ=2\kappa=2) states, for the coupling constant value α=5\alpha=5.

Figure 3. g100g_{10}^0 for the normal state No. 1 of Table 1.

Figure 4. g120g_{12}^0 for the abnormal state No. 3 of Table 1.

The two components g2κ0g_{2\kappa}^0 and g2κ1g_{2\kappa}^1 of the n=2n=2 states are plotted in Figure 5 (state No. 2 with κ=0\kappa=0) and Figure 6 (state No. 4 with κ=0\kappa=0). The component ν=1\nu=1 is dominant in both cases, but for the state No. 4 it is ∼1000\sim1000 times larger (see the scaling factor in Figure 6). This enhancement is due to the Q2Q^2 factor in the denominator of the right-hand-side of Eq. (23), which, in the limit B→0B\to0 and around z=0z=0, behaves as Q2≈B2Q^2\approx B^2. Thus, for n>1n>1 states with small binding energies,

Figure 6. g220g_{22}^{0} and g221g_{22}^{1} (scaled by a factor 10310^{3}) of the abnormal state No. 4 (Table 1).

the component gnκ0g_{n\kappa}^0 that determines M2M^2 is negligibly small with respect to the other components. We will see, however, in the following section that they all play equivalent roles in the construction of the BS amplitude itself and in the form factors.

In the rightest column of Table 1 we have also included the norm N2N_{2} of the two-body contributions in the Fock space, as it is defined in Appendix A. For the states with n=1n=1, N2N_{2} is given by Eq. (A.6) and for n=2n=2, by Eqs. (A.7). We remark therein that the two-body norm N2N_{2} for the abnormal (κ=2\kappa=2) states is much smaller than for the normal (κ=0\kappa=0) ones. This comparison concerns however states covering the two extreme cases in the spectrum: deeply bound states (Nos. 1 and 2) and nearthreshold ones (Nos. 3 and 4). To better understand this difference, we have studied the dependence of N2N_{2} on the binding energy for the first normal and abnormal states. Results are displayed in Figure 7.

The upper panel concerns the normal states. The behaviours of the n=1n=1 (black solid line) and n=2n=2 (red solid line) states are quite similar: N2N_{2} decreases monotonically from N2≈1N_{2}\approx1 when B≈0B\approx0 down to an asymptotic value when B→2mB\to2m. We found numerically N2(2m)≈0.64N_{2}(2m)\approx0.64 for n=1n=1 and N2(2m)≈0.59N_{2}(2m)\approx0.59 for n=2n=2. For the ground state n=1n=1, these limiting values were found analytically in [9], as well as the perturbative expansion in their vicinity. Thus, the limit B→0B\to0 is described by Eq. (A.13), i.e.,

N2(B)=1+1π4Bmln⁡(4Bm).N_{2}(B)=1+\frac{1}{\pi}\sqrt{\frac{4B}{m}}\ln\left(\frac{4B}{m}\right).

At B/m=10−5B/m=10^{-5}, this perturbative expansion gives N2=0.980N_{2}=0.980 in close agreement with the black curve of Figure 7. We conclude from this study that the normal states are dominated by two-body norms. This is particularly true in the limit B→0B\to0, where N2→1N_{2}\to1, but remains also true in all the energy domain, although decreasing with increasing BB.

A very different behaviour is observed with the abnormal states, represented in the lower panel of Figure 7.

As one can see, the two body norm N2N_{2} of these states not only remains comparatively very small, but also vanishes in the non-relativistic limit, making them, in this region, genuine many-body states. The one order of magnitude observed in Table 1 for the binding energies hides in fact a deeper and striking difference between normal and abnormal BS states, independent of their comparison with the non-relativistic spectrum. It is provided by their two-body content: abnormal states do not have in the limit B→0B\to0 any two-body contribution and have, thus, genuine many-body structures. Beyond this limit the norm of the two-body sector remains extremely small. This is the reason why they are absent in the non-relativistic limit reduced to the two-body Schrödinger equation.

The results presented in Table 1 are completed in Table 2 by studying the κ=4\kappa=4 excitations of n=1,2n=1,2 states. The same conclusion holds, even in a more dramatic way. Their two body norms are one order of magnitude smaller than for the κ=2\kappa=2 states of Table 1. This can be expected due to their smaller binding energies and in view of the behaviour described in the lower panel of Figure 7.

The abnormal solution g140g_{14}^{0} for n=1,κ=4n=1,\kappa=4 state (No. 5 in Table 2), is shown in Figure 8, displaying its more involved nodal structure (4 zeros in ]−1,+1[]-1,+1[). The functions g24νg_{24}^{\nu} of the abnormal state n=2,κ=4n=2,\kappa=4 (No. 6 in Table 2), are plotted in Figure 9. The extreme smallness of the binding energy of this state generates a huge enhancement factors in the inhomogeneous equation (23) (through the factor QQ) which results in a huge dominance of the ν=1\nu=1 component in the full BS amplitude. Notice that g241g_{24}^{1} has been reduced by a factor 10510^{5} to become comparable with g240g_{24}^{0}.

Figure 8. g140g_{14}^{0} from state No. 5 in Table 2.

Plot of $g_{24}^{\nu}$ and $g_{24}^{0}/10^{5}$ for the state with $n=2$ and $\kappa=4$

Figure 9. (Color online) g24νg_{24}^{\nu} from state No. 6 in Table 2.

Electromagnetic form factors

We suppose that one of the two constituent particles is charged. The electromagnetic form factor of the system can be expressed in terms of its BS amplitude. It is enough to consider inelastic transitions from an initial ∣i⟩|i\rangle to a final ∣f⟩|f\rangle state. The elastic form factors are obtained from them as a particular case, with f=if=i.

The electromagnetic vertex JμJ_{\mu}, corresponding to a transition ∣i⟩→∣f⟩|i\rangle\to|f\rangle is shown graphically in Figure 10. The corresponding vertex amplitude reads (we use Itzykson and Zuber [15] conventions for the Feynman rules):

Figure 10. Feynman diagram for the electromagnetic form factor.

iJμ=∫d4k(2π)4i[−i(p+p′−2k)μ](k2−m2+iϵ)×i[−iΓ‾(12p′−k,p′)][(p′−k)2−m2+iϵ]i[iΓ(12p−k,p)][(p−k)2−m2+iϵ](25)iJ_{\mu}=\int\frac{d^{4}k}{(2\pi)^{4}}\frac{i[-i(p+p'-2k)_{\mu}]}{(k^{2}-m^{2}+i\epsilon)} \times\frac{i[-i\overline{\Gamma}(\frac{1}{2}p'-k,p')]}{[(p'-k)^{2}-m^{2}+i\epsilon]}\frac{i[i\Gamma(\frac{1}{2}p-k,p)]}{[(p-k)^{2}-m^{2}+i\epsilon]} \tag*{(25)}

where Γ(k,p)\Gamma(k,p) is the vertex function, related to the BS amplitude by the equation

Φ(k,p)=Γ(k,p)[(p2+k)2−m2+iϵ][(p2−k)2−m2+iϵ].(26)\Phi(k,p)=\frac{\Gamma(k,p)}{\left[(\frac{p}{2}+k)^{2}-m^{2}+i\epsilon\right]\left[(\frac{p}{2}-k)^{2}-m^{2}+i\epsilon\right]}. \tag*{(26)}

Γ‾\overline{\Gamma} is the conjugate of Γ\Gamma, obtained from the latter by complex conjugation and use of the antichronological product. A similar definition also holds for the BS amplitude Φ‾\overline{\Phi} [4].

The electromagnetic vertex for the transition i→fi\to f is expressed in terms of the BS amplitude as (see e.g. Eq. (7.1) in [5]):

Jμ=i∫d4k(2π)4(p+p′−2k)μ(k2−m2×Φ‾f(12p′−k,p′)Φi(12p−k,p).(27)\begin{aligned} J_{\mu}=i\int\frac{d^{4}k}{(2\pi)^{4}}(p+p'-2k)_{\mu}(k^{2}-m^{2} \\ &\times\overline{\Phi}_{f}\left(\frac{1}{2}p' - k,p'\right)\Phi_{i}\left(\frac{1}{2}p-k,p\right). \tag*{(27)} \end{aligned}

It has the following general decomposition in terms of two scalar functions (see2 [16]): F(Q2)F(Q^{2}) and G(Q2)G(Q^{2})

Jμ=[(pμ+pμ′)+(pμ′−pμ)Qc2Q2]F(Q2)−(pμ′−pμ)Qc2Q2G(Q2).(28)J_{\mu}=\left[(p_{\mu}+p'_{\mu})+(p'_{\mu}-p_{\mu})\frac{Q_{c}^{2}}{Q^{2}}\right]F(Q^{2})-(p'_{\mu}-p_{\mu})\frac{Q_{c}^{2}}{Q^{2}}G(Q^{2}). \tag*{(28)}

Here, q=p′−pq=p'-p, Q2=−q2=−(p′−p)2Q^{2}=-q^{2}=-(p'-p)^{2} and

Qc2=Mf2−Mi2,(29)Q_{c}^{2}=M_{f}^{2}-M_{i}^{2}, \tag*{(29)}

MiM_{i} and MfM_{f} being the masses of the initial and final states, respectively.

Since q⋅J=Qc2G(Q2)q\cdot J=Q_{c}^{2}G(Q^{2}), the above decomposition does not suppose, in general, current conservation q⋅J=0q\cdot J=0, which implies

G(Q2)≡0.(30)G(Q^{2})\equiv0. \tag*{(30)}

A direct proof of this result from the BS equation is presented in Appendix B. The current conservation becomes a stringent self-consistency criterion of our results: the calculated form factor G(Q2)G(Q^{2}) should be identically zero, or very small within the numerical uncertainties.

We also note that since JμJ_{\mu} [Eq. (28)] is not singular at Q2=0Q^{2}=0, the following relation should hold:

F(0)=G(0).(31)F(0)=G(0). \tag*{(31)}

Equation (30) then implies that F(0)=0F(0)=0 for the transition form factors (for which Qc2≠0Q_{c}^{2}\ne0).

From Eq. (28), the form factors are expressed through JμJ_{\mu} as

F(Q2)=(p+p′)⋅JQ2+q⋅JQc2[(Mf−Mi)2+Q2][(Mf+Mi)2+Q2]F(Q^{2})=\frac{(p+p')\cdot J Q^{2}+q\cdot J Q_{c}^{2}}{[(M_{f}-M_{i})^{2}+Q^{2}][(M_{f}+M_{i})^{2}+Q^{2}]}
G(Q2)=q⋅JQc2.(32)G(Q^{2})=\frac{q\cdot J}{Q_{c}^{2}}. \tag*{(32)}

Considering these formulas at Q2=0Q^{2}=0 (which does not imply q=0q=0), one gets the relation F(0)=q⋅JQc2∣Q2=0=G(0)F(0)=\left.\frac{q\cdot J}{Q_{c}^{2}}\right|_{Q^{2}=0}=G(0), which reproduces Eq. (31).

The expressions for the form factors are obtained by substituting in Eqs. (32) the current JJ [Eq. (27)], then substituting the BS amplitudes (6), using the Feynman parametrization and integrating over kk. In this way, we find the form factors in the form of integrals over products of functions gnn(z)g_{n}^{n}(z) and gn′n′(z′)g_{n'}^{n'}(z'). (Details of similar calculations can be found in Ref. [17].) The result for4: We change the notations in comparison to Ref. [16], where the form factor GG was denoted F′F'.

the transition form factor F(Q2)F(Q^{2}) can be written in the form:

F(Q2)=∑ν=0n−1∑ν′=0n′−1Fnn′νν′,(33)F(Q^{2})=\sum_{\nu=0}^{n-1}\sum_{\nu'=0}^{n'-1}F_{nn'}^{\nu\nu'}, \tag*{(33)}

and similarly for G(Q2)G(Q^{2}). The expressions of the functions of the right-hand-side of this equation for the cases n=n′=1n=n'=1 and n=n′=2n=n'=2 are given in Appendix C.

In the following subsections, we examine the numerical results for the elastic and transition form factors for some states from Tables 1 and 2, corresponding to the coupling constant α=5\alpha=5.

Elastic form factors Fe(Q2)F_{e}(Q^{2})

There are two regions of interest in studying the elastic form factors (FeF_{e}): the region near the origin, giving insight into the size of the system, and the asymptotic region Q2≫m2Q^{2}\gg m^{2}, related to the many-body structure of the wave function [6, 7, 8].

With the elastic form factor of a bound state, normalized to Fe(Q2=0)=1F_{e}(Q^{2}=0)=1, the squared radius is given by

⟨r2⟩=−6(dFedQ2)Q2=0(34)\langle r^{2}\rangle=-6\left(\frac{dF_{e}}{dQ^{2}}\right)_{Q^{2}=0} \tag*{(34)}

and the root mean squared (r.m.s.) radius by R=⟨r2⟩R=\sqrt{\langle r^{2}\rangle}. In the non-relativistic theory, the size of a bound state scales, as a function of its binding energy, as R≈1mBR\approx\frac{1}{\sqrt{mB}}.

On the other hand, as mentioned in the Introduction, the asymptotics of FeF_{e} should qualitatively probe the compositeness of the state. According to [6, 7, 8], the elastic form factors of a nn-body system should decrease as 1/(Q2)n−11/(Q^{2})^{n-1}, where nn is the number of the constituents of the state. It is, however, worth emphasizing here some essential differences of the W-C model with the theoretical framework in which the above asymptotic behaviors have been obtained. The latter have been derived in theories characterized by dimensionless coupling constants, like QCD and the parton model. In the W-C model, bosonic fields interact by the exchange of a scalar particle; the coupling constant gg is then dimensionful, having the dimension of mass. This has an immediate consequence on the behavior of the BS amplitude at large momenta. In the W-C model, as can be checked from Eqs. (4) and (5), the BS amplitude behaves at large momenta as 1/(k2)31/(k^{2})^{3}. In QCD, in the ladder approximation, where quarks interact by means of an exchange of a gluon field, the scalar part of the BS amplitude behaves at large momenta as 1/(k2)21/(k^{2})^{2}, up to logarithms. Extending the comparison to bosonic ϕ4\phi^4-like theories, where the coupling constant is also dimensionless, the interaction between two bosonic constituents is realized either by contact terms, or by the exchange of a two-particle loop; in both cases the large-momentum behavior of the BS amplitude is again 1(k2)2\frac{1}{(k^2)^2}, up to logarithms. The faster decrease of the BS amplitude in the W-C model affects the behaviors of the form factors: one thus expects in this model behaviors of the type 1/(Q2)n1/(Q^2)^n, up to logarithms, instead of 1/(Q2)n−11/(Q^2)^{n-1}.

In QCD, the number nn represents the number of valence quarks (including, eventually, the number of valence gluons, in the case of hybrids). The Fock sectors of the state which contain the sea quarks, therefore with higher nn’s, are expected to decrease asymptotically faster and thus to display rapidly the asymptotic dominance of the valence quark sector, a phenomenon well observed on experimental grounds. In the W-C model, for the normal solutions, one indeed expects the dominance of the two-body sector, as was concluded in § 3. However, for the abnormal solutions, the two-body sector is weakly contributing to the composition of the corresponding states, which are dominated by higher sectors of the Fock space. One therefore expects here a competition between the contributions of the higher sectors of the Fock space, which asymptotically decrease more rapidly, but have large coefficients, and the contribution of the two-body sector, which dominates in the asymptotic region, but with a small coefficient.

For abnormal solutions, or hybrid states, of the W-C model, a refined analysis necessitates the distinction between three regimes in the Q2Q^2 evolution of the form factors, rather than two: the very small Q2Q^2 regime, determined by the binding energy (and equivalently by the rms radius), the intermediate Q2Q^2 region where the decrease is determined by the many-body components (and therefore is fast) and the asymptotic region, where the many-body contribution is exhausted, and only the two-body contribution survives. Since in the W-C model the content of any state is a two-body component plus an indefinite number of exchange particles, the asymptotic behavior of all the elastic form factors is finally determined by its two-body contribution. Therefore, the asymptotic Q2Q^2-dependence should be the same, though with different coefficients, for all elastic form factors. In particular, we predict that the ratios of all elastic form factors should tend to constants at Q2→∞Q^2 \to\infty.

We first examine the n=1n=1 states. They are defined by a single component gnκ0g^0_{n\kappa}, the same that determines the binding energy. We have plotted, in Figure 11, in solid lines, the elastic form factors for the two n=1n=1 states of Table 1: on top, the normal state No. 1 (κ=0\kappa=0, B=0.999B=0.999), and at bottom, the abnormal state No. 3 (κ=2\kappa=2, B=0.00359B=0.00359).

Figure 11. Elastic form factors of n=1n=1 states of Table 1 (solid lines): No. 1 (normal) on top and No. 3 (abnormal) at bottom; the dashed line corresponds to the state with κ=0\kappa=0 with the same binding energy (and rms radius) as the state No. 3 with κ=2\kappa=2.

The corresponding rms radii are R1=1.16R_1=1.16 fm and R3=15.7R_3=15.7 fm, respectively, which roughly scale as 1mB\frac{1}{\sqrt{mB}}. Both states have only one component gnκ0g^0_{n\kappa}, which in turn determines the energy. It is then natural that they have a similar behaviour, close to that of the non-relativistic one. At this level, one cannot see any drastic difference between a normal and an abnormal state. The behaviours of their form factors are quite similar, however with a much faster decrease for the state No. 3, as expected from its smaller binding energy and its many-body structure. At Q2=1Q^2=1, the value of FeF_e for the state No. 3 is three orders of magnitude smaller than that of the state No. 1. In order to disentangle the contributions of the binding energy and of the asymptotic behaviour, we have adjusted, at a second step, α\alpha of the state No. 1 to have the same binding energy as the state No. 3. The result is displayed in dashed line on the lower panel: both curves are tangent to each other at the origin, implying that the rms radii are the same, but one notices that the abnormal state form factor still decreases much faster than that of the normal one, by a factor 10 at Q2=1Q^2 = 1.

One can show that the elastic form factors behave, when Q2/m2→∞Q^2/m^2 \to\infty as

Fe(Q2)=(m2Q2)2[c2ln⁡(Q2m2)+c0]+O((m2Q2)3ln⁡(Q2m2),(m2Q2)3).(35)F_e(Q^2) = \left(\frac{m^2}{Q^2}\right)^2 \left[c_2 \ln\left(\frac{Q^2}{m^2}\right) + c_0\right] + \mathcal{O}\left(\left(\frac{m^2}{Q^2}\right)^3 \ln\left(\frac{Q^2}{m^2}\right),\left(\frac{m^2}{Q^2}\right)^3\right). \tag*{(35)}

For the states n=1n = 1, the coefficient c2c_2 has a simple expression in terms of the BS amplitude:

c2(n=1)=14π2[g1κ′0(−1)]2Ntot,(36)c_2(n = 1) = \frac{1}{4\pi^2}\frac{\left[g_{1\kappa}^{\prime0}(-1)\right]^2}{N_{\mathrm{tot}}}, \tag*{(36)}

where g′(−1)g^\prime(-1) is the derivative of gg with respect to zz at z=−1z = -1 and NtotN_{\mathrm{tot}} is the normalization factor that ensures the condition Fe(0)=1F_e(0) = 1 when gg is arbitrarily normalized (its expression is given in Eq. (C.18)). The expression of c0c_0 is more complicated and depends on the bound state mass MM, as well as on the function gg over the whole region of zz in the interval [−1,+1][-1,+1].

For the normal state n=1n = 1, κ=0\kappa= 0, c0c_0 is negative in general, but changes sign and becomes positive for small binding energies. The expression of gg takes a simple form in the two extreme cases of non-relativistic limit and maximal binding energy (M=0M = 0). Normalizing gg so that g(0)=1g(0) = 1, one has in the first case g(z)=(1−∣z∣)g(z) = (1 - |z|), with Ntot=1/(32πα5)N_{\mathrm{tot}} = 1/(32\pi\alpha^5), and in the second case g(z)=(1−z2)g(z) = (1 - z^2), with Ntot=1/(270π2)N_{\mathrm{tot}} = 1/(270\pi^2) [1, 2]. One obtains, in these two extreme cases, the asymptotic behaviors [9]

Fe(Q2)≃8α5π(m2Q2)2[ln⁡(Q2m2)+2πα](37)F_e(Q^2) \simeq\frac{8\alpha^5}{\pi}\left(\frac{m^2}{Q^2}\right)^2\left[\ln\left(\frac{Q^2}{m^2}\right)+\frac{2\pi}{\alpha}\right] \tag*{(37)}

(non-relativistic limit M→2mM \to2m),

Fe(Q2)≃270(m2Q2)2[ln⁡(Q2m2)−4](38)F_e(Q^2) \simeq270\left(\frac{m^2}{Q^2}\right)^2\left[\ln\left(\frac{Q^2}{m^2}\right)-4\right] \tag*{(38)}

(ultra-relativistic case M=0M = 0).

Notice that in Eq. (37), we have neglected α\alpha-independent constant factors in front of the additive term 2π/α2\pi/\alpha. The fact that c0c_0 is generally negative, except for small binding energies, where it can, however, take a large value (proportional to 1/α1/\alpha), has as a main consequence the screening of the logarithmic tail, requiring, for a numerical analysis of the asymptotic behaviors, very large values of Q2Q^2 (Q2≫100÷1000 m2Q^2 \gg100 \div1000\,m^2).

In order to put in evidence the asymptotic behavior (35) and to determine its leading terms we have computed the “reduced form factor” Fˉe(Q2)\bar{F}_e(Q^2), defined as

Fˉe(Q2)≡1ln⁡(Q2/m2)(Q2m2)2Fe(Q2)\bar{F}_e(Q^2) \equiv\frac{1}{\ln\left(Q^2/m^2\right)}\left(\frac{Q^2}{m^2}\right)^2 F_e(Q^2)
lim⁡Q2→∞Fˉe(Q2)=c2+c0ln⁡(Q2/m2),(39)\lim_{Q^2\to\infty}\bar{F}_e(Q^2) = c_2 + \frac{c_0}{\ln\left(Q^2/m^2\right)}, \tag*{(39)}

which should tend, when Q2→∞Q^2 \to\infty, to a constant (up to logarithmic corrections). The asymptotic coefficients cic_i can be extracted from Fˉe(Q2)\bar{F}_e(Q^2) and its derivative at a given Q2Q^2 with the relations

c0=−dFˉ(Q2)dQ2Q2ln⁡2(Q2m2)(40)c_0 = -\frac{d\bar{F}(Q^2)}{dQ^2}Q^2\ln^2\left(\frac{Q^2}{m^2}\right) \tag*{(40)}

and

c2=Fˉ(Q2)−c0ln⁡(Q2/m2)(41)c_2 = \bar{F}(Q^2) - \frac{c_0}{\ln\left(Q^2/m^2\right)} \tag*{(41)}

The results for the n=1n = 1 state No. 1 from Table 1 are displayed in Figure 12. The upper panel represents the elastic form factor multiplied by QσQ^\sigma with σ=2\sigma= 2 (red line), σ=4\sigma= 4 (blue line) and σ=4\sigma= 4 divided by the logarithmic term, as in Eq. (39), to exhibit the asymptotic behaviour derived in Eq. (35) and the important contribution that the logarithmic term can have in the domain Q2∈[0,1000]Q^2 \in[0,1000], even at Q2∼1000Q^2 \sim1000. The latter is seen in the difference between the blue and black lines, which exactly correspond to Fˉe\bar{F}_e. In the lower panel we have plotted the coefficients c0c_0 and c2c_2 in the “asymptotic” domain Q2∈[100,1000]Q^2 \in[100,1000], together with the full reduced form factor Fˉe\bar{F}_e. They already show a nice convergence at Q2=1000Q^2 = 1000, but the difference between Fˉe\bar{F}_e and its asymptotic value c2c_2 remains sizeable, due to the large contribution of c0c_0, which decreases very slowly. We have also checked the stability of our results with respect to the number of grid points (ngrid=400,800,1600n_{\mathrm{grid}} = 400,800,1600) used in computing the form factors: the sensitivity is not visible by eyes and is not significant in our analysis.

Figure 12. Asymptotic behaviours of the elastic form factor of the n=1n = 1 state No. 1 of Table 1. Upper panel: the elastic form factor FeF_e multiplied by QσQ^\sigma with σ=2\sigma= 2 (red), σ=4\sigma= 4 (blue), and σ=4\sigma= 4 divided by ln⁡Q2\ln Q^2. Lower panel: the asymptotic coefficients c0c_0 and c2c_2 defined in Eq. (39).

Figure 13. Same as in Fig. 12, but for the n=1n = 1 abnormal state No. 3 from Table 1.

Fig. 13 contains the same results for the n=1n = 1 abnormal state No. 3. Due to the faster decrease of the corresponding elastic form factor, (see the lower panel of Figure 11) the asymptotic regime is reached at Q2≈50Q^2 \approx50, with the asymptotic constant c2≈0.0004c_2 \approx0.0004, which is seven orders of magnitude smaller than for the normal state No. 1.

Figures 14 and 15 contain the elastic form factors of the two n=2n = 2 states. The result for the normal state No. 2 (κ=0\kappa= 0, B=0.2084B = 0.2084) is represented in the upper panel of Figure 14. One first observes a much faster decrease than for the n=1n = 1 state No. 1 (upper panel of Figure 11), having comparable binding energies: one order of magnitude at Q2=1Q^2 = 1. One also remarks the appearance of two zeroes in the form factor, which becomes negatives in the range Q2∈[1.5,3.0]Q^2 \in[1.5,3.0]. This is a consequence of the complex structure of the state. Indeed, the different contributions Fνν′F^{\nu\nu'} depending on g20νg20ν′g_{20}^{\nu}g_{20}^{\nu'} are indicated in the lower panel. As one can see, the physical FeF_e results from strong cancellations of terms which have opposite signs and are one order of magnitude larger than the physical value they build. Although these components are of the same order, the contribution due to g200g_{20}^{0} — which determines the binding energy of the state — is far from being dominant.

Figure 14. Upper panel: elastic form factor of the normal state No. 2 of Table 1 (n=2n = 2, κ=0\kappa= 0). The different contributions Fνν′F^{\nu\nu'} depending on g20νg20ν′g_{20}^{\nu}g_{20}^{\nu'} are indicated in the lower panel.

The elastic form factor of the abnormal state No. 4 (κ=2\kappa= 2, B=0.00112B = 0.00112) is displayed in Figure 15. The same remark concerning the faster decrease than the n=1n=1 state No. 3 with comparable binding energy holds. Notice also the non trivial structure – similar to a diffraction pattern – seen at Q2≈0.01Q^2 \approx0.01 and detailed in the lower panel. Such a structure, as well as the zeroes in upper panel of Figure 14, is totally unusual in a two-scalar system interacting by the simple kernel (5) and indicates the complexity of the wave function for any state solution with n>1n>1, be it normal or abnormal. The decomposition of FeF_e in terms of the different components Fνν′F^{\nu\nu'} is similar than for the state No. 2, i.e., strong cancellations occur among opposite sign larger terms.

Figure 15. Elastic form factor of the abnormal state No. 4 of Table 1 (n=2n=2, κ=2\kappa=2). The non-trivial structure at Q2≈0.01Q^2 \approx0.01 is detailed in the lower panel.

The corresponding rms radii, extracted using Eq. (34), are R2=3.8R_2=3.8 fm (state No. 2) and R4=49.0R_4=49.0 fm (state No.4). Just on the basis of their binding energies one should expect twice smaller values. The reason is again the complex structure of the BS amplitude of n>1n>1 states, with a ν>0\nu>0 dominating component (by a factor of 10310^3 in state No. 4) that plays no role in determining the binding energy – and so for the spatial extension – of the system.

As it was the case for the n=1n=1 states, the comparison of the elastic form factors of the n=2n=2 normal and abnormal states (upper panels of Figure 14 and Figure 15) shows that the abnormal state form factors decrease faster than the normal ones as functions of Q2Q^2. At Q2=1Q^2=1 the ratio normal/abnormal is three orders of magnitude. Even after adjusting the coupling constant of state No. 2, to have the same binding energy than the state No. 4, the conclusion remains unchanged.

It is also interesting to examine the asymptotic behaviour, which is supposed to have the same form (39) than for n=1n=1 states. This is done in the two panels of Figure 16. They show again the importance of logarithmmic corrections and the different orders of magnitudes of the asymptotic constant c2c_2 between normal and abnormal sates. Notice the scaling factor introduced in some of the plots to include the comparison in the same frame.

Figure 16. Asymptotic behaviour of the elastic form factors of n=2n=2 states of Table 1 (solid lines): No. 2 (normal) on top and No. 4 (abnormal) at bottom.

The ratios of form factors for the normal states Nos. 1 and 2 and for the abnormal ones 3 and 4 are shown in Figure 17 (upper panel). They indeed tend to constants. The value of this constant is ∼7\sim7. The ratio of form factors for the abnormal states No. 3 and the normal one No. 1 is shown in the lower panel of the same figure. It also tends to a constant. The value of this constant is ∼10−5\sim10^{-5}. Note that the ratio of form factors of different nature (abnormal/normal) is much smaller than normal/normal and abnormal/abnormal, as expected. Surprisingly, the ratios normal/normal and abnormal/abnormal are the same. These asymptotic behaviors of the elastic form factors bring additional arguments in favor of the interpretation of the abnormal states of the W-C model as hybrids.

Ratios of form factors for normal and abnormal states

Figure 17. (Color online) Upper panel: Dotted curve is the ratio of the form factors F(Q2)F(Q^2) for the normal ground state No. 1 and for the normal excited state No. 2. Solid curve is the same for the abnormal states No. 3 and No. 4. Lower panel: The ratio of the form factors F(Q2)F(Q^2) for the abnormal state No. 3 from Table 1 and the normal one No. 1.

Transition form factors Fif(Q2)F_{if}(Q^2)

For the sake of completeness in the study of abnormal solutions of the W-C model we present here the results for the transition form factors. There are four states in Table 1 and, hence, six possible transitions between them. The corresponding transition form factors FF are shown in Figs. 18 and 19. The comparison reveals a hierarchy of the transition form factors. In Figure 18 we can see that the form factor for the transition between two normal states, No. 1 (n=1n = 1, κ=0\kappa= 0) →\to No. 2 (n=2n = 2, κ=0\kappa= 0) (upper panel), dominates, by a factor ∼100\sim100, over the maximal values of the normal →\to abnormal transitions (central and lower panels).

Three-panel transition form factors

Figure 18. Transition form factors from normal state No. 1 (n=1n = 1, κ=0\kappa= 0) to other states listed in Table 1. 1 → 2: No.2 (n=2n = 2, κ=0\kappa= 0, normal) (upper panel); 1 → 3: No. 3 (n=1n = 1, κ=2\kappa= 2, abnormal) (central panel) and 1 → 4: No.4 (n=2n = 2, κ=2\kappa= 2, abnormal) (lower panel).

The transition form factor F(Q2)F(Q^2) between two abnormal states, No. 3 (n=1n = 1, κ=2\kappa= 2) →\to No. 4 (n=2n = 2, \kappa = 2$), is displayed in Figure 19 (upper panel). Its maximal value has the same order of magnitude as the normal-normal one, though it decreases much faster. At last, the form factors for the transitions between the normal and abnormal states (Figure 18 and Figure 19, both central and bottom panels) are approximately 100 times smaller than the abnormal↔\leftrightarrowabnormal form factor. This hierarchy is apparently related to the need of rebuilding the state structure for the normal↔\leftrightarrowabnormal transitions.

Three-panel transition form factors

Figure 19. Same as in Figure 18 between 3 → 4: No. 3(n=1n = 1, κ=2\kappa= 2, abnormal) and No. 4 (n=2n = 2, κ=2\kappa= 2, abnormal) (upper panel); 2 → 3: the state No. 2 (n=2n = 2, κ=0\kappa= 0, normal) and the No. 3 (n=1n = 1, κ=2\kappa= 2, abnormal) (central panel) and 2 → 4: the state No. 2 (n=2n = 2, κ=0\kappa= 0, normal) and No. 4 (n=2n = 2, κ=2\kappa= 2, abnormal) (lower panel).

At first glance, this dominance can be simply due to the very different binding energies: two such states will have a small overlap, without invoking any abnormal character. Indeed, one can hardly separate unambiguously the effect of different structures from the different binding energies (the latter result in different wave functions). That is why we have carried out the complex analysis based on the behavior of the form factor (elastic and inelastic) and on the content of the Fock sectors.

For all the transitions that were considered in this section we have calculated simultaneously the transition form factor G(Q2)G(Q^2). The contraction of the electromagnetic current JJ with the momentum transfer qq results in qJ=Qc2G(Q2)qJ = Q_c^2G(Q^2) [Eq. (32)]. Therefore, the current conservation implies G(Q2)≡0G(Q^2) \equiv0 for any Q2Q^2 [Eq. (30)]. A formal proof of this equality, using the BS equation, is given in Appendix B. Computing a quantity that we know from the first principles (current conservation) that should be identically zero could be in principle considered, at most, as being superfluous. However, as is seen from the derivation given in Appendix B, this property is directly related to the fact that gg is indeed a solution of the BS equation and the form factors, mainly the 3D integrals (C.19) and (C.23), have been accurately computed. It thus constitutes a test for our numerical solutions. Similar tests were successfully carried out in [16] for the form factors corresponding. to the electro-desintegration of a bound system (the transition discrete → continuous spectrum).

The kind of results obtained in computing G(Q2)G(Q^2) is illustrated in Figure 20, in a single example corresponding to the transition between No. 2 (n=2n = 2, κ=0\kappa= 0) → No. 3 (n=1n = 1, κ=2\kappa= 2) states. As in the elastic case, the form factor G(Q2)G(Q^2) results from the sum of two terms: G10(Q2)G^{10}(Q^2) (proportional to g12g10g_1^2g_1^0) and G00(Q2)G^{00}(Q^2) (proportional to g20g10g_2^0g_1^0). They are indicated respectively at upper panel in dashed (G10(Q2)G^{10}(Q^2)) and in dotted (G00(Q2)G^{00}(Q^2)) lines. The sum of them, i.e., the full form factor G(Q2)G(Q^2), is indicated by a thick solid line and is indistinguishable from zero at the scale of the figure. It is in fact ≈10−6\approx10^{-6}. Note however that the sensitivity of G(Q2)G(Q^2) to the accuracy in solving the BS equation, that is, in computing the functions g(z)g(z) and in calculating the 3D integrals (C.23), is very high. The result G(Q2)≡0G(Q^2) \equiv0 is due to delicate cancellations between terms which are several orders of magnitude greater. A small error in these calculations results in non-zero G(Q2)G(Q^2). This is demonstrated in the lower panel of Figure 20. Thus, an error in computing the binding energy, e.g., setting Bi=3.512⋅10−3B_i = 3.512 \cdot10^{-3} instead of Bi=3.51169⋅10−3B_i = 3.51169 \cdot10^{-3} and Bf=0.2084B_f = 0.2084 instead of Bf=0.2084099B_f = 0.2084099, provides G(Q2)≈−0.005G(Q^2) \approx-0.005. This value is comparable with the maximum value of the corresponding form factor F(Q2)F(Q^2) (in central panel of Figure 19). An apparent violation of current conservation would hide in fact a lack of accuracy in the computational procedure. To summarize this section, the comparison of the elastic form factors presented in Figure 11, Figure 14, Figure 15 of the normal states with the abnormal ones, shows that the elastic form factors of the abnormal states vs. Q2Q^2 deincrease much faster than for the normal ones. At Q2∼1Q^2 \sim1, the abnormal form factors are about 10310^3 times smaller than the normal ones, whereas the behaviors of the elastic form factors of the normal states with n=1n = 1 and n=2n = 2 remain very close to each other. These observations confirm that the abnormal states are dominated by the many-body Fock states [6, 7, 8].

Contributions to the $2 \to 3$ transition form factor $G(Q^2)$

Figure 20. Fig. 20 (Color online) Contributions to the 2→32 \to3 transition form factor G(Q2)G(Q^2): the dashed line is G10(Q2)G^{10}(Q^2) and the dotted line G00(Q2)G^{00}(Q^2). The sum of them (solid line) is the full form factor G(Q2)G(Q^2) (upper panel): it is of the order of 10−610^{-6} and results from contributions of several orders of magnitude greater. A small error in computing the binding energy, e.g., Bi=3.512⋅10−3B_i = 3.512 \cdot10^{-3} instead of Bi=3.51169⋅10−3B_i = 3.51169 \cdot10^{-3} and Bf=0.2084B_f = 0.2084 instead of Bf=0.2084099B_f = 0.2084099, would provide G(Q2)≈−0.005G(Q^2) \approx-0.005 (lower panel), a size comparable with F(Q2)F(Q^2) in Figure 19.

The transitions between the normal and abnormal states, in comparison to the normal-normal and abnormal-abnormal transitions, are also suppressed. This suppression indicates that the normal and abnormal states have different structures and the transitions between them require the rebuilding of the states.

The quality of our numerical calculation is quite sufficient to justify the above conclusions. This is demonstrated in Figure 20 for the transition form factor G(Q2)G(Q^2). As is explained in § 4 and proved in Appendix § 7, the electromagnetic current conservation requires G(Q2)=G(Q^2) = 00. This is indeed observed in Figure 20, top, (solid line) as a result of rather delicate cancellations of several contributions. Numerical changes of BB and g(z)g(z), which seem insignificant, may noticeably change the value of G(Q2)G(Q^2) (Figure 20, bottom).

The calculations of the form factors for the transitions to the states given in the Table 2, with the precision used so far, are unstable and require much higher precision. We do not present them here.

No.nnκ\kappaBBN2N_{2}
5141.54091⋅10−51.54091 \cdot10^{-5}6.19⋅10−36.19 \cdot10^{-3}
6244.95065⋅10−64.95065 \cdot10^{-6}2.06⋅10−52.06 \cdot10^{-5}

Table 2. Same as in Table 1, for the abnormal states with κ=4\kappa=4.

Concluding remarks

Our present analysis shows that the abnormal solutions of the W-C model have a different internal structure than the normal ones, which can be traced back to their decomposition properties into Fock space sectors on light-front planes. This constitutes a genuine property of these states and we propose it as an alternative characteristic to the traditional explanation in terms of temporal degrees of freedom excitations. Whereas the normal solutions are dominated by the two-body Fock sector made of the two massive constituents, the abnormal ones are dominated by the Fock sectors made of the two massive constituents and several or many massless exchange particles. This feature is also manifested through the fast decrease of the electromagnetic form factors of the abnormal states, signaling their many-body compositeness. Therefore, the abnormal states do not appear as pathological solutions of the BS equation, but rather as solutions having specifically a relativistic origin, through the dominance, in their internal structure, of the massless exchange particles.

Another particular feature of the abnormal solutions is the relatively large value of the coupling constant needed for their existence (α>π/4\alpha> \pi/4). While the stability condition of the W-C model also requires that α\alpha be bounded by the upper value 2π2\pi, the corresponding window of permissible values does not belong to the domain of perturbation theory and the question of the validity of the ladder approximation can be raised. This question has been examined in Ref. [18] in the light of the incorporation into the model of the renormalization effects. It turns out that the above domain of values of the coupling constant is incompatible with a consistent treatment of such effects. The renormalization constants violate the inequalities which follow from the positivity conditions coming from the spectral functions of the renormalized fields.

The latter result brings us to questioning the effect of the higher-order multiparticle exchange diagrams. This problem has been dealt with in Ref. [19] in a model of QED, where the two massive constituents are static and tied at fixed positions in three-dimensional space. The abnormal solutions corresponding essentially to excitations of degrees of freedom described by the relative time variable (or equivalently, of the relative energy variable), this model should rather preserve their possible existence. It turns out that in this configuration, the two-particle Green’s function is exactly calculable: it does not display any abnormal type of bound state in its structure; only the normal ground-state is present in the spectrum. On the other hand, the BS equation, in the ladder approximation, still continues exhibiting abnormal solutions. A similar conclusion is also obtained from a different approach, based on a three-dimensional reduction of the BS equation with the inclusion of multiparticle exchange diagrams [20].

The above considerations support, as mentioned, another interpretation of the abnormal solutions of the W-C model. It is possible that excitations of the degrees of freedom described by the relative time variable correspond, from the point of view of the Fock decomposition, to filling of the higher Fock sectors. These two interpretations may not contradict each other, but rather be compatible.

In the W-C model, the even-relative-energy abnormal solutions appear as theoretically acceptable states, and are a constitutive part of the corresponding S-matrix. The fact that they are dominated in Fock space by the many-body massless exchange particles may suggest that they are a kind of “hybrid” states. They might represent the Abelian scalar analogs of the hybrids that are searched for in QCD, which are coupled essentially to a pair of quark-antiquark and one or several gluon fields. Here, however, the non-Abelian property of the gauge group, as well as the existence of gluon self-interactions, make the latter states better adapted for experimental, as well as theoretical, investigations. On the other hand, the possible relevance of the multiparticle exchange diagrams in the kernel of the BS equation remains a key ingredient for the ultimate conclusion as to whether the W-C model may have any experimental impact. Note, however, that it is natural to expect that since the multiparticle exchanges add extra exchange particles in the intermediate states, they do not reduce but rather increase the higher Fock components.

In any event, in spite of the fact that the W-C model is an oversimplified model, it nevertheless contains the phenomenon of the particle creation and gives an interesting example of natural generation of hybrid states. Therefore, the hybrid systems can naturally exist in more sophisticated field theories and be detected in appropriate experiments.

The question of a possible existence of abnormal states in the case of massive-particle exchanges is currently under investigation by the present authors.

Acknowledgements

V.A.K. is grateful to the theory group of the Institut de Physique Nucléaire d’Orsay (IPNO) for the kind hospitality and financial support during his visits. H.S. acknowledges support from the EU research and innovation program Horizon 2020, under Grant Agreement No. 824093.

Two-body contribution N2N_2 to the full norm

According to (3), the full normalization is the sum of contributions of all the Fock sectors. If the two-body wave function (ψ2\psi_2 in (2)) is known, then the contribution of the two-body sector in the full norm reads:

N2=1(2π)3∫∣ψ∣2(k⊥,x)d2k⊥ dx2x(1−x).(42)N_2 = \frac{1}{(2\pi)^3}\int\left|\psi\right|^2(\mathbf{k}_{\perp},x)\frac{d^2 k_{\perp}\,dx}{2x(1-x)}. \tag*{(42)}

We recall that we follow the definition of the wave function of light-front dynamics, where the state vector ∣p⟩|p\rangle is defined on a light-front plane. More precisely, we use the explicitly covariant version of light-front dynamics [5], in which the light-front plane is defined by the equation ω⋅x=0\omega\cdot x=0, where ω\omega is a four-vector with the property ω2=0\omega^2=0.

The light-front wave function, which is the coefficient of the two-body contribution ∣2⟩|2\rangle in the Fock decomposition (2), can be extracted from the BS amplitude by projecting it on the light-front plane (see Eq. (3.57) of Ref. [5]):

ψ(k⊥,x)=x(1−x)πNtot∫−∞∞Φ(k+βωω⋅p,p)dβ.(43)\psi(\mathbf{k}_{\perp},x)=\frac{x(1-x)}{\pi\sqrt{N_{\mathrm{tot}}}}\int_{-\infty}^{\infty}\Phi\left(k+\frac{\beta\omega}{\omega\cdot p},p\right)d\beta. \tag*{(43)}

where NtotN_{\mathrm{tot}} is the full normalization factor (the same as in Eqs. (C.19) and (C.23) below) providing the normalization of the elastic form factor F(0)=1F(0)=1.

Substituting here the BS amplitude (24) and making the transformations resulting in Eq. (3.67) of Ref. [5], we obtain, for n=2n=2, an expression similar3 to Eq. (3.67):

ψ(k⊥,x)=−ix(1−x)πNtot∫−∞∞dβ∫−11dz×m3g12(z)[(k⊥2+κ02)(1−zz0)+β(z0−z)−ϵ]3−ix(1−x)πNtot∫−∞∞dβ∫−11dz×m5g22(z)[(k⊥2+κ02)(1−zz0)+β(z0−z)−ϵ]4.(44)\begin{aligned} \psi(\mathbf{k}_{\perp},x)={}&-\frac{ix(1-x)}{\pi\sqrt{N_{\mathrm{tot}}}}\int_{-\infty}^{\infty}d\beta\int_{-1}^{1}dz\\ &\times\frac{m^3g_1^2(z)}{\left[(\mathbf{k}_{\perp}^2+\kappa_0^2)(1-z z_0)+\beta(z_0-z)-\epsilon\right]^3}\\ &-\frac{ix(1-x)}{\pi\sqrt{N_{\mathrm{tot}}}}\int_{-\infty}^{\infty}d\beta\int_{-1}^{1}dz\\ &\times\frac{m^5g_2^2(z)}{\left[(\mathbf{k}_{\perp}^2+\kappa_0^2)(1-z z_0)+\beta(z_0-z)-\epsilon\right]^4}. \tag*{(44)} \end{aligned}

where

z0=1−2x,κ02=m2−14M2,z_0=1-2x,\qquad\kappa_0^2=m^2-\frac{1}{4}M^2,
k⊥2=k⊥2+m24x(1−x)−m2.(45)\mathbf{k}_{\perp}^2=\frac{\mathbf{k}_{\perp}^2+m^2}{4x(1-x)}-m^2. \tag*{(45)}

To calculate these integrals, we use the formulas:

∫−∞∞dβ(βu+v−iϵ)3=iπv2δ(u),∫−∞∞dβ(βu+v−iϵ)4=2iπ3v3δ(u).(46)\int_{-\infty}^{\infty}\frac{d\beta}{(\beta u+v-\mathrm{i}\epsilon)^3}=\frac{\mathrm{i}\pi}{v^2}\delta(u), \qquad \int_{-\infty}^{\infty}\frac{d\beta}{(\beta u+v-\mathrm{i}\epsilon)^4}=\frac{2\mathrm{i}\pi}{3v^3}\delta(u). \tag*{(46)}

In our case, v=(k2+κ02)(1−zz0)v=(\mathbf{k}^2+\kappa_0^2)(1-zz_0), u=z−z0u=z-z_0. In this way, we find

ψ(k⊥,x)=m3x(1−x)g12(1−2x)Ntot[k⊥2+m2−x(1−x)M2]2+2m5x(1−x)g21(1−2x)3Ntot[k⊥2+m2−x(1−x)M2]3.(47)\psi(\mathbf{k}_{\perp},x)=\frac{m^3x(1-x)g_1^2(1-2x)}{\sqrt{N_{\mathrm{tot}}}[\mathbf{k}_{\perp}^2+m^2-x(1-x)M^2]^2} +\frac{2m^5x(1-x)g_2^1(1-2x)}{3\sqrt{N_{\mathrm{tot}}}[\mathbf{k}_{\perp}^2+m^2-x(1-x)M^2]^3}. \tag*{(47)}

Substituting ψ(k⊥,x)\psi(\mathbf{k}_{\perp},x) into the two-body normalization integral (A.1) and integrating, one finds for n=2n=2

N2=13⋅27π2Ntot∫−11dz (1−z2){[g21(z)]2[Q(z)]3+g21(z)g20(z)[Q(z)]4+415[g20(z)]2[Q(z)]5},(48)N_2=\frac{1}{3\cdot2^7\pi^2N_{\mathrm{tot}}}\int_{-1}^{1}dz\,(1-z^2)\left\{\frac{[g_2^1(z)]^2}{[Q(z)]^3}+\frac{g_2^1(z)g_2^0(z)}{[Q(z)]^4}+\frac{4}{15}\frac{[g_2^0(z)]^2}{[Q(z)]^5}\right\}, \tag*{(48)}

with Q(z)Q(z) defined in Eq. (9).

The case n=1n=1, is obtained by keeping in the wave function (A.6) the first term only and replacing in it g21g_2^1 by g10g_1^0. This gives:

N2=1384π2Ntot∫−11(1−z2)[g10(z)]2[Q(z)]3 dz.(49)N_2=\frac{1}{384\pi^2N_{\mathrm{tot}}}\int_{-1}^{1}\frac{(1-z^2)[g_1^0(z)]^2}{[Q(z)]^3}\,dz. \tag*{(49)}

It is interesting to investigate analytically the behavior of N2N_2 in the non-relativistic limit when the binding energy B=2m−MB=2m-M tends to zero. One can show that in this limit at leading order N2(B→0)=1N_2(B\to0)=1 and then find the law of approaching of N2(B)−1N_2(B)-1 to 0. We will demonstrate this in the case n=1n=1 [Eq. (A.8)]. Q(z)Q(z) [Eq. (9)] can be represented as Q(z)≈z2+BmQ(z)\approx z^2+\frac{B}{m}. Then

1[Q(z)]3≈1(z2+ϵ)3≈3π8ϵ5/2δ(z),ϵ=Bm.\frac{1}{[Q(z)]^3}\approx\frac{1}{(z^2+\epsilon)^3}\approx\frac{3\pi}{8\epsilon^{5/2}}\delta(z),\qquad\epsilon=\frac{B}{m}.

We substitute this formula in (A.8) and find the leading term when B→0B\to0:

N2(B→0)=[g10(0)]2m5/2210πNtotB5/2.(50)N_2(B\to0)=\frac{[g_1^0(0)]^2m^{5/2}}{2^{10}\pi N_{\mathrm{tot}}B^{5/2}}. \tag*{(50)}

We then calculate NtotN_{\mathrm{tot}} when B→0B\to0. For n=1n=1, the sum (33) is reduced to one term only, determined by Eq. (C.17) of Appendix C. NtotN_{\mathrm{tot}} is given by Eq. (C.18). Replacing in (C.18) the variable ξ+=12(1+y)\xi_+=\frac{1}{2}(1+y) and keeping the leading term only, we transform the integrand as

m2(5−6ξ+)−2M2ξ+(1−ξ+)[m2−ξ+(1−ξ+)M2]4≈2m6(y2+ϵ)3≈3π4ϵ5/2δ(y).\frac{m^2(5-6\xi_+)-2M^2\xi_+(1-\xi_+)}{[m^2-\xi_+(1-\xi_+)M^2]^4}\approx\frac{2}{m^6(y^2+\epsilon)^3}\approx\frac{3\pi}{4\epsilon^{5/2}}\delta(y).

We omitted the derivative of the delta-function δ′(y)\delta'(y) which enters with a coefficient of the same order as δ(y)\delta(y), however, it does not contribute for symmetric g(z)g(z). Therefore, we get

Ntot=m5/227πB5/2∫−11g(z) dz∫−11g(z′) dz′∫01du u2(1−u)2δ((1+z)u+(1+z′)(1−u)−1).(51)N_{\mathrm{tot}}=\frac{m^{5/2}}{2^7\pi B^{5/2}}\int_{-1}^{1}g(z)\,dz\int_{-1}^{1}g(z')\,dz'\int_0^1du\,u^2(1-u)^2\delta\left((1+z)u+(1+z')(1-u)-1\right). \tag*{(51)}

From the delta-function we find u=−z′/(z−z′)u=-z'/(z-z'). This expression is in the limits 0≤u≤10\leq u\leq1, if −1≤z≤0-1\leq z\leq0, 0≤z′≤10\leq z'\leq1 and 0≤z≤10\leq z\leq1, −1≤z′≤0-1\leq z'\leq0. For symmetric g(z)g(z) these two domains give equal contributions. Therefore, integrating over uu by means of the delta-function, we find for symmetric g(z)g(z):

Ntot(B→0)=3m5/226πB5/2∫01∫01z2z′2g(z)g(z′)(z+z′)5 dz dz′.(52)N_{\mathrm{tot}}(B\to0)=\frac{3m^{5/2}}{2^6\pi B^{5/2}}\int_0^1\int_0^1\frac{z^2z'^2g(z)g(z')}{(z+z')^5}\,dz\,dz'. \tag*{(52)}

When B→0B\to0, the limiting expression of g(z)g(z) is [1, 2]: g(z)≈1−∣z∣g(z)\approx1-|z|. Calculating the integral in (A.11):

∫01∫01z2z′2g(z)g(z′)(z+z′)5 dz dz′=148,\int_0^1\int_0^1\frac{z^2z'^2g(z)g(z')}{(z+z')^5}\,dz\,dz'=\frac{1}{48},

we get

Ntot(B→0)=m5/2210πB5/2.N_{\mathrm{tot}}(B\to0)=\frac{m^{5/2}}{2^{10}\pi B^{5/2}}.

Substituting it in (A.9), we obtain N2(B→0)=1N_2(B\to0)=1, as should be for the normal state in the non-relativistic case.

The next order correction was found in [9], Eq. (24). It reads:

N2−1=2απln⁡(α).(53)N_2-1=\frac{2\alpha}{\pi}\ln(\alpha). \tag*{(53)}

In the given order, we can use the Balmer formula: B=14mα2B=\frac{1}{4}m\alpha^2, that is α=4Bm\alpha=\sqrt{\frac{4B}{m}}. Substituting this α\alpha in (A.12), we find for B≪mB\ll m:

N2−1=1π4Bmln⁡(4Bm).(54)N_2-1=\frac{1}{\pi}\sqrt{\frac{4B}{m}}\ln\left(\frac{4B}{m}\right). \tag*{(54)}

Appendix B: Proof that G(Q2)≡0G(Q^2)\equiv0

The vanishing of the form factor GG for the transition between different states is realized if the initial and final states are associated with the solutions of the BS equation. Let us rewrite the BS equation (4) so that it would contain the functions Φi(12p−k,p)\Phi_i(\frac{1}{2}p-k,p) and Φf(12p′−k,p′)\Phi_f(\frac{1}{2}p'-k,p') entering in the expression (27) for the current. The corresponding equations, obtained from Eq. (4) by the variable shift k→12p−kk \to\frac{1}{2}p-k, are (assuming time-reversal and CP invariances)

[(p−k)2−m2](k2−m2)Φi(12p−k,p)=16iαπm2∫d4k′(2π)4Φi(12p−k′,p)(k−k′)2+iϵ≡K∗Φi,(55)\left[(p-k)^2-m^2\right](k^2-m^2)\Phi_i\left(\frac{1}{2}p-k,p\right) =16i\alpha\pi m^2\int\frac{d^4k'}{(2\pi)^4}\frac{\Phi_i\left(\frac{1}{2}p-k',p\right)}{(k-k')^2+i\epsilon} \equiv K * \Phi_i, \tag*{(55)}
Φ‾f(12p′−k,p′)[(p′−k)2−m2](k2−m2)=16iαπm2∫d4k′(2π)4Φ‾f(12p′−k′,p′)(k−k′)2+iϵ≡Φ‾f∗K,(56)\overline{\Phi}_f\left(\frac{1}{2}p'-k,p'\right)\left[(p'-k)^2-m^2\right](k^2-m^2) =16i\alpha\pi m^2\int\frac{d^4k'}{(2\pi)^4}\frac{\overline{\Phi}_f\left(\frac{1}{2}p'-k',p'\right)}{(k-k')^2+i\epsilon} \equiv\overline{\Phi}_f * K, \tag*{(56)}

where the argument 12p−k\frac{1}{2}p-k is the relative momentum of the particles having the momenta p−kp-k, kk and similarly for 12p′−k\frac{1}{2}p'-k. Also notice that the kernel KK is the same in both equations.

According to Eq. (32):

G(Q2)=q⋅J/Qc2,G(Q^2)=q\cdot J/Q_c^2,

with q=p′−pq=p'-p and JJ determined by Eq. (27). The integrand of q⋅Jq\cdot J [cf. Eq. (27)] contains the scalar product

q⋅(p+p′−2k)=(p′2−p2−2(p′−p)⋅k).q\cdot(p+p'-2k)=\left(p'^2-p^2-2(p'-p)\cdot k\right).

On the other hand:

(p′2−p2−2(p′−p)⋅k)(k2−m2)=[(p′−k)2−m2](k2−m2)−[(p−k)2−m2](k2−m2)={[(p′−k)2−m2](k2−m2)−∗K‾}−{[(p−k)2−m2](k2−m2)−K∗}.\begin{aligned} \left(p'^2-p^2-2(p'-p)\cdot k\right)(k^2-m^2) \\ &=\left[(p'-k)^2-m^2\right](k^2-m^2) -\left[(p-k)^2-m^2\right](k^2-m^2) \\ &=\left\{\left[(p'-k)^2-m^2\right](k^2-m^2)-\overline{*K}\right\} -\left\{\left[(p-k)^2-m^2\right](k^2-m^2)-K^*\right\}. \end{aligned}

After substitution in q⋅Jq\cdot J, the first operator acts on Φ‾f\overline{\Phi}_f and gives zero, and similarly with the second operator acting on Φi\Phi_i. Hence

q⋅J=0⇒G(Q2)=0.q\cdot J=0 \qquad\Rightarrow\qquad G(Q^2)=0.

We emphasize that this result is obtained provided Φi,f\Phi_{i,f} satisfy the BS equation with a given kernel. Numerical uncertainties in Φi,f\Phi_{i,f} result in deviations of G(Q2)G(Q^2) from zero. This is observed in Figure 20 (bottom).

Appendix C: Expressions of the form factors

In general, the elastic and transition form factors are obtained as a sum of n×n′n\times n' terms Fνν′F^{\nu\nu'}, involving the different components of the BS amplitude in the initial (n)(n) and final (n′)(n') states. They have the general form given in Eq. (33):

F(Q2)=∑ν=0n−1∑ν′=0n′−1Fνν′(Q2).(57)F(Q^2)=\sum_{\nu=0}^{n-1}\sum_{\nu'=0}^{n'-1}F^{\nu\nu'}(Q^2). \tag*{(57)}

We present here explicit expressions for calculating the elastic form factors in the cases n=1,2n=1,2 and the transition form factors between n=2n=2 and n′=2n'=2 states.

The elastic form factor, FeF_e, for n=1n=1 states, determined by a single component g≡g01g\equiv g^1_0, is given by

Fe(Q2)=m632π2Ntot∫−11dz g(z)∫−11dz′ g(z′)∫01du u2(1−u)2ND4,(58)F_e(Q^2)=\frac{m^6}{32\pi^2N_{\mathrm{tot}}}\int_{-1}^{1}dz\,g(z)\int_{-1}^{1}dz'\,g(z') \int_0^1du\,u^2(1-u)^2\frac{N}{D^4}, \tag*{(58)}

with the integrands given by

N(z,z′,u,Q2)=A(z,z′,u)+B(z,z′,u)Q2,N(z,z',u,Q^2)=A(z,z',u)+B(z,z',u)Q^2,
D(z,z′,u,Q2)=m2−M2ξ+(1−ξ+)−B(z,z′,u)Q2,D(z,z',u,Q^2)=m^2-M^2\xi_+(1-\xi_+)-B(z,z',u)Q^2,

in terms of functions AA and BB:

ξ±(z,z′,u)=12[(1+z)u±(1+z′)(1−u)],\xi_{\pm}(z,z',u)=\frac{1}{2}\left[(1+z)u\pm(1+z')(1-u)\right],
A(z,z′,u)=m2(5−6ξ+)−2M2ξ+(1−ξ+),A(z,z',u)=m^2(5-6\xi_+)-2M^2\xi_+(1-\xi_+),
B(z,z′,u)=−14(1+z)(1+z′)u(1−u)=−14(ξ+2−ξ−2).\begin{aligned} B(z,z',u)=-\frac{1}{4}(1+z)(1+z')u(1-u) \\ &=-\frac{1}{4}(\xi_+^2-\xi_-^2). \end{aligned}

The total norm NtotN_{\mathrm{tot}}, defined to fulfill F(0)=1F(0)=1, is obtained by

Ntot=m632π2∫−11dz g(z)∫−11dz′ g(z′)∫01du u2(1−u)2m2(5−6ξ+)−2M2ξ+(1−ξ+)[m2−ξ+(1−ξ+)M2]4.(59)N_{\mathrm{tot}}=\frac{m^6}{32\pi^2}\int_{-1}^{1}dz\,g(z)\int_{-1}^{1}dz'\,g(z')\int_0^1du\,u^2(1-u)^2 \frac{m^2(5-6\xi_+)-2M^2\xi_+(1-\xi_+)} {\left[m^2-\xi_+(1-\xi_+)M^2\right]^4}. \tag*{(59)}

For n=2n=2 states, the BS amplitude is determined by two components g20,g21g^0_2,g^1_2 (index κ\kappa is here irrelevant and omitted). The expressions of the different form factors Feνν′F^{\nu\nu'}_e can be written in close analogy with (C.17) and read

Feνν′(Q2)=m632π2Ntot∫−1+1dz g2ν(z)∫−1+1dz′ g2ν′(z′)∫01du u3−ν(1−u)3−ν′(m2D)2−ν−ν′Nνν′D4.(60)F^{\nu\nu'}_e(Q^2)=\frac{m^6}{32\pi^2N_{\mathrm{tot}}}\int_{-1}^{+1}dz\,g^\nu_2(z)\int_{-1}^{+1}dz'\,g^{\nu'}_2(z') \int_0^1du\,u^{3-\nu}(1-u)^{3-\nu'} \left(\frac{m^2}{D}\right)^{2-\nu-\nu'}\frac{N^{\nu\nu'}}{D^4}. \tag*{(60)}

with the integrands expressed in terms of functions Aνν′A_{\nu\nu'} and Bνν′B_{\nu\nu'} as

Nνν′=Aνν′+Bνν′Q2,(61)N^{\nu\nu'}=A_{\nu\nu'}+B_{\nu\nu'}Q^2, \tag*{(61)}

with A11≡AA_{11}\equiv A, B11≡BB_{11}\equiv B and the remaining ones given by

A10=14[4m2(6−7ξ+)−4M2ξ+(2−ξ+−ξ−2)],A_{10}=\frac{1}{4}\left[4m^2(6-7\xi_+)-4M^2\xi_+(2-\xi_+-\xi_-^2)\right],
A00=13[4m2(7−8ξ+)−8Mi2ξ+(1−ξ+2)],A_{00} = \frac{1}{3}\left[4m^2(7-8\xi_+) - 8M_i^2\xi_+(1-\xi_+^2)\right],
B10=−14[(2−ξ+)(ξ+2−ξ−2)],B_{10} = -\frac{1}{4}\left[(2-\xi_+)(\xi_+^2-\xi_-^2)\right],
B00=−13[(3−2ξ+)(ξ+2−ξ−2)],(62)B_{00} = -\frac{1}{3}\left[(3-2\xi_+)(\xi_+^2-\xi_-^2)\right], \tag*{(62)}

and the symmetry relation F10=F01F^{10}=F^{01}. The total norm is given by

Ntot=m632π2∑ν,ν′=01∫−1+1dz g2ν(z)∫−1+1dz′ g2ν′(z′)×∫01du u3−ν(1−u)3−ν′(m2D)2−ν−ν′Nνν′D4∣Q2=0.(63)N_{\mathrm{tot}} = \frac{m^6}{32\pi^2}\sum_{\nu,\nu'=0}^{1}\int_{-1}^{+1} dz\,g_2^\nu(z)\int_{-1}^{+1} dz'\,g_2^{\nu'}(z')\times\int_0^1 du\,u^{3-\nu}(1-u)^{3-\nu'}\left(\frac{m^2}{D}\right)^{2-\nu-\nu'}\left.\frac{N^{\nu\nu'}}{D^4}\right|_{Q^2=0}. \tag*{(63)}

Notice that since A11≡AA_{11}\equiv A, B11≡BB_{11}\equiv B, Fe11(Q2)F_{\mathrm{e}}^{11}(Q^2) from Eq. (C.19) provides the case n=1n=1.

The transition form factor between n=1n=1 and n=2n=2 states offers several possibilities: 1→11\to1, 1→21\to2, 2→12\to1 and 2→22\to2. We give below the expressions of the components Fifνν′F_{if}^{\nu\nu'} of the transition form factors between an initial state n=2n=2 and a final state n=2n=2, determined, respectively, by the components (g20,i,g21,i)(g_2^{0,i},g_2^{1,i}) (g20,f,g21,f)(g_2^{0,f},g_2^{1,f}) and the total norms of the corresponding elastic form factors NtotiN_{\mathrm{tot}}^i, NtotfN_{\mathrm{tot}}^f:

Fifνν′=m632π2NtotiNtotf∫−1+1dz g2ν,i(z)∫−1+1dz′ g2ν′,f(z′)×∫01du u3−ν(1−u)3−ν′(m2D)2−ν−ν′Nifνν′Dif4.(64)F_{if}^{\nu\nu'} = \frac{m^6}{32\pi^2\sqrt{N_{\mathrm{tot}}^iN_{\mathrm{tot}}^f}}\int_{-1}^{+1} dz\,g_2^{\nu,i}(z)\int_{-1}^{+1} dz'\,g_2^{\nu',f}(z')\times\int_0^1 du\,u^{3-\nu}(1-u)^{3-\nu'}\left(\frac{m^2}{D}\right)^{2-\nu-\nu'}\frac{N_{if}^{\nu\nu'}}{D_{if}^4}. \tag*{(64)}

The integrand Nifνν′N_{if}^{\nu\nu'} and DD can be written in the form

Nifνν′(z,z′,u)=Aνν′+Bνν′Q2+Cνν′Qc2,(65)N_{if}^{\nu\nu'}(z,z',u)=A^{\nu\nu'}+B^{\nu\nu'}Q^2+C^{\nu\nu'}Q_c^2, \tag*{(65)}
Dif(z,z′,u)=m2−ξ+(1−ξ+)Mi2−B11Q2+12C11Qc2,(66)D_{if}(z,z',u)=m^2-\xi_+(1-\xi_+)M_i^2-B^{11}Q^2+\frac{1}{2}C^{11}Q_c^2, \tag*{(66)}

involving now Qc2=Mf2−Mi2Q_c^2=M_f^2-M_i^2 and new functions Cνν′C^{\nu\nu'}. The functions AA and BB are the same as for the elastic case with M2≡Mi2M^2\equiv M_i^2 and the functions Cνν′C^{\nu\nu'} are defined as

C11=−(ξ+−ξ−)(1−ξ+),C^{11}=-(\xi_+-\xi_-)(1-\xi_+),
C10=−12(ξ+−ξ−)(2−ξ+−ξ+2),C^{10}=-\frac{1}{2}(\xi_+-\xi_-)(2-\xi_+-\xi_+^2),
C00=−43(ξ+−ξ−)(1−ξ+2).(67)C^{00}=-\frac{4}{3}(\xi_+-\xi_-)(1-\xi_+^2). \tag*{(67)}

The transition form factor G(Q2)G(Q^2) is obtained by an expansion similar to (C.16). The corresponding components Gifνν′G_{if}^{\nu\nu'} are given by (C.23), where the integrands Nifνν′N_{if}^{\nu\nu'} are replaced by

Nˉifνν′(z,z′,u)=Aνν′+Bˉνν′Q2+Cνν′Qc2+Dνν′(Q2Qc2).(68)\bar{N}_{if}^{\nu\nu'}(z,z',u)=A^{\nu\nu'}+\bar{B}^{\nu\nu'}Q^2+C^{\nu\nu'}Q_c^2+D^{\nu\nu'}\left(\frac{Q^2}{Q_c^2}\right). \tag*{(68)}

Their form is similar to that of (C.24), with the same Aνν′A^{\nu\nu'} and Cνν′C^{\nu\nu'}, but with two new coefficients Bˉνν′\bar{B}^{\nu\nu'} (instead of Bνν′B^{\nu\nu'}) and Dνν′D^{\nu\nu'} given by:

Bˉ11=−14(5ξ−2−6ξ+ξ−+ξ+2),\bar{B}^{11}=-\frac{1}{4}(5\xi_-^2-6\xi_+\xi_-+\xi_+^2),
Bˉ10=14(ξ+−ξ−)[ξ−(4+3ξ+)−ξ+(2−ξ+)],\bar{B}^{10}=\frac{1}{4}(\xi_+-\xi_-)[\xi_-(4+3\xi_+)-\xi_+(2-\xi_+)],
Bˉ00=13(ξ+−ξ−)[ξ+(2ξ+−3)+ξ−(3+6ξ+)],\bar{B}^{00}=\frac{1}{3}(\xi_+-\xi_-)[\xi_+(2\xi_+-3)+\xi_-(3+6\xi_+)],
D11=+3ξ−(Mi2ξ+−2m2),D^{11}=+3\xi_-(M_i^2\xi_+-2m^2),
D10=−ξ−4[28m2−4Mi2ξ+(3+ξ+)−(ξ+2−ξ−2)Q2],D^{10}=-\frac{\xi_-}{4}[28m^2-4M_i^2\xi_+(3+\xi_+)-(\xi_+^2-\xi_-^2)Q^2],
D00=+2ξ−3[−16m2+2Mi2ξ+(3+2ξ+)+(ξ+2−ξ−2)Q2].(69)D^{00}=+2\frac{\xi_-}{3}[-16m^2+2M_i^2\xi_+(3+2\xi_+)+(\xi_+^2-\xi_-^2)Q^2]. \tag*{(69)}

A few remarks concerning the results of this section are in order:

– The equality (31) follows from the equality of (C.20) and (C.27) at Q2=0Q^2=0.

– The elastic form factor (C.19) is obtained from the transition form factor (C.23) setting i=fi=f, and hence Ntoti=Ntotf=NtotN_{\mathrm{tot}}^i=N_{\mathrm{tot}}^f=N_{\mathrm{tot}}, Mi=Mf=MM_i=M_f=M and Qc=0Q_c=0.

– The normalization factors NtotiN_{\mathrm{tot}}^i which enter in Eqs. (C.23) are those of the corresponding elastic form factors. The functions gn0g_n^0 can be normalized arbitrarily since any multiplicative factor included in the definition of gg is cancelled by a redefinition of NtotN_{\mathrm{tot}}. The function g21g_2^1 is, according to Eq. (23), determined by the choice for g20g_2^0. We have used, for convenience, the normalization gn0(0)=1g_n^0(0)=1.

– The transitions n=2→n′=1n=2\to n'=1 are obtained by keeping only the terms F00F^{00} and F10F^{10} in (C.23), as well as in the corresponding expression NtotN_{\mathrm{tot}}.

– The transitions n=1→n′=2n=1\to n'=2 are similarly obtained by keeping the terms F00F^{00} and F01F^{01}.

– The transition n=1→n′=1n=1\to n'=1 is given by the single term F00F^{00}.

References

  1. [1]G.C. Wick, Properties of Bethe-Salpeter wave functions, Phys. Rev. 96, 1124 (1954).
  2. [2]R.E. Cutkosky, Solutions of a Bethe-Salpeter equation, Phys. Rev. 96, 1135 (1954).DOI
  3. [3]E.E. Salpeter, H. Bethe, A Relativistic equation for bound-state problems, Phys. Rev. 84, 1232 (1951).DOI
  4. [4]N. Nakanishi, A General survey of the Bethe-Salpeter equation, Prog. Theor. Phys. Suppl. 43, 1 (1969).DOI
  5. [5]J. Carbonell, B. Desplanques, V.A. Karmanov, J.-F. Mathiot, Explicitly covariant light front dynamics and relativistic few-body systems, Phys. Rep. 300, 215 (1998); arXiv:nucl-th/9804029.DOI
  6. [6]V. A. Matveev, R. M. Muradyan, A. N. Tavkhelidze, Automodellism in the large-angle elastic scattering and the structure of hadrons, Lett. Nuovo Cim. 7, 719 (1973).DOI
  7. [7]S. J. Brodsky, G. R. Farrar, Scaling laws at large transverse momentum, Phys. Rev. Lett. 31, 1153 (1973).DOI
  8. [8]A. Radyushkin, Quark counting rules: Old and new approaches, Int. J. Mod. Phys. A 25, 502 (2010); arXiv: 0907.4585.DOI
  9. [9]D. S. Hwang, V. A. Karmanov, Many-body Fock sectors in Wick-Cutkosky model, Nucl. Phys. B 696, 413 (2004); arXiv:hep-th/0405035.DOI
  10. [10]V. A. Karmanov, J. Carbonell, H. Sazdjian, Structure and EM form factors of purely relativistic systems, PoS(LC2009)050; arXiv:2001.00401.
  11. [11]G. Feldman, T. Fulton, J. Townsend, Wick equation, the infinite-momentum frame, and perturbation theory, Phys. Rev. D 7, 1814 (1973).DOI
  12. [12]M. Mangin-Brinet, J. Carbonell, Solutions of the Wick-Cutkosky model in the Light Front Dynamics, Phys. Lett. B 474, 237 (2000)
  13. [13]M. Ciafaloni, P. Menotti, Operator analysis of the Bethe-Salpeter equation, Phys. Rev. 140, B929 (1965).DOI
  14. [14]S. Naito, S-matrix and abnormal solutions of the Bethe-Salpeter equation, Prog. Theor. Phys. 40, 628 (1968).DOI
  15. [15]C. Itzykson, J.-B. Zuber, Quantum field theory (Dover Publications, NY, 1989).
  16. [16]J. Carbonell, V.A. Karmanov, Transition electromagnetic form factor and current conservation in the Bethe-Salpeter approach, Phys. Rev. D 91, 076010 (2015); arXiv:1504.02450.DOI
  17. [18]S. Ahlig, R. Alkofer, (In-)consistencies in the relativistic description of excited states in the Bethe-Salpeter equation, Annals of Phys., 275, 113 (1999); arXiv:hep-th/9810241.DOI
  18. [19]H. Jallouli, H. Sazdjian, There are no abnormal solutions of the Bethe-Salpeter equation in the static model, J. Phys. G 22, 1119 (1996); arXiv:hep-th/9512172.DOI
  19. [20]J. Bijtebier, Bethe-Salpeter equation: 3-D reductions, heavy mass limits and abnormal solutions, Nucl. Phys. A 623, 498 (1997); arXiv:nucl-th/9703028.DOI

Paper details

vlnc:W_vfhzewv2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents