1 sec:Introduction↩︎

Changes in propagators, vertices caused by external fields are significant for coherent color-field models, early heavy-ion physics, and cosmological scenarios involving large classical gauge fields. A classical solution of the gauge field with quantized gluon fluctuations is very helpful.

Quantum field theory in specific backgrounds dates back to Schwinger’s external-field formulation [1], which described how strong classical fields alter perturbation theory. While the Abelian case has been extensively developed, particularly in strong-field QED [2], [3], [4], extending it to Yang-Mills theory is more complex and involved because of gauge self-interactions and the color structure. The background-field method provides a consistent framework in which Green’s functions and effective actions can be constructed while maintaining gauge invariance with respect to the classical configuration [5], [6].

In noncovariant gauges, however, perturbation theory introduces gauge-dependent singularities in the gluon propagator. Their consistent treatment requires a causal prescription; throughout this work, we employ the Mandelstam-Leibbrandt regularization [7], [8]. Recent progress shows that it is possible to do exact or semi-exact analysis in plane-wave and other nontrivial Yang-Mills backgrounds [9]. This allows direct calculation of radiative quantities in analytical background solutions.

Background fields also alter the vacuum’s order parameters. Less attention has been given to the simultaneous treatment of condensates, self-energies, and vertex corrections in analytically tractable non-Abelian plane-wave backgrounds.

In this work, we analyze a fermion propagating in an external plane-wave Yang-Mills field that satisfies the classical equations of motion [10]. Based on the exact fermion propagator found in [11], we calculate the one-loop fermion-gluon vertex, the renormalized fermion self-energy, the associated effective mass shift in the external Yang-Mills gauge field.

A single gauge condition is applied to both the background and the operator part of the external field by choosing the axial gauge for the external Yang-Mills gauge field. The Mandelstam-Leibbrandt prescription is used to handle all singular structures. For the effective mass shift, using the exact fermionic propagator in the external Yang-Mills plane wave enables us to separate the contribution from the field-free part from that caused by the external field. In particular, the effective mass is directly determined by the renormalized self-energy constructed from the exact propagator, and background corrections appear as characteristic oscillatory patterns associated with the plane-wave solution of the Yang-Mills equation.

The paper is organized as follows. Section 2 reviews the gluon propagator in the chosen gauge. Section 3 discusses the fermion-gluon renormalized vertex functions (the one-loop vertex correction). Section 4 derives the fermion’s effective mass from the renormalized self-energy. Section 5 analyzes the fermion condensate. Section 6 presents the discussion and outlook.

Portions of the manuscript were edited with the assistance of an AI-based language model. The author holds full responsibility for the content, analysis, and conclusions provided. The scientific content is exclusively the author’s responsibility.

2 Propagator of the Yang-Mills gauge field↩︎

The external Yang-Mills gauge field \(A_{a}^{\mu }\) satisfies the equations\[\partial _{\mu }F_{a}^{\nu \mu }\left( x\right) -gf_{ab}{}^{c}A_{\mu }^{b}F_{c}^{\nu \mu }\left( x\right) =0, \label{YMeq}\tag{1}\] \[F_{a}^{\nu \mu }\left( x\right) =\partial ^{\nu }A_{a}^{\mu }-\partial ^{\mu }A_{a}^{\nu }\left( x\right) -gf_{a}{}^{bc}A_{b}^{\nu }\left( x\right) A_{c}^{\mu }\left( x\right) , \label{Fmunudef}\tag{2}\] where \(x^{\mu }=\left( x^{0},\vec{x}\right)\), \(\partial _{\mu }=\left( \partial /\partial t,\nabla \right)\), Roman letters are used to number the basis in the space of the \(SU(N)\) fundamental representation group; \(a,b,c=1,\cdots ,N^{2}-1\), diag\(\left( G^{\mu \nu }\right) =\left( 1,-1,-1,-1\right)\), \(G^{\mu \nu }\) is the metric tensor. For equations (1 ) and (2 ), the solution can be expressed as a plane wave moving along the light cone [10].\[A_{+}^{a}\left( x\right) =f^{a}\left( x^{+}\right) x^{1}+g^{a}\left( x^{+}\right) x^{2},\;A_{-}^{a}=A_{1}^{a}=A_{2}^{a}=0, \label{Acoleman}\tag{3}\] \[A_{\mu }^{a}\left( x\right) =A_{\mu }^{a}\left( qx\right) ,\;q_{\mu }q^{\mu }=0, \label{Amuq}\tag{4}\] The axial gauge is proper in this case [10].\[\partial ^{\mu }A_{\mu }^{a}\left( x\right) =q^{\mu }\dot{A}_{\mu }^{a}\left( x\right) =\partial ^{+}A_{+}^{a}\left( x\right) =0\;\Rightarrow \;q^{\mu }A_{\mu }^{a}\left( x\right) =0. \label{gaugefixing}\tag{5}\] Let us split the external Yang-Mills gauge field into the background field \(\mathcal{A}_{\mu }^{a}\left( x\right)\)and the operator part \(\mathsf{A}_{\mu }^{a}\left( x\right)\)as \(A_{\mu }^{a}\left( x\right) =\mathcal{A}_{\mu }^{a}\left( x\right) +\mathsf{A}_{\mu }^{a}\left( x\right)\). The operator part describes the quantum fluctuations. In the case of the plane wave and the axial gauge, we can write the following decomposition: \[\mathsf{A}_{a}^{\mu }\left( x\right) =\int \frac{d^{3}q}{\left( 2\pi \right) ^{3}}\frac{1}{\sqrt{2E_{q}}}\sum_{\lambda =1}^{2}\left[ \epsilon _{a}^{\mu }\left( \vec{q},\lambda \right) \hat{c}_{\lambda a}\left( \vec{q}\right) \exp \left( -iqx\right) +\epsilon _{a}^{\ast \mu }\left( \vec{q},\lambda \right) \hat{c}_{\lambda a}^{\dagger }\left( \vec{q}\right) \exp \left( iqx\right) \right] , \label{decomposition}\tag{6}\] where \(\hat{c}_{\lambda a}\left( \vec{q}\right)\) and \(\hat{c}_{\lambda a}^{\dagger }\left( \vec{q}\right)\) are the operators of cancellation and creation, respectively. \(\epsilon _{a}^{\mu }\left( \vec{q},\lambda \right)\) describes the polarization of the Yang-Mills gauge field. We have the following conditions for them:\[\begin{align} \left[ \hat{c}_{\lambda a}^{\dagger }\left( \vec{q}\right) ,\hat{c}_{\lambda ^{\prime }b}\left( \vec{q}^{\prime }\right) \right] =&-\delta _{ab}\eta _{\lambda \lambda ^{\prime }}\left( 2\pi \right) ^{3}\delta ^{\left( 3\right) }\left( \vec{q}-\vec{q}^{\prime }\right) , \notag \\ \epsilon _{a}^{\ast \mu }\left( \vec{q},\lambda \right) \epsilon _{\mu a}\left( \vec{q},\lambda ^{\prime }\right) =&\eta _{\lambda \lambda ^{\prime }},\;\epsilon _{a}^{\mu }\left( \vec{q},\lambda \right) q_{\mu }=0, \notag \\ \langle 0|\hat{c}_{\lambda a}\left( \vec{q}\right) \hat{c}_{\lambda a}^{\dagger }\left( \vec{q}\right) |0\rangle =&1, \label{opercond} \end{align}\tag{7}\] In this gauge, we have that only 2 physical polarizations propagate, and\[\sum_{\lambda =1}^{2}\epsilon _{\nu }^{a}\left( -\vec{q},\lambda \right) \epsilon _{\mu }^{\ast b}\left( -\vec{q},\lambda \right) =\delta ^{ab}\left[ -g_{\mu \nu }+\frac{q^{\mu }n^{\nu }+q^{\nu }n^{\mu }}{q\cdot n}+\frac{n^{2}}{\left( q\cdot n\right) ^{2}}q^{\mu }q^{\nu }\right] . \label{polarsum}\tag{8}\] where \(n^{\mu }\) is a fixed unit 4-vector.

The gluon propagator in axial gauges is\[D_{\mu \nu }^{ab}\left( x,x^{\prime }\right) =i\left( \theta \left( x^{0}-x^{0\prime }\right) \langle 0|\mathsf{A}_{\mu }^{a}\left( x\right) \mathsf{A}_{\nu }^{b}\left( x^{\prime }\right) |0\rangle +i\theta \left( x^{0\prime }-x^{0}\right) \langle 0|\mathsf{A}_{\nu }^{a}\left( x^{\prime }\right) \mathsf{A}_{\mu }^{b}\left( x\right) |0\rangle \right) . \label{gluonprop}\tag{9}\] Substituting (6 ) and (8 ) into the (9 ) and utilizing (7 ), we derive the following expression for the gluon propagator (see, for instance, [12], [13])\[D_{\mu \nu }^{ab}\left( x,x^{\prime }\right) =-\int_{C}\frac{d^{4}q}{\left( 2\pi \right) ^{4}}\frac{e^{-q\left( x-x^{\prime }\right) }}{q^{2}+i\epsilon }\delta ^{ab}\left[ -g_{\mu \nu }+\frac{q^{\mu }n^{\nu }+q^{\nu }n^{\mu }}{q\cdot n}+\frac{n^{2}}{\left( q\cdot n\right) ^{2}}q^{\mu }q^{\nu }\right] , \label{glprop1}\tag{10}\] where the contour \(C\) is the standard contour of integration. Through this decision, gauge redundancy is removed without the typical introduction of Fadeev-Popov ghosts (see, for instance, [14]). Using the additional vector light-like \(n^{2}=0\), we get the expression:\[\tilde{D}_{\mu \nu }^{ab}\left( x,x^{\prime }\right) =-iD_{\mu \nu }^{ab}\left( x,x^{\prime }\right) =\int_{C}\frac{d^{4}q}{\left( 2\pi \right) ^{4}}\frac{i\delta ^{ab}}{q^{2}+i\epsilon }\left[ -g_{\mu \nu }+\frac{q^{\mu }n^{\nu }+q^{\nu }n^{\mu }}{q\cdot n}\right] e^{-q\left( x-x^{\prime }\right) }. \label{glprop2}\tag{11}\]

According to a prescription independently suggested by Mandelstam [7] and Leibbrandt [8] (ML)\[\frac{1}{q\cdot n}\rightarrow \frac{1}{\left[ qn\right] }=\lim_{\epsilon _{3}\rightarrow 0}\frac{n^{\ast }q}{\left( n^{\ast }q\right) \left( q\cdot n\right) +i\epsilon _{3}} \label{ML}\tag{12}\] where \(n^{\ast }\) is the vector conjugate to the vector \(n\), we have the following expression for \(D_{\mu \nu }^{ab}\left( x,x^{\prime }\right)\) and \(\tilde{D}_{\mu \nu }^{ab}\left( x,x^{\prime }\right)\) \[\tilde{D}_{\mu \nu }^{ab}\left( x,x^{\prime }\right) =-iD_{\mu \nu }^{ab}\left( x,x^{\prime }\right) =\int_{C}\frac{d^{4}q}{\left( 2\pi \right) ^{4}}\frac{i\delta ^{ab}}{q^{2}+i\epsilon }\left[ -g_{\mu \nu }+\frac{\left( q^{\mu }n^{\nu }+q^{\nu }n^{\mu }\right) n^{\ast }q}{\left( n^{\ast }q\right) \left( q\cdot n\right) +i\epsilon _{3}}\right] e^{-q\left( x-x^{\prime }\right) }. \label{gluonpropML}\tag{13}\]

3 Renormalized vertex functions↩︎

The one-loop vertex correction in momentum space is\[\begin{align} v^{\dagger }\bar{u}\left( p^{\prime }\right) \Gamma ^{c\mu }\left( p,k\right) u\left( p\right) v=& v^{\dagger }\bar{u}\left( p^{\prime }\right) \left( -ig\gamma ^{\alpha }T^{a}\right) \tilde{G}_{F}\left( q+k\right) \gamma ^{\mu }T^{c}\tilde{G}_{F}\left( q\right) \notag \\ & \times \left( -ig\gamma ^{\beta }T^{b}\right) u\left( p\right) v\tilde{D}_{\alpha \beta }^{ab}\left( p-q\right) , \label{vertexmomspace} \end{align}\tag{14}\] where \(p^{\prime }=p+k\), \(u\left( p\right)\) and \(v\) are spinors (indices omitted), which are elements of the spaces of the appropriate representations [15], and for \(u_{\sigma }\left( p\right)\), we use the following normalization: \[\bar{u}_{\sigma }\left( p\right) u_{\lambda }\left( p^{\prime }\right) =\pm 2m\delta _{\sigma \lambda }\delta _{pp^{\prime }},\;p^{2}=m^{2}, \label{fermnorm}\tag{15}\] where the Dirac scalar production of the spinors \(u_{\sigma }\left( p\right)\) and \(u_{\sigma }\left( -p\right)\), respectively, is represented by the plus and minus signs. The spinor \(v_{\alpha }\) is normalized by condition \(v_{\alpha }^{\dagger }v_{\beta }=\delta _{\alpha \beta }\).

For \(\tilde{G}_{F}\left( x,y\right)\), we used results from [11]. \[\tilde{G}_{F}\left( p\right) =\frac{i\left( \cancel{p}+m\right) U\left( p\right) }{p^{2}-m^{2}+i\epsilon }, \label{Fermionpropmomspace}\tag{16}\] \[\begin{align} U\left( p\right) &=&U\left( p,\varphi ,\varphi ^{\prime }\right) =\cos \left( \theta \left( p,\varphi \right) \right) \cos \left( \theta \left( p,\varphi ^{\prime }\right) \right) \notag \\ &&\times \left\{ 1+\frac{\tan \left( \theta \left( p,\varphi ^{\prime }\right) \right) }{\theta \left( p,\varphi ^{\prime }\right) }\frac{g\left( \left( \gamma ^{\sigma }\right) ^{\dagger }\mathcal{A}_{\sigma }^{e}\left( \varphi ^{\prime }\right) \right) \left( \left( \gamma ^{\rho }\right) ^{\dagger }k_{\rho }\right) }{2\left( pk\right) }T_{e}\right. \notag \\ &&\left. +\frac{g\left( \gamma ^{\nu }k_{\nu }\right) \left( \gamma ^{\lambda }\mathcal{A}_{\lambda }^{b}\left( \varphi \right) \right) }{2\left( pk\right) }\frac{\tan \left( \theta \left( p,\varphi \right) \right) }{\theta \left( p,\varphi \right) }T_{b}+\frac{g\left( \gamma ^{\nu }k_{\nu }\right) \left( \gamma ^{\lambda }\mathcal{A}_{\lambda }^{b}\left( \varphi \right) \right) }{2\left( pk\right) }\frac{\tan \left( \theta \left( p,\varphi \right) \right) }{\theta \left( p,\varphi \right) }\right. \notag \\ &&\left. \times \frac{\tan \left( \theta \left( p,\varphi ^{\prime }\right) \right) }{\theta \left( p,\varphi ^{\prime }\right) }\frac{g\left( \left( \gamma ^{\sigma }\right) ^{\dagger }\mathcal{A}_{\sigma }^{e}\left( \varphi ^{\prime }\right) \right) \left( \left( \gamma ^{\rho }\right) ^{\dagger }k_{\rho }\right) }{2\left( pk\right) }T_{b}T_{e}\right\} , \label{Udef} \end{align}\tag{17}\] \[\theta \left( p\right) =\theta \left( p,\varphi \right) =\frac{g}{\left( pk\right) }\sqrt{\frac{1}{2N}}\left( \int_{0}^{\varphi }d\varphi ^{\prime \prime }\left( \mathcal{A}_{\mu }^{a}\left( \varphi ^{\prime \prime }\right) p^{\mu }\right) \int_{0}^{\varphi }d\varphi ^{\prime \prime \prime }\left( \mathcal{A}_{a}^{\mu }\left( \varphi ^{\prime \prime \prime }\right) p_{\mu }\right) \right) ^{\frac{1}{2}}, \label{thetadef}\tag{18}\]

\(\varphi =kx,\;k^{\mu }\mathcal{A}_{\mu }^{a}\left( x\right) =0\). According to (13 ), the gluon propagator given in momentum space is\[\tilde{D}_{\mu \nu }^{ab}\left( q\right) =\frac{i\delta ^{ab}}{q^{2}+i\epsilon _{3}}\left[ -g_{\alpha \beta }+\frac{\left( q_{\alpha }n_{\beta }+q_{\beta }n_{\alpha }\right) \left( n^{\ast }\cdot q\right) }{\left( n^{\ast }\cdot q\right) \left( q\cdot n\right) +i\varepsilon }\right] . \label{gluonmomspace}\tag{19}\] Substituting (16 ) and (13 ) into (14 ), we obtain\[\begin{align} \Gamma ^{c\mu }\left( p,k\right) &=&\int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}v^{\dagger }\bar{u}\left( p^{\prime }\right) \left( -ig\gamma ^{\alpha }T^{a}\right) \frac{i\left( \cancel{P}+m\right) U\left( P\right) }{P^{2}-m^{2}+i\varepsilon }\gamma ^{\mu }T^{c}\frac{i\left( \cancel{r}+m\right) U\left( r\right) }{r^{2}-m^{2}+i\varepsilon } \notag \\ &&\times \left( -ig\gamma ^{\beta }T^{b}\right) u\left( p\right) v\frac{i\delta ^{ab}}{q^{2}+i\varepsilon }\left( -g_{\alpha \beta }+\frac{\left( q_{\alpha }n_{\beta }+q_{\beta }n_{\alpha }\right) q\cdot n^{\ast }}{q\cdot n^{\ast }q\cdot n+i\varepsilon }\right) . \label{vertexmomentspaceexact} \end{align}\tag{20}\] where \(p^{\prime }=p+k\) and loop momentum \(r\), and we have denoted \(P\equiv r+k\) and gluon momentum \(q\equiv p-r\).

It is clear that as the limit \(\mathcal{A\rightarrow }0\), \(U\rightarrow 1\), and we substitute \(T^{a}=1\), \(g=e\), and \(v=1\), which comes from formula (16 ),we obtain\[\begin{align} \Gamma _{El}^{\mu }\left( p,k\right) &=&e^{2}\bar{u}\left( p^{\prime }\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}\gamma ^{\alpha }\frac{\left( \cancel{P}+m\right) }{P^{2}-m^{2}+i\varepsilon }\gamma ^{\mu }\frac{\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\epsilon }\gamma ^{\beta } \notag \\ &&\times \frac{1}{q^{2}+i\varepsilon }\left( -g_{\alpha \beta }+\frac{\left( q_{\alpha }n_{\beta }+q_{\beta }n_{\alpha }\right) \left( q\cdot n^{\ast }\right) }{q\cdot n^{\ast }q\cdot n+i\epsilon }\right) u\left( p\right) , \label{vertexEl} \end{align}\tag{21}\] This matches the one-loop vertex correction in momentum space of electromagnetic interaction.

We can write the weak-field expansion as \(U=1+\mathcal{O}\left( g\mathcal{A}\right) +\mathcal{O}\left( \left( g\mathcal{A}\right) ^{2}\right)\), and substitute it into the expression for the one-loop vertex correction in momentum space. Thus,\[\begin{align} U\left( p,\varphi ,\varphi ^{\prime }\right) &=&\cos \left( \theta \left( p,\varphi \right) \right) \cos \left( \theta \left( p,\varphi ^{\prime }\right) \right) \left\{ 1+\left( \cdots \right) _{\mathcal{O}\left( g\mathcal{A}\right) }+\left( \cdots \right) _{\mathcal{O}\left( \left( g\mathcal{A}\right) ^{2}\right) }\right\} , \notag \\ \cos \theta &=&1-\frac{\theta ^{2}}{2}+\mathcal{O}\left( \theta ^{4}\right) ,\;\frac{\tan \theta }{\theta }=1+\frac{\theta ^{2}}{3}+\mathcal{O}\left( \theta ^{4}\right) , \label{expanUtet} \end{align}\tag{22}\] thus, we can set \(\left( \frac{\tan \theta }{\theta }\right) \rightarrow 1\) at the linear order. We denote\[\begin{align} X\left( l;\varphi \right) &=&\frac{g}{2\left( lk\right) }\left( \gamma ^{\nu }k_{\nu }\right) \left( \gamma ^{\lambda }\mathcal{A}_{\lambda }^{b}\left( \varphi \right) \right) T_{b}, \notag \\ Y\left( l;\varphi ^{\prime }\right) &=&\frac{g}{2\left( lk\right) }\left( \left( \gamma ^{\sigma }\right) ^{\dagger }\mathcal{A}_{\sigma }^{e}\left( \varphi ^{\prime }\right) \right) \left( \left( \gamma ^{\rho }\right) ^{\dagger }k_{\rho }\right) T_{e}, \notag \\ XY\left( l;\varphi ,\varphi ^{\prime }\right) &=&\left( \frac{g}{2\left( lk\right) }\right) ^{2}\left( \gamma ^{\nu }k_{\nu }\right) \left( \gamma ^{\lambda }\mathcal{A}_{\lambda }^{b}\left( \varphi \right) \right) \left( \left( \gamma ^{\sigma }\right) ^{\dagger }\mathcal{A}_{\sigma }^{e}\left( \varphi ^{\prime }\right) \right) \left( \left( \gamma ^{\rho }\right) ^{\dagger }k_{\rho }\right) T_{b}T_{e}, \label{XYdef} \end{align}\tag{23}\] so that \(X\), \(Y=O\left( g\mathcal{A}\right)\). The scalar prefactor correlation from \(\cos \left( \theta \left( p,\varphi \right) \right) \cos \left( \theta \left( p,\varphi ^{\prime }\right) \right)\) at \(\mathcal{O}\left( \left( g\mathcal{A}\right) ^{2}\right)\) is color-singlet\[Z\left( l;\varphi ,\varphi ^{\prime }\right) \equiv -\frac{1}{2}\left( \theta \left( l,\varphi \right) ^{2}+\theta \left( l,\varphi ^{\prime }\right) ^{2}\right) \mathbf{1.} \label{Zldefinition}\tag{24}\]

Then, up to quadratic order, a compact and correct expansion is \[U\left( l;\varphi ,\varphi ^{\prime }\right) =\mathbf{1}+X\left( l;\varphi \right) +Y\left( l;\varphi ^{\prime }\right) +X\left( l;\varphi \right) Y\left( l;\varphi ^{\prime }\right) +Z\left( l;\varphi ,\varphi ^{\prime }\right) +O\left( \left( g\mathcal{A}\right) ^{3}\right) . \label{expanU}\tag{25}\] For \(\Gamma ^{c\mu }\), we write \(\Gamma ^{c\mu }=\Gamma _{\left( 0\right) }^{c\mu }+\Gamma _{\left( 1\right) }^{c\mu }+\Gamma _{\left( 2\right) }^{c\mu }+O\left( \left( g\mathcal{A}\right) ^{3}\right)\), where the subscripts denote the order in the background amplitude.

For the part corresponding to the zero external field (\(U\rightarrow 1\)), we have\[\begin{align} \Gamma _{\left( 0\right) }^{c\mu } &=&\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}v^{\dagger }\bar{u}\left( p^{\prime }\right) \gamma ^{\alpha }T^{a}\frac{i\left( \cancel{P}+m\right) }{P^{2}-m^{2}+i\epsilon }\gamma ^{\mu }T^{c}\frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\epsilon } \notag \\ &&\times \gamma ^{\beta }T^{a}u\left( p\right) v\frac{i}{q^{2}+i\epsilon }H_{\alpha \beta }\left( q\right) , \label{Gammacmyuzro} \end{align}\tag{26}\] where \[H_{\alpha \beta }\left( q\right) =\left( -g_{\alpha \beta }+\frac{\left( q_{\alpha }n_{\beta }+q_{\beta }n_{\alpha }\right) q\cdot n^{\ast }}{q\cdot n^{\ast }q\cdot n+i\epsilon }\right) . \label{Halefbetta}\tag{27}\] Only one of the two \(U\)’s is expanded to first order for the linear background component \(O\left( g\mathcal{A}\right)\), so we have:\[\begin{align} \Gamma _{\left( 1\right) }^{c\mu } &=&\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}v^{\dagger }\bar{u}\left( p^{\prime }\right) \gamma ^{\alpha }T^{a}\frac{i\left( \cancel{P}+m\right) }{P^{2}-m^{2}+i\epsilon } \notag \\ &&\left\{ X\left( P,\varphi \right) \gamma ^{\mu }T^{c}\frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\varepsilon }+Y\left( P,\varphi ^{\prime }\right) \gamma ^{\mu }T^{c}\frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\epsilon }\right. \notag \\ &&\left. +\gamma ^{\mu }T^{c}\frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\varepsilon }X\left( r,\varphi \right) +\gamma ^{\mu }T^{c}\frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\epsilon }Y\left( r,\varphi ^{\prime }\right) \right\} \notag \\ &&\times \gamma ^{\beta }T^{b}u\left( p\right) v\frac{i\delta ^{ab}}{q^{2}+i\epsilon }H_{\alpha \beta }\left( q\right) . \label{Gammacmu1} \end{align}\tag{28}\]

The quadratic term \(\mathcal{O}\left( \left( g\mathcal{A}\right) ^{2}\right)\) has three structural sources: a second-order insertion on \(P\) only: \(\left[ XY+Z\right] \left( P\right)\), a second-order insertion on r only: \(\left[ XY+Z\right] \left( r\right)\), and a first-order insertion on each line: \(\left[ X+Y\right] \left( P\right) \left[ X+Y\right] \left( r\right)\). Second-order terms are\[\begin{align} \Gamma _{\left( 2\right) ,P;XY}^{c\mu } &=&\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}v^{\dagger }\bar{u}\left( p^{\prime }\right) \gamma ^{\alpha }T^{a}\frac{i\left( \cancel{P}+m\right) }{P^{2}-m^{2}+i\epsilon }\left[ XY\left( P;\varphi ,\varphi ^{\prime }\right) \right] \notag \\ &&\times \gamma ^{\mu }T^{c}\frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\varepsilon }\gamma ^{\beta }T^{b}u\left( p\right) v\frac{i\delta ^{ab}}{q^{2}+i\epsilon }H_{\alpha \beta }\left( q\right) , \label{Gam2PXY} \end{align}\tag{29}\] \[\begin{align} \Gamma _{\left( 2\right) ,P;Z}^{c\mu } &=&\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}v^{\dagger }\bar{u}\left( p^{\prime }\right) \gamma ^{\alpha }T^{a}\frac{i\left( \cancel{P}+m\right) }{P^{2}-m^{2}+i\epsilon }\left[ Z\left( P;\varphi ,\varphi ^{\prime }\right) \right] \notag \\ &&\times \gamma ^{\mu }T^{c}\frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\varepsilon }\gamma ^{\beta }T^{b}u\left( p\right) v\frac{i\delta ^{ab}}{q^{2}+i\epsilon }H_{\alpha \beta }\left( q\right) , \label{Gam2PZ} \end{align}\tag{30}\] \[\begin{align} \Gamma _{\left( 2\right) ,r;XY}^{c\mu } &=&\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}v^{\dagger }\bar{u}\left( p^{\prime }\right) \gamma ^{\alpha }T^{a}\frac{i\left( \cancel{P}+m\right) }{P^{2}-m^{2}+i\epsilon }\gamma ^{\mu }T^{c} \notag \\ &&\times \frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\varepsilon }\left[ XY\left( r;\varphi ,\varphi ^{\prime }\right) \right] \gamma ^{\beta }T^{b}u\left( p\right) v\frac{i\delta ^{ab}}{q^{2}+i\epsilon }H_{\alpha \beta }\left( q\right) , \label{Gam2rXY} \end{align}\tag{31}\] \[\begin{align} \Gamma _{\left( 2\right) ,r;Z}^{c\mu } &=&\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}v^{\dagger }\bar{u}\left( p^{\prime }\right) \gamma ^{\alpha }T^{a}\frac{i\left( \cancel{P}+m\right) }{P^{2}-m^{2}+i\epsilon }\gamma ^{\mu }T^{c} \notag \\ &&\times \frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\varepsilon }\left[ Z\left( P;\varphi ,\varphi ^{\prime }\right) \right] \gamma ^{\beta }T^{b}u\left( p\right) v\frac{i\delta ^{ab}}{q^{2}+i\epsilon }H_{\alpha \beta }\left( q\right) , \label{Gam2rZ} \end{align}\tag{32}\] Now we expand \(\left( X+Y\right) \left( P\right) \left( X+Y\right) \left( r\right)\), resulting in four cross-terms. For example:\[\begin{align} \Gamma _{\left( 2\right) ,P,X;r,X}^{c\mu } &=&\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}v^{\dagger }\bar{u}\left( p^{\prime }\right) \gamma ^{\alpha }T^{a}\frac{i\left( \cancel{P}+m\right) }{P^{2}-m^{2}+i\epsilon }X\left( P,\varphi \right) \notag \\ &&\times \gamma ^{\mu }T^{c}\frac{i\left( \cancel{r}+m\right) }{r^{2}-m^{2}+i\varepsilon }X\left( r,\varphi \right) \gamma ^{\beta }T^{b}u\left( p\right) v\frac{i\delta ^{ab}}{q^{2}+i\epsilon }H_{\alpha \beta }\left( q\right) \label{G2X43YPX43Yr} \end{align}\tag{33}\] For \(\Gamma _{\left( 2\right) ,P,X;r,Y}^{c\mu }\), \(\Gamma _{\left( 2\right) ,P,Y;r,X}^{c\mu }\), \(\Gamma _{\left( 2\right) ,P,Y;r,Y}^{c\mu }\), we can write similarly.

Now, we analyze the periodic plane-wave background: \(\mathcal{A}_{\mu }^{a}\left( \varphi \right) =\varepsilon _{\mu }^{a}\cos \left( \varphi \right)\), \(\varphi =\kappa \cdot x\), \(\kappa ^{2}=0\), \(\kappa \cdot \varepsilon ^{a}=0\). To prevent collisions, we use \(k\) to represent the emitted gluon momentum and \(\kappa\) to denote the background wavevector. Every background-dependent object built from \(\mathcal{A}\left( \varphi \right)\) admits a Fourier series in \(\varphi\) (and \(\varphi ^{\prime }\)) due to the periodicity of \(\mathcal{A}\left( \varphi \right)\). In particular, the standard Floquet form can be used to describe the exact dressed fermion factors for a monochromatic plane wave\[U\left( l;\varphi ,\varphi ^{\prime }\right) =\sum_{n=-\infty }^{\infty }U_{n}\left( l\right) e^{-in\left( \varphi -\varphi ^{\prime }\right) }, \label{UexpandFloquet}\tag{34}\] with coefficients \(U_{n}\left( l\right)\) expressed in terms of Bessel functions \(J_{n}\) of an “intensity parameter”(the non-Abelian generalization of the Volkov parameter). At weak fields, this reduces to a finite number of harmonics at each order: \(O\left( \varepsilon \right)\): only \(n=\pm 1\), \(O\left( \varepsilon ^{2}\right)\): only \(n=0,\pm 2\), etc. When such a Fourier expansion is substituted into the coordinate-space amplitude, and a Fourier transform to momentum space is performed, the phase factor \(e^{-i\kappa \cdot x}\) results in shifted momentum-conserving delta functions. We can write,\[\int d^{4}xe^{i\left( p^{\prime }-p-k\right) \cdot x}e^{-in\kappa \cdot x}=\left( 2\pi \right) ^{4}\delta ^{\left( 4\right) }\left( p^{\prime }-p-k-n\kappa \right) . \label{delta4exter}\tag{35}\] Therefore, the precise selection rule is \(\left( 2\pi \right) ^{4}\delta ^{\left( 4\right) }\left( p^{\prime }-p-k-n\kappa \right)\) for each harmonic \(n\). Momentum conservation is approximately \(p^{\prime }\approx p+k+n\kappa\).

We denote\[\alpha \left( l\right) \equiv \frac{g}{l\cdot \kappa }\sqrt{\frac{\left( l\cdot \varepsilon ^{a}\right) \left( l\cdot \varepsilon _{a}\right) }{2N}}. \label{alfal}\tag{36}\] We are using the two harmonic weights: \(f_{s}\left( \alpha \left( l\right) \right)\)= even-harmonic weight from \(\cos \left( \alpha _{l}\sin \left( \varphi \right) \right)\), \(c_{s}\left( \alpha _{l}\right)\)= odd-harmonic weight from \(\cos \left( \varphi \right) \cos \left( \alpha \left( l\right) \sin \left( \varphi \right) \right)\). We define the selectors \(\mathcal{E}_{s}=\frac{1+\left( -1\right) ^{s}}{2}\) (1 if \(s\) even), \(\mathcal{O}_{s}=\frac{1-\left( -1\right) ^{s}}{2}\) (1 if \(s\) odd). We have the Fourier expansion for \(\cos \left( \alpha \left( l\right) \sin \left( \varphi \right) \right)\): \(\cos \left( \alpha \left( l\right) \sin \left( \varphi \right) \right) =\sum_{s\in Z}f_{s}\left( \alpha \left( l\right) \right) e^{is\varphi }\), with explicit parity selection \(f_{s}\left( \alpha \left( l\right) \right) =\mathcal{E}_{s}J_{s}\left( \alpha \left( l\right) \right)\), where \(J_{s}\) is the Bessel function. So \(f_{s}=0\) for odd \(s\).

By definition, \(\cos \left( \varphi \right) \cos \left( \alpha \left( l\right) \sin \left( \varphi \right) \right) =\sum_{s\in Z}c_{s}\left( \alpha \left( l\right) \right) e^{is\varphi }\), and we get the the shift identity\[c_{s}\left( \alpha \left( l\right) \right) =\frac{1}{2}\left( f_{s-1}\left( \alpha \left( l\right) \right) +f_{s+1}\left( \alpha \left( l\right) \right) \right) =\frac{1}{2}\left( \mathcal{E}_{s-1}J_{s-1}\left( \alpha \left( l\right) \right) +\mathcal{E}_{s+1}J_{s+1}\left( \alpha \left( l\right) \right) \right) . \label{shiftident}\tag{37}\] So \(c_{s}=0\) for even \(s\). In the weak-field (small \(\alpha \left( l\right) \sim g\varepsilon\)) version, using the expansion of the Bessel function, for even weights, we get\[f_{0}\left( \alpha \right) =1-\frac{\alpha ^{2}}{4}+\mathcal{O}\left( \alpha ^{4}\right) ,\;f_{\pm 2}=\frac{\alpha ^{2}}{8}+\mathcal{O}\left( \alpha ^{4}\right) ,\;f_{|s|\geqslant 4}=\mathcal{O}\left( \alpha ^{|s|}\right) , \label{evenveights}\tag{38}\] and \(f_{odd}=0\). For odd weights, \(c_{s}\), we get\[\begin{align} c_{\pm 1}\left( \alpha \right) &=&\frac{1}{2}\left( f_{0}+f_{\pm 2}\right) =\frac{1}{2}\left( 1-\frac{\alpha ^{2}}{4}+\frac{\alpha ^{2}}{8}\right) +\mathcal{O}\left( \alpha ^{4}\right) =\frac{1}{2}-\frac{\alpha ^{2}}{16}+\mathcal{O}\left( \alpha ^{4}\right) , \notag \\ c_{\pm 3}\left( \alpha \right) &=&\frac{1}{2}\left( f_{\pm 2}+f_{\pm 4}\right) =\frac{\alpha ^{2}}{16}+\mathcal{O}\left( \alpha ^{4}\right) , \label{oddweights} \end{align}\tag{39}\] and \(c_{even}=0\). So, to \(O\left( \alpha ^{2}\right)\), only \(c_{\pm 1}\left( \alpha \right) \approx \frac{1}{2}-\frac{\alpha ^{2}}{16}\), \(c_{\pm 3}\left( \alpha \right) \approx \frac{\alpha ^{2}}{16}\) matter.

Now, we define the linear operator insertion (Dirac+color)\[\mathcal{O}^{\left( 1\right) }\left( l\right) \equiv \frac{g}{2\left( l\cdot \kappa \right) }\left( \left( \gamma ^{\nu }\kappa _{\nu }\right) \left( \gamma ^{\lambda }\mathcal{\varepsilon }_{\lambda }^{b}\right) T_{b}+\left( \left( \gamma ^{\sigma }\right) ^{\dagger }\mathcal{\varepsilon }_{\sigma }^{e}\right) \left( \left( \gamma ^{\rho }\right) ^{\dagger }\kappa _{\rho }\right) T_{e}\right) . \label{O1ldef}\tag{40}\] Then the weak-field Bessel-dressed linear harmonic is\[\begin{align} &&\Gamma _{\left( 1\right) ,s}^{c\mu }\left( p,k\right) =\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}\frac{iH_{\alpha \beta }\left( q\right) }{D_{P}D_{r}D_{g}} \notag \\ &&\times \left[ c_{s}\left( \alpha \left( _{P}\right) \right) \gamma ^{\alpha }T^{a}\left( i\left( \cancel{P}+m\right) \right) \mathcal{O}^{\left( 1\right) }\left( P\right) \gamma ^{\mu }T^{c}\left( i\left( \cancel{r}+m\right) \right) \gamma ^{\beta }T^{a}\right. \notag \\ &&\left. +c_{s}\left( \alpha \left( r\right) \right) \gamma ^{\alpha }T^{a}\left( i\left( \cancel{P}+m\right) \right) \gamma ^{\mu }T^{c}\left( i\left( \cancel{r}+m\right) \right) \mathcal{O}^{\left( 1\right) }\left( r\right) \gamma ^{\beta }T^{a}\right] , \label{Gamma1s} \end{align}\tag{41}\] where \(q=p-r\),\(\;P=r+k\),\(\;D_{P}=P^{2}-m^{2}+i\epsilon\), \(D_{r}=r^{2}-m^{2}+i\epsilon\),\(\;D_{g}=q^{2}+i\epsilon\), with parity selection \(c_{s}=0\) for even \(s\).

At \(\mathcal{O}\left( \left( g\varepsilon \right) ^{2}\right)\), there are three contributions. We examine the \(XY\) insertion (even harmonics, uses \(f_{s}\)). We define\[\mathcal{O}^{\left( 2\right) }\left( l\right) \equiv \left( \frac{g}{2\left( l\cdot \kappa \right) }\right) ^{2}\left( \gamma ^{\nu }\kappa _{\nu }\right) \left( \gamma ^{\lambda }\mathcal{\varepsilon }_{\lambda }^{b}\right) \left( \left( \gamma ^{\sigma }\right) ^{\dagger }\mathcal{\varepsilon }_{\sigma }^{e}\right) \left( \left( \gamma ^{\rho }\right) ^{\dagger }\kappa _{\rho }\right) T_{b}T_{e} \label{O2ldef}\tag{42}\] Then\[\begin{align} &&\Gamma _{\left( 2\right) ,s}^{c\mu }|_{\mathcal{O}^{\left( 2\right) }}=\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}\frac{iH_{\alpha \beta }\left( q\right) }{D_{P}D_{r}D_{g}} \notag \\ &&\left[ f_{s}\left( \alpha \left( P\right) \right) \gamma ^{\alpha }T^{a}\left( i\left( \cancel{P}+m\right) \right) \mathcal{O}^{\left( 2\right) }\left( P\right) \gamma ^{\mu }T^{c}\left( i\left( \cancel{r}+m\right) \right) \gamma ^{\beta }T^{a}\right. \notag \\ &&\left. +f_{s}\left( \alpha \left( r\right) \right) \gamma ^{\alpha }T^{a}\left( i\left( \cancel{P}+m\right) \right) \gamma ^{\mu }T^{c}\left( i\left( \cancel{r}+m\right) \right) \mathcal{O}_{XY}^{\left( 2\right) }\left( r\right) \gamma ^{\beta }T^{a}\right] , \label{Gamma2ssame} \end{align}\tag{43}\] with parity selection \(f_{s}\left( \alpha \right) =0\) for odd \(s\).

Now, let’s analyze the case where one inserts one linear term on each line (even harmonics, convolution of odd weights). This has weight \(c_{s_{1}}\left( \alpha _{P}\right) c_{s_{2}}\left( \alpha _{r}\right)\) with \(s_{1}+s_{2}=s\). Explicitly:\[\begin{align} &&\Gamma _{\left( 2\right) ,s}^{c\mu }|_{\mathcal{O}^{\left( 1\right) }\mathcal{O}^{\left( 1\right) }}=\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}\frac{iH_{\alpha \beta }\left( q\right) }{D_{P}D_{r}D_{g}} \notag \\ &&\times \sum_{s_{1}+s_{2}=s}c_{s_{1}}\left( \alpha \left( P\right) \right) c_{s2}\left( \alpha r\right) \gamma ^{\alpha }T^{a}\left( i\left( \cancel{P}+m\right) \right) \notag \\ &&\times \mathcal{O}^{\left( 1\right) }\left( P\right) \gamma ^{\mu }T^{c}\left( i\left( \cancel{r}+m\right) \right) \mathcal{O}^{\left( 1\right) }\left( r\right) \gamma ^{\beta }T^{a}, \label{Gamma2scross} \end{align}\tag{44}\] and since \(c\) is odd-only, the sum automatically forces \(s\) to be even: \(\Gamma _{\left( 2\right) ,s}^{c\mu }|_{\mathcal{O}^{\left( 1\right) }\mathcal{O}^{\left( 1\right) }}=0\) for odd \(s\). At order \(O\left( \alpha ^{2}\right)\), only \(s_{1}=\pm 1\) and \(s_{2}=\mp 1\) contribute to \(s=0\), while \(s_{1}=\pm 1\) and \(s_{2}=\pm 1\) contribute to \(s=\pm 2\), with \(c_{\pm 1}\approx \frac{1}{2}\). For the monochromatic background \(\mathcal{A}_{\mu }^{a}\left( \varphi \right) =\varepsilon _{\mu }^{a}\cos \left( \varphi \right)\), we have, for any momentum \(\ell\), \(\theta \left( l,\varphi \right) =\alpha _{l}\sin \left( \varphi \right)\). The scalar prefactor in our weak-field expansion is\[Z\left( l;\varphi ,\varphi ^{\prime }\right) =-\frac{1}{2}\alpha \left( l\right) ^{2}\left( \sin ^{2}\left( \varphi \right) +\sin ^{2}\left( \varphi ^{\prime }\right) \right) \mathbf{1.} \label{Zldefin1}\tag{45}\] Expressing \(\sin ^{2}\left( \varphi \right)\) in terms of exponents, we can write \(Z\left( l;\varphi ,\varphi ^{\prime }\right)\) as a harmonic sum in \(e^{is\varphi }\) (and similarly for \(\varphi ^{\prime }\)). The coefficient multiplying the harmonic \(\delta ^{\left( 4\right) }\left( p^{\prime }-p-k-n\kappa \right)\) is\[Z_{s}\left( l\right) =\alpha \left( l\right) ^{2}z_{s}\mathbf{1,\;}z_{0}=-\frac{1}{2},\;z_{+2}=z_{-2}=+\frac{1}{4}, \label{Zslvol}\tag{46}\] \(z_{s}=0\) for all other \(s\). Equivalently, written out:\[Z_{0}\left( l\right) =-\frac{1}{2}\alpha \left( l\right) ^{2}\mathbf{1,\;}Z_{\pm 2}\left( l\right) =+\frac{1}{4}\alpha _{l}^{2}\mathbf{1,} \label{Z02lvol}\tag{47}\] \(Z_{s}\left( l\right) =0\) (\(s\neq 0,\pm 2\)). Here, \(\mathbf{1}\) represents the identity in Dirac and color space; \(Zs\) is a scalar dressing factor. So\[\begin{align} \Gamma _{\left( 2\right) ,s}^{c\mu }|_{Z} &=&\left( -ig^{2}\right) \int \frac{d^{4}r}{\left( 2\pi \right) ^{4}}\frac{iH_{\alpha \beta }\left( q\right) }{D_{P}D_{r}D_{g}} \notag \\ &&\times \left[ \gamma ^{\alpha }T^{a}\left( i\left( \cancel{P}+m\right) \right) Z_{s}\left( P\right) \gamma ^{\mu }T^{c}\left( i\left( \cancel{r}+m\right) \right) \gamma ^{\beta }T^{a}\right. \notag \\ &&\left. +\gamma ^{\alpha }T^{a}\left( i\left( \cancel{P}+m\right) \right) \gamma ^{\mu }T^{c}\left( i\left( \cancel{r}+m\right) \right) Z_{s}\left( r\right) \gamma ^{\beta }T^{a}\right] , \label{Gamma2sZ} \end{align}\tag{48}\] with \(s=0\), \(\pm 2\) only. So \(\Gamma _{\left( 2\right) ,s}^{c\mu }\left( p,k\right) =\Gamma _{\left( 2\right) ,s}^{c\mu }|_{\mathcal{O}^{\left( 1\right) }\mathcal{O}^{\left( 1\right) }}+\Gamma _{\left( 2\right) ,s}^{c\mu }|_{\mathcal{O}^{\left( 2\right) }}+\Gamma _{\left( 2\right) ,s}^{c\mu }|_{Z}\).

The full vertex in the periodic background decomposes as\[\Gamma ^{c\mu }\left( p^{\prime },p\right) =\sum_{n\in \mathcal{Z}}\left( 2\pi \right) ^{4}\delta ^{\left( 4\right) }\left( p^{\prime }-p-k-n\kappa \right) \Gamma _{\left[ n\right] }^{c\mu }\left( p,k\right) . \label{fullvertexperiod}\tag{49}\] where\[\Gamma _{\left[ \pm 1\right] }^{c\mu }=\Gamma _{\left( 1\right) ,\pm 1}^{c\mu },\;\Gamma _{\left[ 0\right] }^{c\mu }=\Gamma _{\left[ 0\right] }^{c\mu }=\Gamma _{\left( 0\right) }^{c\mu }+\Gamma _{\left( 2\right) ,0}^{c\mu },\;\Gamma _{\left[ \pm 2\right] }^{c\mu }=\Gamma _{\left( 2\right) ,\pm 2}^{c\mu }. \label{Gam102}\tag{50}\]

Now, we examine the UV behavior of the one-loop correction. For large Euclidean \(r\) (or large Minkowski invariant \(r^{2}\)), the vacuum one-loop vertex behaves as usual and is logarithmically UV divergent. Each linear insertion carries a factor \(U^{\left( 1\right) }\left( l\right) \sim \frac{g}{l\cdot \kappa }\cancel{\kappa}\cancel{\varepsilon}\) \(\Rightarrow\) \(U^{\left( 1\right) }\left( l\right) \sim \mathcal{O}\left( \frac{1}{r}\right)\) (\(r\rightarrow \infty\)), because \(l\cdot \kappa \sim r\) for generic directions. Quadratic pieces scale as \(U^{\left( 2\right) }\left( l\right) \sim \mathcal{O}\left( \frac{1}{\left( l\cdot \kappa \right) ^{2}}\right) =\mathcal{O}\left( \frac{1}{r^{2}}\right)\). As a result, \(\Gamma _{(0)}\) has the same UV divergence as the vacuum QCD vertex. The integral is UV-convergent since we obtain an extra factor of \(1/r\) when comparing \(\Gamma _{\left( 1\right) }\) to the vacuum. The convergence of \(\Gamma _{\left( 2\right) }\) is increased since we obtain \(1/r^{2}\) (or two factors of \(1/r\)). So, the UV divergent part is entirely contained within the vacuum-like component: \(\left[ \Gamma ^{c\mu }\right] _{\mathrm{div}}=\left[ \Gamma _{\left( 0\right) }^{c\mu }\right] _{\mathrm{div}}\),\(\;\left[ \Gamma _{\left( 1\right) }^{c\mu }\right] _{\mathrm{div}}=0\),\(\;\left[ \Gamma _{\left( 2\right) }^{c\mu }\right] _{\mathrm{div}}=0\), up to possible scheme-dependent subtleties with the axial-gauge prescription (which influence how you handle spurious poles but do not introduce new UV divergences related to the background). Since \(\Gamma _{\left( 1\right) }\) and \(\Gamma _{\left( 2\right) }\) are UV finite, the background only affects the finite remainder: \(\Gamma ^{c\mu }=\Gamma _{\left( 0\right) }^{c\mu }+\Gamma _{\left( 1\right) }^{c\mu }|_{finite}+\Gamma _{\left( 2\right) }^{c\mu }|_{finite}+\cdots\), and those finite pieces come with harmonic delta functions that enforce \(p^{\prime }-p-k=n\kappa\).

4 Fermion effective mass↩︎

The fermion self-energy is given by\[\Sigma \left( p\right) =i\int \frac{d^{4}q}{\left( 2\pi \right) ^{4}}\left( -ig\gamma ^{\mu }T^{a}\right) \tilde{G}\left( p-q\right) \left( -ig\gamma ^{\nu }T^{b}\right) \tilde{D}_{\mu \nu }^{ab}\left( q\right) \label{fermionselfernergydef}\tag{51}\]

Substituting \(\tilde{G}\) and \(\tilde{D}_{\mu \nu }^{ab}\) into (51 ), we obtain\[\Sigma \left( p\right) =-ig^{2}\int \frac{d^{4}q}{\left( 2\pi \right) ^{4}}\left\{ \mathcal{I}_{1}-\mathcal{I}_{2}-\mathcal{I}_{3}\right\} , \label{fermionWdef}\tag{52}\] where\[\begin{align} \mathcal{I}_{1} &=&\frac{\gamma ^{\mu }T^{a}\left( \cancel{r}+m\right) U\left( r\right) \gamma _{\mu }T^{a}}{D_{F}\left( r\right) D_{G}\left( q\right) }, \notag \\ \mathcal{I}_{2} &=&\frac{\cancel{q}T^{a}\left( \cancel{r}+m\right) U\left( r\right) \cancel{n}T^{a}\left( n^{\ast }\cdot q\right) }{D_{F}\left( r\right) D_{G}\left( q\right) D_{ax}\left( q\right) }, \notag \\ \mathcal{I}_{3} &=&\frac{\cancel{n}T^{a}\left( \cancel{r}+m\right) U\left( r\right) \cancel{q}T^{a}\left( n^{\ast }\cdot q\right) }{D_{F}\left( r\right) D_{G}\left( q\right) D_{ax}\left( q\right) }, \notag \\ D_{F}\left( r\right) &\equiv &\left( r^{2}-m^{2}+i\epsilon \right) ,\;D_{G}\left( q\right) \equiv \left( q^{2}+i\epsilon \right) , \notag \\ D_{ax}\left( q\right) &\equiv &\left( n^{\ast }\cdot q\right) \left( q\cdot n\right) +i\epsilon ,\;r\equiv p-q, \label{Wdef} \end{align}\tag{53}\] We use the cyclicity of the trace: \(\mathrm{Tr}\left[ \left( \cancel{p}+m\right) \gamma ^{\mu }L\gamma _{\mu }\right] =\mathrm{Tr}\left[ \left( -2\cancel{p}+4m\right) L\right]\), and define\[\delta m\equiv \frac{1}{4m}\mathrm{Tr}\left[ \left( \cancel{p}+m\right) \Sigma \left( p\right) \right] |_{p^{2}=m^{2}}, \label{defdeltam}\tag{54}\] so that (to one-loop order) \(m_{eff}=m+\delta m\). It is clear that as the limit \(\mathcal{A\rightarrow }0\), \(U\rightarrow 1\), and we substitute \(T^{a}=1\), \(g=e\), we obtain\[\Sigma _{El}\left( p\right) =-ie^{2}\int \frac{d^{4}q}{\left( 2\pi \right) ^{4}}\left[ \frac{\gamma ^{\mu }\left( \cancel{r}+m\right) \gamma _{\mu }}{D_{F}\left( r\right) D_{G}\left( q\right) }-\frac{\cancel{q}\left( \cancel{r}+m\right) \cancel{n}\left( n^{\ast }\cdot q\right) }{D_{F}\left( r\right) D_{G}\left( q\right) D_{ax}\left( q\right) }-\frac{\cancel{n}\left( \cancel{r}+m\right) \cancel{q}\left( n^{\ast }\cdot q\right) }{D_{F}\left( r\right) D_{G}\left( q\right) D_{ax}\left( q\right) }\right] . \label{elecselfenergy}\tag{55}\] This corresponds to the electron’s self-energy in electromagnetic interactions.

We expand \(U(r)\), keeping \(C\left( r\right) \equiv \cos \left( \theta \left( p,\varphi \right) \right) \cos \left( \theta \left( p,\varphi ^{\prime }\right) \right)\), \[U\left( r\right) =C\left( r\right) \left[ 1+\Delta ^{\left( 1\right) }\left( r\right) +\Delta ^{\left( 2\right) }\left( r\right) \right] , \label{Uexpand}\tag{56}\] where\[\begin{align} \alpha ^{\prime } &\equiv &\frac{g}{2\left( r\cdot \kappa \right) }\frac{\tan \left( \theta \left( p,\varphi ^{\prime }\right) \right) }{\theta \left( p,\varphi ^{\prime }\right) },\;\beta \equiv \frac{g}{2\left( r\cdot \kappa \right) }\frac{\tan \left( \theta \left( p,\varphi \right) \right) }{\theta \left( p,\varphi \right) }, \notag \\ \Delta ^{\left( 1\right) }\left( r\right) &=&\alpha ^{\prime }\mathcal{\cancel{A}}^{e}\left( \varphi ^{\prime }\right) \cancel{\kappa}T_{e}+\beta \cancel{\kappa}\mathcal{\cancel{A}}^{b}\left( \varphi \right) T_{b}, \notag \\ \Delta ^{\left( 2\right) }\left( r\right) &=&\alpha ^{\prime }\beta \left( \mathcal{\cancel{A}}^{e}\left( \varphi ^{\prime }\right) \cancel{\kappa}\right) \left( \cancel{\kappa}\mathcal{\cancel{A}}^{b}\left( \varphi \right) \right) T_{b}T_{e}. \label{Delta1Delta2def} \end{align}\tag{57}\] We have for the "1" in \(U\), \(T^{a}\left( \cdots \right) T^{a}\rightarrow C_{F}\left( \cdots \right)\) with \(C_{F}=\frac{N^{2}-1}{2N}\) (in the fundamental representation of \(SU(N)\)). For \(\delta m_{\left( 1\right) }\) we have \(\delta m_{\left( 1\right) }\equiv \frac{1}{4m}\mathrm{Tr}\left[ \left( \cancel{p}+m\right) \gamma ^{\mu }\left( \cancel{r}+m\right) U\left( r\right) \gamma _{\mu }\right] =\frac{1}{4m}\mathrm{Tr}\left[ \left( -2\cancel{p}+4m\right) \left( \cancel{r}+m\right) U\left( r\right) \right]\). Replacing \(U(r)\), we get for a vacuum-like segment (no explicit \(\mathcal{A}\) insertion beyond \(C(r)\)) the expression:\[\delta m_{\left( 1\right) }^{\left( 0\right) }=C_{F}C\left( r\right) \left( 4m-\frac{2}{m}p\cdot r\right) \label{deltamvac1}\tag{58}\] then we compute linear-in-\(\mathcal{A}\) insertions from \(\Delta ^{\left( 1\right) }\). For \(\mathcal{\cancel{A}}^{e}\left( \varphi ^{\prime }\right) \cancel{\kappa}\), and \(\cancel{\kappa}\mathcal{\cancel{A}}^{b}\) using \(k^{2}=0\), \(\kappa \mathcal{A}^{e}\left( \varphi ^{\prime }\right) =0\), \(\kappa \mathcal{A}^{b}\left( \varphi \right) =0\).\[\begin{align} \frac{1}{4m}\mathrm{Tr}\left[ \left( -2\cancel{p}+4m\right) \left( \cancel{r}+m\right) \mathcal{\cancel{A}}^{e}\left( \varphi ^{\prime }\right) \cancel{\kappa}\right] &=&\frac{2}{m}\left[ \left( p\cdot \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \right) \left( r\cdot \kappa \right) -\left( p\cdot \kappa \right) \left( r\cdot \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \right) \right] , \notag \\ \frac{1}{4m}\mathrm{Tr}\left[ \left( -2\cancel{p}+4m\right) \left( \cancel{r}+m\right) \cancel{\kappa}\mathcal{\cancel{A}}^{b}\left( \varphi \right) \right] &=&\frac{2}{m}\left[ \left( p\cdot \kappa \right) \left( r\cdot \mathcal{A}^{b}\left( \varphi \right) \right) -\left( p\cdot \mathcal{A}^{b}\left( \varphi \right) \right) \left( r\cdot \kappa \right) \right] . \label{Tr40kA41} \end{align}\tag{59}\] So the linear part of \(\delta m_{\left( 1\right) }\) is\[\begin{align} \delta m_{\left( 1\right) }^{\left( 1\right) } &=&C\left( r\right) \left( C_{F}-\frac{C_{A}}{2}\right) \left[ \alpha ^{\prime }\frac{2}{m}\left( \left( p\cdot \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \right) \left( r\cdot \kappa \right) -\left( p\cdot \kappa \right) \left( r\cdot \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \right) \right) T_{e}\right. \notag \\ &&\left. +\beta \frac{2}{m}\left( \left( p\cdot \kappa \right) \left( r\cdot \mathcal{A}^{b}\left( \varphi \right) \right) -\left( p\cdot \mathcal{A}^{b}\left( \varphi \right) \right) \left( r\cdot \kappa \right) \right) T_{b}\right] . \label{deltam11} \end{align}\tag{60}\] where we used \(T^{a}T^{b}T^{a}=\left( C_{F}-\frac{C_{A}}{2}\right) T^{b}\),\(\;C_{A}=N\). For the vacuum-like axial trace (again exact in \(C\left( r\right)\)) with \(U\rightarrow C\left( r\right)\) and color \(T^{a}\left( \cdots \right) T^{a}\rightarrow C_{F}\left( \cdots \right)\), we have the linear axial trace from \(\Delta ^{\left( 1\right) }\) (for the \(\alpha ^{\prime }\mathcal{\cancel{A}}^{e}\left( \varphi ^{\prime }\right) \cancel{k}T_{e}\) insertion):\[\begin{align} \delta m_{\left( 2+3\right) }^{\left( 1\right) }|_{\mathcal{A}^{\prime }} &=&C\left( r\right) \alpha ^{\prime }\frac{2}{m}\left[ -\left( \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \cdot n\right) \left( p\cdot q\right) \right. \notag \\ &&\left. +\left( p\cdot \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \right) \left( r\cdot \kappa \right) \left( n\cdot q\right) -\left( \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \cdot q\right) \left( r\cdot \kappa \right) \left( n\cdot p\right) \right. \notag \\ &&\left. +\left( \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \cdot r\right) \left( \kappa \cdot n\right) \left( p\cdot q\right) -\left( \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \cdot r\right) \left( \kappa \cdot p\right) \left( n\cdot q\right) \right. \notag \\ &&\left. +\left( \mathcal{A}^{e}\left( \varphi ^{\prime }\right) \cdot r\right) \left( \kappa \cdot q\right) \left( n\cdot p\right) \right] \left( C_{F}-\frac{C_{A}}{2}\right) T_{e}, \label{deltam123alfsh} \end{align}\tag{61}\] (for the \(\beta \mathcal{\cancel{A}}^{e}\left( \varphi ^{\prime }\right) \cancel{k}T_{b}\) insertion):\[\begin{align} \delta m_{\left( 2+3\right) }^{\left( 1\right) }|_{\mathcal{A}} &=&C\left( r\right) \beta \frac{2}{m}\left[ +\left( \mathcal{A}^{b}\left( \varphi \right) \cdot n\right) \left( r\cdot \kappa \right) \left( p\cdot q\right) \right. \notag \\ &&\left. -\left( p\cdot \mathcal{A}^{b}\left( \varphi \right) \right) \left( r\cdot \kappa \right) \left( n\cdot q\right) +\left( \mathcal{A}^{b}\left( \varphi \right) \cdot q\right) \left( r\cdot \kappa \right) \left( n\cdot p\right) \right. \notag \\ &&\left. -\left( \mathcal{A}^{b}\left( \varphi \right) \cdot r\right) \left( \kappa \cdot n\right) \left( p\cdot q\right) +\left( \mathcal{A}^{b}\left( \varphi \right) \cdot r\right) \left( \kappa \cdot p\right) \left( n\cdot q\right) \right. \notag \\ &&\left. -\left( \mathcal{A}^{b}\left( \varphi \right) \cdot q\right) \left( r\cdot \kappa \right) \left( n\cdot p\right) \right] \left( C_{F}-\frac{C_{A}}{2}\right) T_{b}. \label{deltam123bet} \end{align}\tag{62}\] Under the same plane-wave constraints \(k^{2}=0\), \(k\mathcal{A}^{b}\left( \varphi \right) =0\), \(k\mathcal{A}^{e}\left( \varphi ^{\prime }\right) =0\), the scalar mass projector from \(\Delta ^{\left( 2\right) }\) also vanishes: \(\delta m_{\left( 2+3\right) }^{\left( 2\right) }=0\). Collecting the traced results, the complete one-loop on-shell mass shift is\[\begin{align} \delta m &=&-ig^{2}\int \frac{d^{4}q}{\left( 2\pi \right) ^{4}}\frac{1}{D_{F}\left( r\right) D_{G}\left( q\right) } \notag \\ &&\times \left\{ \delta m_{\left( 1\right) }^{\left( 0\right) }+\delta m_{\left( 1\right) }^{\left( 1\right) }-\left[ -\frac{\left( n^{\ast }\cdot q\right) }{D_{ax}\left( q\right) }\right] \left( \delta m_{\left( 2+3\right) }^{\left( 0\right) }+\delta m_{\left( 2+3\right) }^{\left( 1\right) }\right) \right\} . \label{deltamfull} \end{align}\tag{63}\]

For transverse plane wave the phase integral provides \(\int_{0}^{\varphi }d\varphi ^{\prime }\left( \mathcal{A}^{a}\cdot r\right) =\left( \varepsilon ^{a}\cdot r\right) \sin \left( \varphi \right)\) and for \(\theta \left( r,\varphi \right)\), we have \(\theta \left( r,\varphi \right) =\alpha \left( r\right) \sin \left( \varphi \right)\), where \(\alpha \left( r\right) =\frac{g}{r\cdot \kappa }\sqrt{\frac{\left( \varepsilon ^{a}\cdot r\right) \left( \varepsilon _{a}\cdot r\right) }{2N}}\) and \[\langle \cos \left( \alpha \left( r\right) \varphi \right) \rangle _{\varphi }=\frac{1}{2\pi }\int_{0}^{2\pi }d\varphi \cos \left( \alpha \left( r\right) \sin \left( \varphi \right) \right) =J_{0}\left( \alpha \left( r\right) \right) , \label{cosaver}\tag{64}\] \[\langle C\left( r\right) \rangle _{\varphi ,\varphi ^{\prime }}=\langle \cos \left( \alpha \left( r\right) \varphi \right) \rangle _{\varphi }^{2}=J_{0}^{2}\left( \alpha \left( r\right) \right) . \label{60C62}\tag{65}\] Using the tracing results obtained above, the cycle-averaged mass displacement on the shell surface is\[\langle \delta m\rangle =-ig^{2}C_{F}\int \frac{d^{4}q}{\left( 2\pi \right) ^{4}}\frac{\mathcal{W}\left( p-q\right) \left[ \mathcal{L}_{F}\left( p,q\right) +\mathcal{L}_{ML}\left( p,q;n,n^{\ast }\right) \right] }{\left[ \left( p-q\right) ^{2}-m^{2}+i\epsilon \right] \left( q^{2}+i\epsilon \right) } \label{deltamass}\tag{66}\] where \(\mathcal{W}\left( r\right) =J_{0}^{2}\left( \alpha \left( r\right) \right)\). The traced numerators (vacuum-like pieces) are \(\mathcal{L}_{F}\left( p,q\right) =2m+\frac{2}{m}\left( p\cdot q\right)\), and the ML/axial term (from our \(\mathcal{I}_{2}+\mathcal{I}_{3}\)) is\[\begin{align} \mathcal{L}_{ML}\left( p,q;n,n^{\ast }\right) &=&\frac{n^{\ast }\cdot q}{\left( n^{\ast }\cdot q\right) \left( n\cdot q\right) +i\epsilon }\frac{2}{m} \notag \\ &&\times \left[ m^{2}\left( n\cdot q\right) -\left( m^{2}-p\cdot q\right) \left( q\cdot n\right) \right. \notag \\ &&\left. +\left( p\cdot n\right) \left( p\cdot q\right) +\left( p\cdot q\right) \left( p\cdot n-q\cdot n\right) \right] . \label{NML} \end{align}\tag{67}\]

To apply the Wick rotation, we can express the loop energy as \(q^{0}\rightarrow iq_{4\text{,}}\;d^{4}q\rightarrow id^{4}q_{E}\),\(\;q^{2}=\left( q^{0}\right) ^{2}-\mathbf{q}^{2}\rightarrow -q_{E}^{2}\). So, the Euclidean form is\[\langle \delta m\rangle =g^{2}C_{F}\int \frac{d^{4}q_{E}}{\left( 2\pi \right) ^{4}}\frac{\mathcal{W}\left( r_{E}\right) \left[ \mathcal{L}_{F}\left( p_{E},q_{E}\right) +\mathcal{L}_{ML}\left( p_{E},q_{E};n_{E},n_{E}^{\ast }\right) \right] }{q_{E}^{2}\left( \left( p_{E}-q_{E}\right) ^{2}+m^{2}\right) }, \label{Euclform}\tag{68}\] where\[\begin{align} \mathcal{W}\left( r_{E}\right) &=&J_{0}^{2}\left( \alpha _{E}\left( r_{E}\right) \right) , \notag \\ \alpha _{E}\left( r_{E}\right) &=&\frac{g}{r_{E}\kappa _{E}}\sqrt{\frac{\left( \varepsilon ^{a}\cdot r_{E}\right) \left( \varepsilon _{a}\cdot r_{E}\right) }{2N}},\;\kappa _{E}\varepsilon ^{a}=0,\;\kappa _{E}^{2}=0, \label{WEucl} \end{align}\tag{69}\] \[\mathcal{L}_{F,E}\left( p_{E},q_{E}\right) =4m-\frac{2}{m}p_{E}\cdot r_{E}, \label{LFEucl}\tag{70}\] \[\begin{align} \mathcal{L}_{ML,E}\left( p_{E},q_{E};n_{E},n_{E}^{\ast }\right) &=&\frac{n_{E}^{\ast }\cdot q_{E}}{\left( n_{E}^{\ast }\cdot q_{E}\right) \left( n_{E}\cdot q_{E}\right) }\frac{2}{m} \notag \\ &&\times \left[ m^{2}\left( n_{E}\cdot q_{E}\right) -\left( m^{2}-p_{E}\cdot q_{E}\right) \left( q_{E}\cdot n_{E}\right) \right. \notag \\ &&\left. +\left( p_{E}\cdot n_{E}\right) \left( p_{E}\cdot q_{E}\right) +\left( p_{E}\cdot q_{E}\right) \left( p_{E}\cdot n_{E}-q_{E}\cdot n_{E}\right) \right] . \label{LMLEucl} \end{align}\tag{71}\] We introduce the Feynman parameter \(x\in \left[ 0,1\right]\) to combine the denominators:\[\begin{align} &&\frac{1}{q_{E}^{2}\left( \left( p_{E}-q_{E}\right) ^{2}+m^{2}\right) } \notag \\ &=&\int_{0}^{1}dx\frac{1}{\left[ \left( q_{E}-xp_{E}\right) ^{2}+x\left( 1-x\right) p_{E}^{2}+xm^{2}\right] ^{2}}. \label{xdenominat} \end{align}\tag{72}\] Denoting \(l_{E}=q_{E}-xp_{E}\), for \(r_{E}\) we have \(r_{E}=p_{E}-q_{E}=\left( 1-x\right) p_{E}-l_{E}\). Then\[\langle \delta m\rangle =g^{2}C_{F}\int_{0}^{1}dx\int \frac{d^{4}l_{E}}{\left( 2\pi \right) ^{4}}\frac{\mathcal{W}\left( \left( 1-x\right) p_{E}-l_{E}\right) \left[ \mathcal{\tilde{L}}_{F,E}+\mathcal{\tilde{L}}_{ML,E}\right] }{\left( l_{E}^{2}+\Delta \left( x\right) \right) ^{2}} \label{60deltamEuc62}\tag{73}\] with \(\Delta \left( x\right) =x\left( 1-x\right) p_{E}^{2}+xm^{2}\). and on shell \(p_{E}^{2}\rightarrow -m^{2}\) so \(\Delta \left( x\right) =x^{2}m^{2}\). At a large Euclidean loop momentum \(\alpha _{E}\left( r_{E}\right) \sim \frac{1}{r_{E}\cdot \kappa _{E}}\times \left( \varepsilon \cdot r_{E}\right) \sim \mathcal{O}\left( 1\right)\), the Bessel factor remains bounded, and in practice, we can safely split: \(\langle \delta m\rangle _{E}=\delta m_{E}^{free}+\delta m_{E}^{bg\left( finit\right) }\), where \(\delta m_{E}^{free}\) is the standard one-loop mass shift, and \(\delta m_{E}^{bg\left( finit\right) }\) is derived from the difference integral\[\delta m_{E}^{bg\left( finit\right) }=g^{2}C_{F}\int_{0}^{1}dx\int \frac{d^{4}l_{E}}{\left( 2\pi \right) ^{4}}\frac{\left[ \mathcal{W}\left( r\left( l_{E},x\right) \right) -1\right] }{\left( l_{E}^{2}+x^{2}m^{2}\right) ^{2}}\left[ \mathcal{\tilde{L}}_{F,E}\left( l_{E},x\right) +\mathcal{\tilde{L}}_{ML,E}\left( l_{E},x\right) \right] . \label{deltamEbg}\tag{74}\] The background-dependent correction is finite because \(\mathcal{W}\left( r\left( l_{E},x\right) \right) -1\) kills the local UV piece. The small-field expansion is \(\mathcal{W}\left( r\left( l_{E},x\right) \right) -1=-\frac{1}{2}\alpha _{E}\left( r_{E}\right) ^{2}+\mathcal{O}\left( \alpha _{E}^{4}\right) ,\) so to \(O\left( \varepsilon ^{4}\right)\),\[\delta m_{E}^{\left( 2\right) }=-\frac{1}{2}g^{2}C_{F}\int_{0}^{1}dx\int \frac{d^{4}l_{E}}{\left( 2\pi \right) ^{4}}\frac{\alpha _{E}\left( r_{E}\right) ^{2}}{\left( l_{E}^{2}+x^{2}m^{2}\right) ^{2}}\left[ \mathcal{\tilde{L}}_{F,E}\left( l_{E},x\right) +\mathcal{\tilde{L}}_{ML,E}\left( l_{E},x\right) \right] . \label{deltamE2}\tag{75}\] All background dependence appears through UV-finite factors, such as \(\cos \left( \theta \right)\), \(J_{0}\left( \alpha \right) ^{2}\), and \(\left( \varepsilon \cdot p\right) ^{2}/\left( p\cdot \kappa \right) ^{2}\), confirming that mass renormalization remains local. Therefore, external classical fields do not alter the UV structure of renormalizable QFT.

5 Fermion condensate in an external Yang-Mills gauge field↩︎

The field-induced condensate is computed by explicitly subtracting \(\langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A}}-\langle 0|\bar{\psi}\psi |0\rangle _{free}\), which removes the divergence of the free propagator. This method aligns with renormalization schemes where free-field subtractions regularize VEVs before incorporating interaction effects. We write

\[\langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}=\langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A}}-\langle 0|\bar{\psi}\psi |0\rangle _{free} \label{normalordered}\tag{76}\] where \[\langle \bar{\psi}\psi \rangle _{\mathcal{A}}=-\int \frac{d^{4}p}{\left( 2\pi \right) ^{4}}\mathrm{Tr}\left[ \tilde{G}_{F}\left( p\right) \right] , \label{ferconddef}\tag{77}\]

Substituting (16 ) into (fer@eq:ferconddef ) with \(x=x^{\prime }\), we obtain\[\langle \bar{\psi}\psi \rangle _{\mathcal{A}}=-\int \frac{d^{4}p}{\left( 2\pi \right) ^{4}}\mathrm{Tr}\left[ \frac{i\left( \cancel{p}+m\right) \tilde{U}\left( p\right) }{p^{2}-m^{2}+i\epsilon }\right] , \label{fercondU}\tag{78}\] where \[\begin{align} &&U\left( p\right) |_{x=x^{\prime }} \notag \\ &=&\tilde{U}\left( p\right) =\cos ^{2}\left( \theta \left( p,\varphi \right) \right) \notag \\ &&\times \left\{ 1+g\frac{\tan \left( \theta \left( p,\varphi \right) \right) }{2\left( p\kappa \right) \theta \left( p,\varphi \right) }\left( \left( \left( \gamma ^{\sigma }\right) ^{\dagger }\mathcal{A}_{\sigma }^{e}\left( \varphi \right) \right) \left( \left( \gamma ^{\rho }\right) ^{\dagger }k_{\rho }\right) T_{e}+\left( \gamma ^{\nu }k_{\nu }\right) \left( \gamma ^{\lambda }\mathcal{A}_{\lambda }^{b}\left( \varphi \right) \right) T_{b}\right) \right. \notag \\ &&\left. +g^{2}\left( \frac{\tan \left( \theta \left( p,\varphi \right) \right) }{2\left( p\kappa \right) \theta \left( p,\varphi \right) }\right) ^{2}\left( \gamma ^{\nu }k_{\nu }\right) \left( \gamma ^{\lambda }\mathcal{A}_{\lambda }^{b}\left( \varphi \right) \right) \left( \left( \gamma ^{\sigma }\right) ^{\dagger }\mathcal{A}_{\sigma }^{e}\left( \varphi \right) \right) \left( \left( \gamma ^{\rho }\right) ^{\dagger }k_{\rho }\right) T_{b}T_{e}\right\} , \label{Ux} \end{align}\tag{79}\] and \[\theta \left( p\right) =\theta \left( p,\varphi \right) =\frac{g}{\left( p\kappa \right) }\sqrt{\frac{1}{2N}}\left( \int_{0}^{\varphi }d\varphi ^{\prime }\left( \mathcal{A}_{\mu }^{a}\left( \varphi ^{\prime }\right) p^{\mu }\right) \int_{0}^{\varphi }d\varphi ^{\prime \prime }\left( \mathcal{A}_{a}^{\mu }\left( \varphi ^{\prime \prime }\right) p_{\mu }\right) \right) ^{\frac{1}{2}}, \label{tetx}\tag{80}\] When \(U(p)\rightarrow 1\), we obtain\[\mathrm{Tr}\left[ i\left( \cancel{p}+m\right) \right] =i\mathrm{Tr}_{D}\left( \cancel{p}+m\right) \mathrm{Tr}_{c}\left( \mathbf{1}\right) =i\left( 0+4m\right) N=4imN \label{trfree}\tag{81}\] so\[\langle \bar{\psi}\psi \rangle _{free}=-4imN\int \frac{d^{4}p}{\left( 2\pi \right) ^{4}}\frac{1}{p^{2}-m^{2}+i\epsilon } \label{condfree}\tag{82}\] This is UV divergent and must be regularized.

Considering that only terms with an even number of gamma matrices contribute and \(\mathrm{tr}_{c}\left( T^{a}\right) =0\), and in axial gauge with \(\kappa ^{\mu }\mathcal{A}_{\mu }^{a}=0\) and \(\kappa ^{2}=0\), we have\[\begin{align} \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free} &=&-4imN\int \frac{d^{4}p}{\left( 2\pi \right) ^{4}}\frac{\cos ^{2}\left( \theta \left( p,\varphi \right) \right) -1}{p^{2}-m^{2}+i\epsilon } \notag \\ &=&2imN\int \frac{d^{4}p}{\left( 2\pi \right) ^{4}}\frac{1-\cos \left( 2\theta \left( p,\varphi \right) \right) }{p^{2}-m^{2}+i\epsilon }, \label{fercondcos} \end{align}\tag{83}\]

The condensate vanishes identically when the background is switched off (\(\theta \rightarrow 0\)). So this is a genuine response functional of the fermionic vacuum to the background gauge field. The dependence on the background enters only through the gauge-invariant scalar phase \(\theta \left( p,\varphi \right)\). The contribution of field-free divergence is removed through subtraction. The integrand has a clear sign in Euclidean space because \(1-\cos (2\theta )=2\sin ^{2}\theta \geq 0\). The background always makes the scalar density bigger than it would be without the field.

In the case of the monochromatic plane wave, it is useful to package the quadratic form\[\left( \varepsilon ^{a}\cdot p\right) \left( \varepsilon _{a}\cdot p\right) =p_{\mu }\Xi ^{\mu \nu }p_{\nu },\;\Xi ^{\mu \nu }\equiv \varepsilon ^{a\mu }\varepsilon _{a}^{\nu }. \label{epep}\tag{84}\] So\[2\theta \left( p,\varphi \right) =\frac{2g}{p\kappa }\sqrt{\frac{1}{2N}}\sqrt{p_{\mu }\Xi ^{\mu \nu }p_{\nu }}\sin \left( \varphi \right) . \label{2tetpfi}\tag{85}\] For a monochromatic wave, the clearest “oscillatory contribution”is the average over \(\varphi \in \left[ 0,2\pi \right]\). We use the exact Bessel identity \(\frac{1}{2\pi }\int_{0}^{2\pi }d\varphi \cos \left( a\sin \varphi \right) =J_{0}\left( a\right)\). Then\[\langle \cos \left( 2\theta \left( p,\varphi \right) \right) \rangle _{\varphi }=J_{0}\left( \frac{2g}{p\kappa }\sqrt{\frac{1}{2N}}\sqrt{p\Xi p}\right) . \label{cosaverage}\tag{86}\] Therefore, the precise phase-averaged condensate is\[\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }=2imN\int \frac{d^{4}p}{\left( 2\pi \right) ^{4}}\frac{1-J_{0}\left( \frac{2g}{p\kappa }\sqrt{\frac{1}{2N}}\sqrt{p\Xi p}\right) }{p^{2}-m^{2}+i\epsilon }. \label{exactphasavercond}\tag{87}\]

In \(d=4-2\eta\), we define the regulated (phase-averaged) integral:\[\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }^{DR}=2imN\mu ^{2\eta }\int \frac{d^{4-2\eta }p}{\left( 2\pi \right) ^{4-2\eta }}\frac{1-J_{0}\left( \frac{2g}{p\kappa }\sqrt{\frac{1}{2N}}\sqrt{p\Xi p}\right) }{p^{2}-m^{2}+i\epsilon }. \label{DRavercond}\tag{88}\]

Pauli–Villars (PV) implements\[\frac{1}{p^{2}-m^{2}+i\epsilon }\rightarrow \frac{1}{p^{2}-m^{2}+i\epsilon }-\frac{1}{p^{2}-M^{2}+i\epsilon },\;M\gg m, \label{PVreg}\tag{89}\] so the regulated, phase-averaged condensate is\[\begin{align} &&\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }^{PV} \notag \\ &=&2imN\int \frac{d^{4}p}{\left( 2\pi \right) ^{4}}\left( \frac{1}{p^{2}-m^{2}+i\epsilon }-\frac{1}{p^{2}-M^{2}+i\epsilon }\right) \left[ 1-J_{0}\left( \frac{2g}{p\kappa }\sqrt{\frac{1}{2N}}\sqrt{p\Xi p}\right) \right] . \label{PVaverregul} \end{align}\tag{90}\]

Now we choose the coordinates of the light front:\[p^{\pm }=p^{0}\pm p^{3},\;p^{2}=p^{+}p^{-}-p_{\perp }^{2}. \label{lightfroncoor}\tag{91}\] and\[\kappa ^{\mu }=\left( \varpi ,0,\mathbf{0}_{\perp }\right) \;\Rightarrow \;p\kappa =\frac{1}{2}\varpi p^{-}. \label{lightfrontmom}\tag{92}\] In axial gauge for a plane wave, we choose \(\varepsilon ^{a}\) purely transverse, so\[\left( \varepsilon ^{a}\cdot p\right) =-\varepsilon _{\perp }^{a}\cdot p_{\perp },\;\Rightarrow \;p\Xi p=p_{\perp }^{i}C_{ij}p_{\perp }^{j},\;C_{ij}\equiv \varepsilon _{\perp }^{ai}\varepsilon _{\perp }^{aj}. \label{Cdef}\tag{93}\] We begin by integrating \(p^{-}\) over residues using the phase-averaged Minkowski form:\[\int^{+}\frac{dp^{-}}{2\pi }\frac{i}{p^{+}p^{-}-p_{\perp }^{2}-m^{2}+i\epsilon }=\frac{1}{p^{+}}\theta \left( p^{+}\right) \frac{1}{2}, \label{integrpmin}\tag{94}\] where theta is the Heaviside function. The condensate then becomes\[\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }=mN\int_{p^{+}>0}\frac{dp^{+}d^{2}p_{\perp }}{\left( 2\pi \right) ^{3}}\frac{1}{p^{+}}\left[ 1-J_{0}\left( \frac{4g}{\varpi }\sqrt{\frac{1}{2N}}\frac{\sqrt{p_{\perp }Cp_{\perp }}}{p_{on}^{-}}\right) \right] , \label{condlightfront}\tag{95}\] where\[p_{on}^{-}=\frac{p_{\perp }^{2}+m^{2}}{p^{+}}. \label{pminuson}\tag{96}\] We then use different light-front UV cutoff surfaces: ” rectangular”cutoffs in \(\left( p^{+},p_{\perp }\right)\) and covariant “invariant mass” (Brodsky–Lepage) cutoff. We select the following conditions for “rectangular”cutoffs \(\delta \leq p^{+}\leq \Lambda ^{+},\;|p_{\perp }|\leq \Lambda _{\perp }\). Then \[\begin{align} &&\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }^{LF\;rect} \notag \\ &=&mN\int_{\delta }^{\Lambda ^{+}}\frac{dp^{+}}{p^{+}}\int_{|p_{\perp }|}\frac{d^{2}p_{\perp }}{\left( 2\pi \right) ^{3}}\left[ 1-J_{0}\left( \frac{4g}{\varpi }\sqrt{\frac{1}{2N}}\frac{p^{+}\sqrt{p_{\perp }Cp_{\perp }}}{p_{\perp }^{2}+m^{2}}\right) \right] . \label{condensateLFrect} \end{align}\tag{97}\] This makes it completely clear how UV sensitivity occurs through large \(p_{\perp }\) and large \(p^{+}\) (and also IR sensitivity through \(\delta \rightarrow 0\)).

For a one-loop vacuum-type integral like ours, a commonly used invariant-mass restriction is to bound \(\frac{p_{\perp }^{2}+m^{2}}{p^{+}}\leq \Lambda _{inv}\) or, in terms of the on-shell invariant \(p^{+}p_{on}^{-}=p_{\perp }^{2}+m^{2}\), \[p_{\perp }^{2}+m^{2}\leq \Lambda _{inv}^{2}. \label{covariantcutoff}\tag{98}\] Either way, the integration region is no longer a rectangle but a “covariant”domain in LF variables. A clear, explicit choice is\[\delta \leq p^{+}\leq \Lambda ^{+},\;p_{\perp }^{2}+m^{2}\leq \Lambda _{inv}^{2}, \label{covariantdomain}\tag{99}\] giving\[\begin{align} \langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }^{LF\;inv} &=&mN\int_{\delta }^{\Lambda ^{+}}\frac{dp^{+}}{p^{+}}\int_{p_{\perp }^{2}\leq \Lambda _{inv}^{2}-m^{2}}\frac{d^{2}p_{\perp }}{\left( 2\pi \right) ^{3}} \notag \\ &&\times \left[ 1-J_{0}\left( \frac{4g}{\varpi }\sqrt{\frac{1}{2N}}\frac{p^{+}\sqrt{p_{\perp }Cp_{\perp }}}{p_{\perp }^{2}+m^{2}}\right) \right] . \label{condensatecovar} \end{align}\tag{100}\] Let’s consider the case of large \(\Lambda\) and large \(\Lambda ^{+}\) for the light front cutoff invariant mass. If the transverse polarization/color structure is isotropic, we can take \(C_{ij}=\tau \delta _{ij}\), so \(\sqrt{p_{\perp }Cp_{\perp }}=\sqrt{\tau }|p_{\perp }|\). We define the momentum fraction with respect to the external positive scale:\[x\equiv \frac{p^{+}}{\varpi }\in \left( 0,1\right) ,\;\frac{dp^{+}}{p^{+}}=\frac{dx}{x}. \label{defx}\tag{101}\] Then the Bessel argument simplifies nicely and becomes\[\left( 4g\sqrt{\tau }\sqrt{\frac{1}{2N}}\right) \frac{x\rho }{\rho ^{2}+m^{2}},\;\rho \equiv |p_{\perp }|. \label{Besselarg}\tag{102}\] The invariant-mass cutoff for two-particle kinematics is \(\frac{m^{2}+\rho ^{2}}{x\left( 1-x\right) }\leq \Lambda ^{2}\), therefore \(\;0\leq \rho \leq \rho _{\max }\left( x\right) \equiv \sqrt{\Lambda ^{2}x\left( 1-x\right) -m^{2}}\). \(\;\)This requires \(\Lambda ^{2}x\left( 1-x\right) \geq m^{2}\), i.e.\[x\in \left[ x_{-},x_{+}\right] ,\;x_{\pm }=\frac{1}{2}\left( 1\pm \sqrt{1-\frac{4m^{2}}{\Lambda ^{2}}}\right) , \label{xxplusminus}\tag{103}\] So we also see right away that we need \(\Lambda >2m\) for a nonempty domain. Putting it all together yields the fully explicit invariant-mass–cutoff result:\[\begin{align} &&\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }^{inv.\;mass} \notag \\ &=&\frac{mN}{4\pi ^{2}}\int_{x_{-}}^{x_{+}}\frac{dx}{x}\int_{0}^{\sqrt{\Lambda ^{2}x\left( 1-x\right) -m^{2}}}\rho d\rho \left[ 1-J_{0}\left( \zeta \frac{x\rho }{\rho ^{2}+m^{2}}\right) \right] , \label{condeninvmass} \end{align}\tag{104}\] where \(\zeta =4g\sqrt{\tau }\sqrt{\frac{1}{2N}}\). The UV region is \(\rho \gg m\). There,\[z\left( x,\rho \right) \equiv \zeta \frac{x\rho }{\rho ^{2}+m^{2}}\simeq \zeta \frac{x}{\rho }\ll 1, \label{zxro}\tag{105}\] and the only thing we need is the small-argument expansion \(J_{0}\left( z\right) =1-\frac{z^{2}}{4}+\mathcal{O}\left( z^{4}\right)\), therefore \(1-\;J_{0}\left( z\right) =\frac{z^{2}}{4}+\mathcal{O}\left( z^{4}\right)\), and the UV tail of the \(\rho\)-integral behaves as\[\rho d\rho \left[ 1-J_{0}\left( z\right) \right] \simeq \frac{\zeta ^{2}x^{2}}{4}\frac{d\rho }{\rho }, \label{rointegral}\tag{106}\] producing a logarithm cut off at \(\rho _{\max }\left( x\right)\). We match the lower end at \(\rho \sim m\) to clearly isolate the UV log (any \(\mathcal{O}\left( m\right)\) choice shifts only a finite constant). Then, the UV-controlled part is\[\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }^{inv}|_{UV}\simeq \frac{mN}{2\pi ^{2}}\int_{x_{-}}^{x_{+}}\frac{dx}{x}\frac{\zeta ^{2}x^{2}}{4}\ln \frac{\rho _{\max }\left( x\right) }{m}. \label{condeninvUV}\tag{107}\] We denote \(L\equiv \Lambda /m\). Since \(\rho _{\max }\left( x\right) =\sqrt{L^{2}x\left( 1-x\right) -m^{2}}\),\[\ln \frac{\rho _{\max }\left( x\right) }{m}=\frac{1}{2}\ln \left( L^{2}x\left( 1-x\right) -m^{2}\right) . \label{lnromax}\tag{108}\] The subtle NLO effect (next-to-leading order (NLO)) arises because the invariant-mass cutoff excludes the endpoint regions \(x\rightarrow 0,1\): x\(\in \left[ x_{-},x_{+}\right]\) with \(x_{-}\sim 1/L^{2}\). Keeping that exact and expanding for \(L\gg 1\), one finds\[\int_{x_{-}}^{x_{+}}dxx\ln \frac{\rho _{\max }\left( x\right) }{m}=\frac{1}{2}\ln \frac{\Lambda }{m}-\frac{1}{2}-\frac{m^{2}}{\Lambda ^{2}}\ln \frac{\Lambda }{m}+\mathcal{O}\left( \frac{m^{2}}{\Lambda ^{2}}\right) . \label{intxdxlnromaxm}\tag{109}\] Substituting this into the condensate yields:\[\langle \langle 0|\colon \bar{\psi}\psi \colon |0\rangle _{\mathcal{A}}\rangle _{\varphi }^{inv}=\frac{mg^{2}\kappa }{4\pi ^{2}}\left[ \ln \frac{\Lambda }{m}-1-2\frac{m^{2}}{\Lambda ^{2}}\ln \frac{\Lambda }{m}+\mathcal{O}\left( \frac{m^{2}}{\Lambda ^{2}}\right) \right] +\left( finite\;IR\;piece\right) . \label{condenexterinv}\tag{110}\]

Let’s now examine the hard UV cutoff. When \(C_{ij}=\tau \delta _{ij}\), \(\kappa ^{\mu }=\left( \frac{\varpi }{2},0,0,\frac{\varpi }{2}\right)\) \(\Rightarrow\) \(\kappa ^{+}=\varpi\), \(\kappa ^{2}=0\), and transverse polarization, \(\sqrt{\left( \varepsilon ^{a}\cdot p\right) \left( \varepsilon _{a}\cdot p\right) }=\sqrt{\tau }|p_{\perp }|\), we can rewrite (83 ) as\[\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }=2imN\int \frac{d^{4}p}{\left( 2\pi \right) ^{4}}\frac{1-J_{0}\left( a\left( p\right) \right) }{p^{2}-m^{2}+i\epsilon }, \label{condensUVcutoff}\tag{111}\] where\[a\left( p\right) =\frac{2g\sqrt{\tau }}{p\kappa }\sqrt{\frac{1}{2N}}|p_{\perp }|. \label{a40p41}\tag{112}\] Let’s point out that\[p^{2}-m^{2}+i\epsilon =\left( p^{0}\right) ^{2}-\mathbf{p}^{2}-m^{2}+i\epsilon =\left( p^{0}\right) ^{2}-E_{\mathbf{p}}^{2}+i\epsilon ,\;E_{\mathbf{p}}\equiv \sqrt{\mathbf{p}^{2}+m^{2}}. \label{Ep}\tag{113}\] The standard identity holds as long as we consider the background factor \(J_{0}\left( a\left( p\right) \right)\) to be evaluated at the pole (i.e., \(p^{0}\rightarrow E_{\mathbf{p}}\) in the vacuum contraction):z\[\int \frac{dp^{0}}{2\pi }\frac{i}{\left( p^{0}\right) ^{2}-E_{\mathbf{p}}^{2}+i\epsilon }=\frac{1}{2E_{\mathbf{p}}}. \label{integralpzro}\tag{114}\] So we get\[\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }=mN\int \frac{d^{3}p}{\left( 2\pi \right) ^{3}}\frac{1}{E_{\mathbf{p}}}\left[ 1-J_{0}\left( a_{on}\left( \mathbf{p}\right) \right) \right] , \label{condensUVcutoffd3p}\tag{115}\] where \(a_{on}\left( \mathbf{p}\right)\) is \(a_{on}\left( p\right)\) evaluated at the on-shell energy \(p^{0}=E_{\mathbf{p}}\). With \(k^{\mu }=\left( \frac{\varpi }{2},0,0,\frac{\varpi }{2}\right)\), \(p\cdot k=\frac{\varpi }{2}\left( p^{0}-p^{3}\right)\). Hence,\[a_{on}\left( \mathbf{p}\right) =\beta \frac{|p_{\perp }|}{E_{\mathbf{p}}-p^{3}}. \label{aonp}\tag{116}\] where \(\beta =\frac{4g\sqrt{\tau }}{\varpi }\sqrt{\frac{1}{2N}}\frac{|p_{\perp }|}{E_{\mathbf{p}}-p^{3}}\). Note that the would-be collinear singular point \(E_{\mathbf{p}}-p^{3}\rightarrow 0\) only occurs when \(p_{\perp }\rightarrow 0\) and \(m\rightarrow 0\); for \(m>0\), \(E_{\mathbf{p}}-p^{3}=\left( m^{2}+p_{\perp }^{2}\right) /\left( E_{\mathbf{p}}+p^{3}\right)\), and the ratio remains finite as \(p_{\perp }\rightarrow 0\). Let \(p^{3}\equiv p_{z}\), \(|p_{\perp }|\equiv \rho\), so \(d^{3}p=2\pi \rho d\rho dp_{z}\) (in cylindrical coordinates), and the spherical cutoff is \(\rho ^{2}+p_{z}^{2}<\Lambda ^{2}\). Then\[\begin{align} &&\langle \langle 0|\bar{\psi}\psi |0\rangle _{\mathcal{A-}free}\rangle _{\varphi }^{\left( |\mathbf{p}|<\Lambda \right) } \notag \\ &=&\frac{mN}{4\pi ^{2}}\int_{-\Lambda }^{\Lambda }dp_{z}\int_{0}^{\sqrt{\Lambda ^{2}-p_{z}^{2}}}\frac{\rho d\rho }{\sqrt{\rho ^{2}+p_{z}^{2}+m^{2}}}\left[ 1-J_{0}\left( \beta \frac{\rho }{\sqrt{\rho ^{2}+p_{z}^{2}+m^{2}}-p_{z}}\right) \right] . \label{condensateUVcutoff} \end{align}\tag{117}\]

The condensate is influenced by collinear and light-front regions. This is why light-front cutoffs, invariant-mass regulators, and covariant DR all give different finite parts. This property makes the object a good way to test the structure of the light-front vacuum. The condensate shows how a coherent gluonic background changes the fermion vacuum, just as the Euler-Heisenberg effective action in QED and fermion polarization in a laser background do. In this case, though, the background is non-Abelian, color comes in through \(\theta\), and the group structure affects the outcome through \(N\).

6 Discussion and applications↩︎

In our paper, we used the exact Green function (16 ) for fermions, where the interaction with the external plane-wave background \(\mathcal{A}_{\mu }^{a}\left( \varphi \right)\) proceeds to all orders by the dressing matrix \(U\left( p,\varphi ,\varphi ^{\prime }\right)\). Gauge-covariant coupling to the background is guaranteed at all orders since all interactions with the classical field are represented in \(U\). Loop corrections are still described in powers of \(g\), but with background effects considered exactly instead than as external insertions.

All dressed quantities (propagator, vertex, self-energy, and condensate) have the same Floquet structure for a periodic plane-wave background. This appears in the vertex as 49 . This demonstrates that: Discrete quanta of momentum \(n\kappa\) are exchanged through the background. Vertex corrections, mass shifts, and condensate oscillations are all governed by the same harmonic selection rules. In this sense, the background acts as a coherent medium rather than as a perturbative external insertion.

We have derived the full one-loop, on-shell fermion mass shift in an external Yang–Mills plane-wave background within the axial gauge, employing the Mandelstam–Leibbrandt prescription to control spurious gauge singularities. The final result is 63 , which encapsulates, in a gauge-consistent manner, both vacuum renormalization and finite background-induced effects. An important aspect of the result is the essential difference between the axial-completion terms \(\delta m_{(2+3)}\) and the Feynman-like contribution \(\delta m_{(1)}\). While each term depends on the gauge-defining vectors \(n^{\mu }\) and \(n^{\ast \mu }\), their sum yields a well-defined pole mass shift. After cycle averaging over the plane-wave phases, all terms linear in the background field vanish. The leading physical correction arises at quadratic order in the background amplitude and can be written, in the small-field/high-energy limit, as \[\delta m=\delta m_{free}\left[ 1+\frac{g^{2}}{4N}\frac{\left( \varepsilon ^{a}\cdot p\right) \left( \varepsilon _{a}\cdot p\right) }{\left( p\cdot \kappa \right) ^{2}}+\mathcal{O}\left( \varepsilon ^{4}\right) \right] . \label{deltamdeltavac}\tag{118}\] The correction is suppressed by the invariant \(\left( p\cdot \kappa \right) ^{-2}\) and weighted by the color-summed polarization tensor \(\varepsilon _{\mu }^{a}\varepsilon _{\nu a}\), reflecting the non-Abelian nature of the background. The finite, background-dependent correction shows an interaction with a classical color field, modifying the fermion’s propagation. The correction is finite and kinematically suppressed.

The explicit axial-gauge formulation and the division into vacuum and background terms make the result especially suitable for light-front quantization. The calculation extends the Volkov solution and associated effective mass concept from QED to non-Abelian gauge theories, offering a systematic framework for studying fermions in coherent Yang–Mills backgrounds beyond perturbation theory in the field amplitude. The background makes finite oscillatory corrections (the phase structure \(\theta \left( p,\varphi \right)\) in \(U\)). We would like to emphasize that this feature is crucial because background effects only show up in real, quantifiable quantities, not in counterterms, and all combined operators constructed from the dressed propagator are appropriately adjusted. The exact propagator, effective mass, and vertex together form a closed, consistent set of building blocks for future studies of quantum processes in strong, coherent non-Abelian fields. We can use this unified framework to directly study scattering and radiation processes in coherent Yang–Mills fields, spin-dependent and polarization-sensitive observables, effective-field-theory matching in external gauge backgrounds, and systematic extensions to higher loops or non-equilibrium scenarios.

Our results are relevant to studying heavy-ion collisions, non-Abelian Schwinger pair production, and early-universe cosmology involving strong gauge fields [16], [17].

7 Acknowledgement↩︎

This work was partially supported by the grant No. 25RG-1C157 of the Higher Education and Science Committee of the Ministry of Education, Science, Culture and Sport RA.

References↩︎