Introduction
Relativistic heavy ion collisions provide a unique opportunity to study strong interaction matter, namely the quark gluon plasma (QGP), under extreme conditions. Strong electromagnetic (EM) fields are created in high energy collisions. In peripheral collisions, the dominant component is the magnetic field along the direction of the reaction plane [1–4]. In Au+Au collisions at s N N = 200 \sqrt{s_{\mathrm{NN}}}=200 s NN = 200 GeV, the highest collision energy at the Relativistic Heavy Ion Collider (RHIC) of Brookhaven National Laboratory, the magnetic field at the moment after the collision can reach 10 17 10^{17} 1 0 17 -10 18 10^{18} 1 0 18 Gauss [1–5]. In Pb+Pb collisions with s N N = 2.76 \sqrt{s_{\mathrm{NN}}}=2.76 s NN = 2.76 TeV at the Large Hadron Collider (LHC) of the European Organization for Nuclear Research (CERN), the magnetic field can reach 10 times as high as that at RHIC. In the early stage of the collision, the magnetic field decays with time as ∼ t − 3 \sim t^{-3} ∼ t − 3 . However, the finite Ohm conductivity of the QGP slows down the damping of the magnetic field in the later stage, resulting in a ∼ t − 1 \sim t^{-1} ∼ t − 1 behavior [6–16]. This may significantly extend the lifetime of the magnetic field, so one can expect sizeable effects from the interaction between the magnetic field and the QGP. These effects include the charge-odd directed flow v 1 o d d v_1^{\mathrm{odd}} v 1 odd induced by the Lorentz force acting on charged particles [18 , 19 ] [7, 11, 17], the splitting of Λ \Lambda Λ and Λ ˉ \bar{\Lambda} Λ ˉ hyperons’ global polarizations induced by the coupling between the magnetic field and hyperon’s magnetic moment [20 , 21 , 22 , 23 , 24 ] , and anomalous transport phenomena including the chiral magnetic effect (CME) [25 , 26 , 27 , 28 , 29 ] , the chiral separation effect (CSE) [30 , 31 , 32 ] , and the chiral magnetic wave (CMW) [33 , 34 ] , see Refs. [35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 ] for reviews. Study of strong interaction matter in strong magnetic fields may also help us understand compact stars such as magnetars [43 , 44 , 45 , 46 ] , in which strong magnetc fields and QGP coexist in the cores of compact stars.
In strong magnetic fields, quark spectrum will be described by Landau energy levels [47 ] . As a result, the transport properties of the QGP will be affected when the field strength is much larger than the energy scale of the system such as the temperature or chemical potential. In past few years, many studies have been made on magnetic field effects on transport quantities such as shear and bulk viscosities [48 , 49 , 50 , 51 , 52 , 53 ] , diffusion constants [48 , 51 , 52 , 53 , 54 , 55 ] , jet quenching parameters [56 , 57 , 58 ] , and electrical conductivities [51 , 55 , 59 , 60 , 61 , 62 , 63 ] . There are also some effects of strong magnetic fields on photon emission in heavy-ion collisions [64 , 65 , 66 , 67 ] . Among all these quantities, electrical conductivities are of special importance because they directly control the lifetime of magnetic fields and are also necessary parameters in the numerical simulation of magnetohydrodynamics [68 , 69 ] .
When the strength of the magnetic field is sufficiently large, the energy gap between the lowest Landau level and higher Landau levels becomes so large that nearly all fermions occupy the lowest Landau level. As a consequence, the system is greatly simplified to a ( 1 + 1 ) (1+1) ( 1 + 1 ) -dimension system from a ( 3 + 1 ) (3+1) ( 3 + 1 ) -dimension system, as the transverse motion (perpendicular to the direction of magnetic field) of fermions is restricted [47 ] . Electrical conductivities in the approximation of the lowest Landau level have been calculated in Ref. [59 , 60 ] with quantum field theory at finite temperature, and in Ref. [63 ] with chiral kinetic theory. On the other hand, conductivities in weak magnetic fields have been studied in the kinetic approach [55 , 61 ] , holograph models [70 , 71 ] , and the Lattice QCD [72 , 73 ] . In the medium magnetic field, one has to sum over all Landau levels. Electrical conductivities in this region have been computed by solving the linearized Boltzmann equations with explicit 1 ↔ 2 1 \leftrightarrow2 1 ↔ 2 processes in Ref. [62 ] and by employing the relaxation time approximation [74 ] and the holographic QCD model [75 , 76 ] .
The present work is based on covariant Wigner functions [77 , 78 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 86 ] , for recent reviews, see e.g. Refs. [42 , 87 ] . In the constant magnetic field, the Wigner function can be analytically derived using the exact solution to the Dirac equation [88 , 89 ] , which can serve as a tool to calculate quantities such as the pair production [90 ] and chiral susceptibility [91 ] . For particle collisions or interaction among particles in the Wigner function approach, we apply the relaxation time approximation as in Ref. [92 , 93 ] .
This paper is organized as follows. In Sec. II, we give an introduction to covariant Wigner functions and kinetic equations under the relaxation time approximation. In Sec. III, we discuss Wigner functions in equilibrium to give analytical solutions and equilibrium conditions. Section IV is devoted to the kinetic equation and solution for the off-equilibrium part. One can show that the off-equilibrium part of the Wigner function is fully determined by its vector and axial vector components. We focus on these two components in § 0.5 and § 0.6 , respectively. Analytical solutions are given when perturbative fields only depend on the proper time. The Ohm and Hall conductivities are studied in § 0.5 in respect to their dependence on the evolution and relaxation time, the particle mass, and the strength of the background magnetic field. We make conclusions in the last section, § 0.7 .
Throughout this work, we choose the convention for the metric tensor g μ ν = diag { + , − , − , − } g_{\mu\nu} = \operatorname{diag}\{+,-,-,-\} g μν = diag { + , − , − , − } and the Levi-Civita tensor ϵ μ ν ρ σ \epsilon^{\mu\nu\rho\sigma} ϵ μν ρ σ , ϵ 0123 = − ϵ 0123 = + 1 \epsilon^{0123} = -\epsilon_{0123} = +1 ϵ 0123 = − ϵ 0123 = + 1 . The local collective velocity u μ = γ ( 1 , v ) u^\mu= \gamma(1,\mathbf{v}) u μ = γ ( 1 , v ) satisfies u μ u μ = 1 u^\mu u_\mu= 1 u μ u μ = 1 with γ \gamma γ being the Lorentz contraction factor. Relative to u μ u^\mu u μ , we define the orthogonal projector Δ μ ν = g μ ν − u μ u ν \Delta^{\mu\nu} = g^{\mu\nu} - u^\mu u^\nu Δ μν = g μν − u μ u ν . The electric charge q q q is set to 1 throughout this paper for simplicity.
In this paper, we consider a weak perturbation field f μ ν f^{\mu\nu} f μν in additional to a strong background field F 0 μ ν F_0^{\mu\nu} F 0 μν . The background field is taken as a constant magnetic field in the comoving frame with respect to the velocity u μ u^\mu u μ , which is given by
F 0 μ ν = B 0 ϵ μ ν α β u α b β , (1) F_0^{\mu\nu} = B_0 \epsilon^{\mu\nu\alpha\beta}u_\alpha b_\beta, \tag*{(1)} F 0 μν = B 0 ϵ μν α β u α b β , (1)
where B 0 B_0 B 0 is the background field’s strength with b μ b^\mu b μ being its direction which is orthogonal to u μ u^\mu u μ , u μ b μ = 0 u_\mu b^\mu= 0 u μ b μ = 0 . We define another projection tensor
Ξ μ ν ≡ g μ ν − u μ u ν + b μ b ν . (2) \Xi^{\mu\nu} \equiv g^{\mu\nu} - u^\mu u^\nu+ b^\mu b^\nu. \tag*{(2)} Ξ μν ≡ g μν − u μ u ν + b μ b ν . (2)
which can project a four-vector to the direction perpendicular to both u μ u^\mu u μ and b μ b^\mu b μ .
Covariant Wigner functions
The Wigner function is defined as the Fourier transformation of the two-point correlation function [77 , 78 ]
W α β ( x , p ) = ∫ d 4 y ( 2 π ℏ ) 4 exp ( − i ℏ p ⋅ y ) ⟨ : ψ ˉ β ( x 1 ) U ( x 1 , x 2 ) ψ α ( x 2 ) : ⟩ , (3) W_{\alpha\beta}(x,p) = \int\frac{d^4y}{(2\pi\hbar)^4} \exp\left(-\frac{i}{\hbar}p\cdot y\right) \left\langle:\bar{\psi}_{\beta}(x_1)U(x_1,x_2)\psi_{\alpha}(x_2):\right\rangle, \tag*{(3)} W α β ( x , p ) = ∫ ( 2 π ℏ ) 4 d 4 y exp ( − ℏ i p ⋅ y ) ⟨ : ψ ˉ β ( x 1 ) U ( x 1 , x 2 ) ψ α ( x 2 ) : ⟩ , (3)
where ⟨ : O : ⟩ \langle: O :\rangle ⟨ : O : ⟩ denotes the expectation value of a normal-ordered operator O O O . Here x 1 x_1 x 1 and x 2 x_2 x 2 are two space-time coordinates, with y μ = x 1 μ − x 2 μ y^\mu= x_1^\mu- x_2^\mu y μ = x 1 μ − x 2 μ and x μ = ( x 1 μ + x 2 μ ) / 2 x^\mu= (x_1^\mu+ x_2^\mu)/2 x μ = ( x 1 μ + x 2 μ ) /2 being the relative and central coordinate respectively. With the vector potential A μ A_\mu A μ for the classical electromagnetic field, the gauge link in Eq. (3) is defined as
U ( x 1 , x 2 ) = exp [ − i ℏ y μ ∫ − 1 / 2 1 / 2 d t A μ ( x + t y ) ] . (4) U(x_1,x_2) = \exp\left[-\frac{i}{\hbar}y^\mu\int_{-1/2}^{1/2}dt\,A_\mu(x+ty)\right]. \tag*{(4)} U ( x 1 , x 2 ) = exp [ − ℏ i y μ ∫ − 1/2 1/2 d t A μ ( x + t y ) ] . (4)
The free-streaming Dirac field satisfies the Dirac equation and its adjoint form
( i ℏ γ ⋅ D − m ) ψ = ψ ˉ ( i ℏ γ ⋅ D † + m ) = 0 , (5) (i\hbar\gamma\cdot D-m)\psi=\bar{\psi}(i\hbar\gamma\cdot D^\dagger+m)=0, \tag*{(5)} ( i ℏ γ ⋅ D − m ) ψ = ψ ˉ ( i ℏ γ ⋅ D † + m ) = 0 , (5)
where D μ = ∂ x μ + ( i / ℏ ) A μ D_\mu= \partial_{x^\mu} + (i/\hbar)A_\mu D μ = ∂ x μ + ( i /ℏ ) A μ is the covariant derivative. The kinetic equation for the Wigner function has a similar form
( γ ⋅ K − m ) W ( x , p ) = 0 , (6) (\gamma\cdot K-m)W(x,p)=0, \tag*{(6)} ( γ ⋅ K − m ) W ( x , p ) = 0 , (6)
which can be derived from the Dirac equation. The operator K μ K^{\mu} K μ is given by
K μ = Π μ + i ℏ 2 ∇ μ , (7) K^{\mu} = \Pi^{\mu} + \frac{i\hbar}{2}\nabla^{\mu}, \tag*{(7)} K μ = Π μ + 2 i ℏ ∇ μ , (7)
where the generalized momentum operator Π μ \Pi^{\mu} Π μ and space-time derivative ∇ μ \nabla^{\mu} ∇ μ are
Π μ = p μ − ℏ 2 j 1 ( Δ ) F μ ν ( x ) ∂ p ν , ∇ μ = ∂ x μ − j 0 ( Δ ) F μ ν ( x ) ∂ p ν , (8) \begin{aligned}
\Pi^{\mu} &= p^{\mu} - \frac{\hbar}{2}j_{1}(\Delta)F^{\mu\nu}(x)\partial_{p^{\nu}},\\
\nabla^{\mu} &= \partial_{x}^{\mu} - j_{0}(\Delta)F^{\mu\nu}(x)\partial_{p^{\nu}}, \tag*{(8)}
\end{aligned} Π μ ∇ μ = p μ − 2 ℏ j 1 ( Δ ) F μν ( x ) ∂ p ν , = ∂ x μ − j 0 ( Δ ) F μν ( x ) ∂ p ν , (8)
with j 0 ( Δ ) j_{0}(\Delta) j 0 ( Δ ) and j 1 ( Δ ) j_{1}(\Delta) j 1 ( Δ ) are spherical Bessel functions and Δ = ( ℏ / 2 ) ∂ p ν ∂ x ν \Delta= (\hbar/2)\partial_{p}^{\nu}\partial_{x\nu} Δ = ( ℏ/2 ) ∂ p ν ∂ xν . We note that the derivative ∂ x ν \partial_{x\nu} ∂ xν in Δ \Delta Δ only acts on F μ ν ( x ) F^{\mu\nu}(x) F μν ( x ) but not on the Wigner function.
Concerning interaction among particles, we use the relaxation time approximation in addition to Eq. (6) [92 , 93 ] ,
( γ ⋅ K − m ) W ( x , p ) = − i ℏ 2 γ μ u μ δ W ( x , p ) τ , (9) (\gamma\cdot K - m)W(x,p) = -\frac{i\hbar}{2}\gamma^{\mu}u_{\mu}\frac{\delta W(x,p)}{\tau}, \tag*{(9)} ( γ ⋅ K − m ) W ( x , p ) = − 2 i ℏ γ μ u μ τ δ W ( x , p ) , (9)
where τ \tau τ is the relaxation time, δ W \delta W δ W represents the deviation from thermodynamical equilibrium
δ W ( x , p ) ≡ W ( x , p ) − W e q ( x , p ) , (10) \delta W(x,p) \equiv W(x,p) - W_{\mathrm{eq}}(x,p), \tag*{(10)} δ W ( x , p ) ≡ W ( x , p ) − W eq ( x , p ) , (10)
and u μ u_{\mu} u μ is the local collective velocity
u μ ∝ ∫ d 4 p Tr [ γ μ W e q ( x , p ) ] . (11) u_{\mu} \propto\int d^{4}p\operatorname{Tr}[\gamma^{\mu}W_{\mathrm{eq}}(x,p)]. \tag*{(11)} u μ ∝ ∫ d 4 p Tr [ γ μ W eq ( x , p )] . (11)
In this work, the equilibrium Wigner function is taken as the solution in a constant F 0 μ ν F_{0}^{\mu\nu} F 0 μν to the following equation
{ γ μ [ p μ + i ℏ 2 ( ∂ x μ − F 0 μ ν ∂ p ν ) ] − m } W e q ( x , p ) = 0. (12) \left\{\gamma_{\mu}\left[p^{\mu}+\frac{i\hbar}{2}\left(\partial_{x}^{\mu}-F_{0}^{\mu\nu}\partial_{p^{\nu}}\right)\right]-m\right\}W_{\mathrm{eq}}(x,p)=0. \tag*{(12)} { γ μ [ p μ + 2 i ℏ ( ∂ x μ − F 0 μν ∂ p ν ) ] − m } W eq ( x , p ) = 0. (12)
Then from Eq. (9) we obtain the equation for δ W \delta W δ W
{ γ μ p μ + i ℏ 2 γ μ ( ∂ x μ − F 0 μ ν ∂ p ν + 1 τ u μ ) − m } δ W ( x , p ) = i ℏ 2 γ μ F 0 μ ν ∂ p ν W e q + O ( ℏ 2 ) , (13) \left\{\gamma_{\mu}p^{\mu}+\frac{i\hbar}{2}\gamma_{\mu}\left(\partial_{x}^{\mu}-F_{0}^{\mu\nu}\partial_{p^{\nu}}+\frac{1}{\tau}u^{\mu}\right)-m\right\}\delta W(x,p)=\frac{i\hbar}{2}\gamma_{\mu}F_{0}^{\mu\nu}\partial_{p^{\nu}}W_{\mathrm{eq}}+O(\hbar^{2}), \tag*{(13)} { γ μ p μ + 2 i ℏ γ μ ( ∂ x μ − F 0 μν ∂ p ν + τ 1 u μ ) − m } δ W ( x , p ) = 2 i ℏ γ μ F 0 μν ∂ p ν W eq + O ( ℏ 2 ) , (13)
where we have made a semi-classical expansion with respect to ℏ \hbar ℏ and truncate it at second order in ℏ \hbar ℏ .
Wigner functions in constant magnetic field: equilibrium
In this section, we discuss about the equilibrium form of the Wigner function W e q ( x , p ) W_{\mathrm{eq}}(x,p) W eq ( x , p ) as the solution to Eq. (12) in the constant magnetic field (1) without interaction among particles.
The Wigner function satisfies W † = γ 0 W γ 0 W^{\dagger}=\gamma^{0}W\gamma^{0} W † = γ 0 W γ 0 and therefore it can be expanded in terms of the generators of Clifford algebra Γ i = { 1 4 , i γ 5 , γ μ , γ 5 γ μ , ( 1 / 2 ) σ μ ν } \Gamma_i=\{1_4,i\gamma^5,\gamma^\mu,\gamma^5\gamma^\mu,(1/2)\sigma^{\mu\nu}\} Γ i = { 1 4 , i γ 5 , γ μ , γ 5 γ μ , ( 1/2 ) σ μν } as
W ( x , p ) = 1 4 ( 1 4 F + i γ 5 P + γ μ V μ + γ 5 γ μ A μ + 1 2 σ μ ν S μ ν ) (14) W(x,p)=\frac{1}{4}\left(1_4\mathcal{F}+i\gamma^5\mathcal{P}+\gamma^\mu\mathcal{V}_\mu+\gamma^5\gamma^\mu\mathcal{A}_\mu+\frac{1}{2}\sigma^{\mu\nu}\mathcal{S}_{\mu\nu}\right) \tag*{(14)} W ( x , p ) = 4 1 ( 1 4 F + i γ 5 P + γ μ V μ + γ 5 γ μ A μ + 2 1 σ μν S μν ) (14)
where σ μ ν = ( i / 2 ) [ γ μ , γ ν ] \sigma^{\mu\nu}=(i/2)[\gamma^\mu,\gamma^\nu] σ μν = ( i /2 ) [ γ μ , γ ν ] , and all expansion coefficients { F , P , V μ , A μ , S μ ν } \{\mathcal{F},\mathcal{P},\mathcal{V}^{\mu},\mathcal{A}^{\mu},\mathcal{S}^{\mu\nu}\} { F , P , V μ , A μ , S μν } are real functions in phase space. Each function contains an equilibrium and an off-equilibrium part, for example,
F = F e q + δ F . (15) \mathcal{F}=\mathcal{F}_{\mathrm{eq}}+\delta\mathcal{F}. \tag*{(15)} F = F eq + δ F . (15)
So W e q ( x , p ) W_{\mathrm{eq}}(x,p) W eq ( x , p ) also gives its equilibrium components { F e q , P e q , V e q μ , A e q μ , S e q μ ν } \{\mathcal{F}_{\mathrm{eq}},\mathcal{P}_{\mathrm{eq}},\mathcal{V}^{\mu}_{\mathrm{eq}},\mathcal{A}^{\mu}_{\mathrm{eq}},\mathcal{S}^{\mu\nu}_{\mathrm{eq}}\} { F eq , P eq , V eq μ , A eq μ , S eq μν } similarly as in Eq. (14) .
Projecting Eq. (12) onto Γ i \Gamma_i Γ i and decoupling real and imaginary parts, we obtain the following set of partial differential equations
p μ V e q μ − m F e q = 0 , (16) p_\mu\mathcal{V}^{\mu}_{\mathrm{eq}}-m\mathcal{F}_{\mathrm{eq}}=0, \tag*{(16)} p μ V eq μ − m F eq = 0 , (16)
ℏ 2 ( ∂ x ν − F 0 ν α ∂ p α ) A μ e q + m P e q = 0 , (17) \frac{\hbar}{2}\left(\partial_x^\nu-F_0^{\nu\alpha}\partial_{p\alpha}\right)\mathcal{A}^{\mathrm{eq}}_\mu+m\mathcal{P}_{\mathrm{eq}}=0, \tag*{(17)} 2 ℏ ( ∂ x ν − F 0 ν α ∂ p α ) A μ eq + m P eq = 0 , (17)
p μ F e q − ℏ 2 ( ∂ x ν − F 0 ν α ∂ p α ) S ν μ e q − m V μ e q = 0 , (18) p_\mu\mathcal{F}_{\mathrm{eq}}-\frac{\hbar}{2}\left(\partial_x^\nu-F_0^{\nu\alpha}\partial_{p\alpha}\right)\mathcal{S}^{\mathrm{eq}}_{\nu\mu}-m\mathcal{V}^{\mathrm{eq}}_\mu=0, \tag*{(18)} p μ F eq − 2 ℏ ( ∂ x ν − F 0 ν α ∂ p α ) S ν μ eq − m V μ eq = 0 , (18)
− ℏ 2 ( ∂ x μ − F 0 , μ ν ∂ p ν ) P e q + 1 2 ϵ μ ν α β p ν S e q α β + m A μ e q = 0 , (19) -\frac{\hbar}{2}\left(\partial_{x\mu}-F_{0,\mu\nu}\partial_p^\nu\right)\mathcal{P}_{\mathrm{eq}}+\frac{1}{2}\epsilon_{\mu\nu\alpha\beta}p^\nu\mathcal{S}^{\alpha\beta}_{\mathrm{eq}}+m\mathcal{A}^{\mathrm{eq}}_\mu=0, \tag*{(19)} − 2 ℏ ( ∂ xμ − F 0 , μν ∂ p ν ) P eq + 2 1 ϵ μν α β p ν S eq α β + m A μ eq = 0 , (19)
ℏ 2 ( ∂ x μ − F 0 , μ α ∂ p α ) V ν e q − ℏ 2 ( ∂ x ν − F 0 , ν α ∂ p α ) V μ e q − ϵ μ ν α β p α A e q β − m S μ ν e q = 0 , (20) \begin{aligned}
\frac{\hbar}{2}\left(\partial_{x\mu}-F_{0,\mu\alpha}\partial_p^\alpha\right)\mathcal{V}^{\mathrm{eq}}_\nu
-\frac{\hbar}{2}\left(\partial_{x\nu}-F_{0,\nu\alpha}\partial_p^\alpha\right)\mathcal{V}^{\mathrm{eq}}_\mu\\
-\epsilon_{\mu\nu\alpha\beta}p^\alpha\mathcal{A}^{\beta}_{\mathrm{eq}}-m\mathcal{S}^{\mathrm{eq}}_{\mu\nu}=0, \tag*{(20)}
\end{aligned} 2 ℏ ( ∂ xμ − F 0 , μα ∂ p α ) V ν eq − 2 ℏ ( ∂ xν − F 0 , ν α ∂ p α ) V μ eq − ϵ μν α β p α A eq β − m S μν eq = 0 , (20)
ℏ ( ∂ x μ − F 0 μ ν ∂ p ν ) V μ e q = 0 , (21) \hbar\left(\partial_x^\mu-F_0^{\mu\nu}\partial_{p\nu}\right)\mathcal{V}^{\mathrm{eq}}_\mu=0, \tag*{(21)} ℏ ( ∂ x μ − F 0 μν ∂ p ν ) V μ eq = 0 , (21)
p μ A μ e q = 0 , (22) p^\mu\mathcal{A}^{\mathrm{eq}}_\mu=0, \tag*{(22)} p μ A μ eq = 0 , (22)
ℏ 2 ( ∂ x μ − F 0 , μ ν ∂ p ν ) F e q + p ν S ν μ e q = 0 , (23) \frac{\hbar}{2}\left(\partial_{x\mu}-F_{0,\mu\nu}\partial_p^\nu\right)\mathcal{F}_{\mathrm{eq}}+p^\nu\mathcal{S}^{\mathrm{eq}}_{\nu\mu}=0, \tag*{(23)} 2 ℏ ( ∂ xμ − F 0 , μν ∂ p ν ) F eq + p ν S ν μ eq = 0 , (23)
p μ P e q + ℏ 4 ϵ μ ν α β ( ∂ x ν − F 0 ν ρ ∂ p ρ ) S e q α β = 0 , (24) p_\mu\mathcal{P}_{\mathrm{eq}}+\frac{\hbar}{4}\epsilon_{\mu\nu\alpha\beta}\left(\partial_x^\nu-F_0^{\nu\rho}\partial_{p\rho}\right)\mathcal{S}^{\alpha\beta}_{\mathrm{eq}}=0, \tag*{(24)} p μ P eq + 4 ℏ ϵ μν α β ( ∂ x ν − F 0 ν ρ ∂ pρ ) S eq α β = 0 , (24)
p [ μ V ν ] e q + ℏ 2 ϵ μ ν α β ( ∂ x α − F 0 α ρ ∂ p ρ ) A e q β = 0. (25) p_{[\mu}\mathcal{V}^{\mathrm{eq}}_{\nu]}+\frac{\hbar}{2}\epsilon_{\mu\nu\alpha\beta}\left(\partial_x^\alpha-F_0^{\alpha\rho}\partial_{p\rho}\right)\mathcal{A}^{\beta}_{\mathrm{eq}}=0. \tag*{(25)} p [ μ V ν ] eq + 2 ℏ ϵ μν α β ( ∂ x α − F 0 α ρ ∂ pρ ) A eq β = 0. (25)
With the help of Eqs. (16) , (17) , and (20) , we are able to express F e q \mathcal{F}_{\rm eq} F eq , P e q \mathcal{P}_{\rm eq} P eq , and S e q μ ν \mathcal{S}^{\mu\nu}_{\rm eq} S eq μν in terms of the remaining components V e q μ \mathcal{V}^{\mu}_{\rm eq} V eq μ and A e q μ \mathcal{A}^{\mu}_{\rm eq} A eq μ as
F e q = 1 m p μ V e q μ , \mathcal{F}_{\rm eq}=\frac{1}{m}p_{\mu}\mathcal{V}^{\mu}_{\rm eq}, F eq = m 1 p μ V eq μ ,
P e q = − ℏ 2 m ( ∂ x μ − F 0 μ ν ∂ p ν ) A μ e q , \mathcal{P}_{\rm eq}=-\frac{\hbar}{2m}\left(\partial^{\mu}_{x}-F_{0}^{\mu\nu}\partial_{p\nu}\right)\mathcal{A}^{\rm eq}_{\mu}, P eq = − 2 m ℏ ( ∂ x μ − F 0 μν ∂ p ν ) A μ eq ,
S μ ν e q = ℏ 2 m ( ∂ x μ − F 0 , μ α ∂ p α ) V ν e q − ℏ 2 m ( ∂ x ν − F 0 , ν α ∂ p α ) V μ e q − 1 m ϵ μ ν α β p α A e q β . (26) \mathcal{S}^{\rm eq}_{\mu\nu}=\frac{\hbar}{2m}\left(\partial_{x\mu}-F_{0,\mu\alpha}\partial_{p}^{\alpha}\right)\mathcal{V}^{\rm eq}_{\nu}-\frac{\hbar}{2m}\left(\partial_{x\nu}-F_{0,\nu\alpha}\partial_{p}^{\alpha}\right)\mathcal{V}^{\rm eq}_{\mu}-\frac{1}{m}\epsilon_{\mu\nu\alpha\beta}p^{\alpha}\mathcal{A}^{\beta}_{\rm eq}. \tag*{(26)} S μν eq = 2 m ℏ ( ∂ xμ − F 0 , μα ∂ p α ) V ν eq − 2 m ℏ ( ∂ xν − F 0 , ν α ∂ p α ) V μ eq − m 1 ϵ μν α β p α A eq β . (26)
Substituting above relations into Eqs. (23) and (24) , we obtain the kinetic equations for V e q μ \mathcal{V}^{\mu}_{\rm eq} V eq μ and A e q μ \mathcal{A}^{\mu}_{\rm eq} A eq μ ,
p ν ( ∂ x ν − F 0 , ν α ∂ p α ) V μ e q − F 0 , μ ν V e q ν = 0 , p^{\nu}\left(\partial_{x\nu}-F_{0,\nu\alpha}\partial_{p}^{\alpha}\right)\mathcal{V}^{\rm eq}_{\mu}-F_{0,\mu\nu}\mathcal{V}^{\nu}_{\rm eq}=0, p ν ( ∂ xν − F 0 , ν α ∂ p α ) V μ eq − F 0 , μν V eq ν = 0 ,
p ν ( ∂ x ν − F 0 , ν α ∂ p α ) A μ e q − F 0 , μ ν A e q ν = 0. (27) p^{\nu}\left(\partial_{x\nu}-F_{0,\nu\alpha}\partial_{p}^{\alpha}\right)\mathcal{A}^{\rm eq}_{\mu}-F_{0,\mu\nu}\mathcal{A}^{\nu}_{\rm eq}=0. \tag*{(27)} p ν ( ∂ xν − F 0 , ν α ∂ p α ) A μ eq − F 0 , μν A eq ν = 0. (27)
On the other hand, substituting F e q \mathcal{F}_{\rm eq} F eq , P e q \mathcal{P}_{\rm eq} P eq , and S e q μ ν \mathcal{S}^{\mu\nu}_{\rm eq} S eq μν into Eqs. (18) and (19) , we obtain the mass-shell constraints
0 = ( p 2 − m 2 ) V μ e q + ℏ 2 m ϵ μ ν α β F 0 ν α A e q β , 0=\left(p^{2}-m^{2}\right)\mathcal{V}^{\rm eq}_{\mu}+\frac{\hbar}{2m}\epsilon_{\mu\nu\alpha\beta}F_{0}^{\nu\alpha}\mathcal{A}^{\beta}_{\rm eq}, 0 = ( p 2 − m 2 ) V μ eq + 2 m ℏ ϵ μν α β F 0 ν α A eq β ,
0 = ( p 2 − m 2 ) A μ e q + ℏ 2 m ϵ μ ν α β F 0 ν α V e q β . (28) 0=\left(p^{2}-m^{2}\right)\mathcal{A}^{\rm eq}_{\mu}+\frac{\hbar}{2m}\epsilon_{\mu\nu\alpha\beta}F_{0}^{\nu\alpha}\mathcal{V}^{\beta}_{\rm eq}. \tag*{(28)} 0 = ( p 2 − m 2 ) A μ eq + 2 m ℏ ϵ μν α β F 0 ν α V eq β . (28)
which indicate that the particle’s mass-shell is modified by the magnetic field.
The Wigner function in a constant magnetic field has been given in Ref. [88 , 94 ] in a non-covariant form. It is straightforward to generalize the results to the covariant form,
V e q μ = ∑ n = 0 ∞ [ p μ − ( 1 − 2 n ℏ B 0 p T 2 ) Ξ μ ν p ν ] V n ( p ) Λ + ( n ) ( p T ) , \mathcal{V}^{\mu}_{\rm eq}=\sum_{n=0}^{\infty}\left[p^{\mu}-\left(1-\frac{2n\hbar B_{0}}{p_{T}^{2}}\right)\Xi^{\mu\nu}p_{\nu}\right]V_{n}(p)\Lambda_{+}^{(n)}(p_{T}), V eq μ = n = 0 ∑ ∞ [ p μ − ( 1 − p T 2 2 n ℏ B 0 ) Ξ μν p ν ] V n ( p ) Λ + ( n ) ( p T ) ,
A e q μ = [ ( u ⋅ p ) b μ − ( b ⋅ p ) u μ ] ∑ n = 0 ∞ V n ( p ) Λ − ( n ) ( p T ) , (29) \mathcal{A}^{\mu}_{\rm eq}=\left[(u\cdot p)b^{\mu}-(b\cdot p)u^{\mu}\right]\sum_{n=0}^{\infty}V_{n}(p)\Lambda_{-}^{(n)}(p_{T}), \tag*{(29)} A eq μ = [ ( u ⋅ p ) b μ − ( b ⋅ p ) u μ ] n = 0 ∑ ∞ V n ( p ) Λ − ( n ) ( p T ) , (29)
where b μ b^{\mu} b μ is the direction of the magnetic field, Ξ μ ν \Xi^{\mu\nu} Ξ μν is the projection operator in Eq. (2) , p T = − p α Ξ α β p β p_{T}=\sqrt{-p_{\alpha}\Xi^{\alpha\beta}p_{\beta}} p T = − p α Ξ α β p β is the magnitude of the transverse momentum, and Λ ± ( n ) ( p T ) \Lambda_{\pm}^{(n)}(p_{T}) Λ ± ( n ) ( p T ) are defined as
Λ ± ( n ) ( p T ) = ( − 1 ) n [ L n ( 2 p T 2 ℏ B 0 ) ∓ L n − 1 ( 2 p T 2 ℏ B 0 ) ] exp ( − p T 2 ℏ B 0 ) . (30) \Lambda_{\pm}^{(n)}(p_{T})=(-1)^{n}\left[L_{n}\left(\frac{2p_{T}^{2}}{\hbar B_{0}}\right)\mp L_{n-1}\left(\frac{2p_{T}^{2}}{\hbar B_{0}}\right)\right]\exp\left(-\frac{p_{T}^{2}}{\hbar B_{0}}\right). \tag*{(30)} Λ ± ( n ) ( p T ) = ( − 1 ) n [ L n ( ℏ B 0 2 p T 2 ) ∓ L n − 1 ( ℏ B 0 2 p T 2 ) ] exp ( − ℏ B 0 p T 2 ) . (30)
where L n ( x ) L_{n}(x) L n ( x ) with n ≥ 0 n\ge0 n ≥ 0 are Laguerre polynomials and L − 1 ( x ) = 0 L_{-1}(x)=0 L − 1 ( x ) = 0 . In Eq. (29) , V n ( p ) V_{n}(p) V n ( p ) is a function that only depends on u ⋅ p u\cdot p u ⋅ p and b ⋅ p b\cdot p b ⋅ p ,
V n ( p ) ≡ 2 ( 2 π ) 3 ( 2 − δ n 0 ) δ [ ( u ⋅ p ) 2 − ( b ⋅ p ) 2 − m 2 − 2 n ℏ B 0 ] { f e q ( + ) ( u ⋅ p ) θ ( u ⋅ p ) + f e q ( − ) ( − u ⋅ p ) θ ( − u ⋅ p ) } . (31) V_{n}(p)\equiv\frac{2}{(2\pi)^{3}}(2-\delta_{n0})\delta\left[(u\cdot p)^{2}-(b\cdot p)^{2}-m^{2}-2n\hbar B_{0}\right]\left\{f^{(+)}_{\rm eq}(u\cdot p)\theta(u\cdot p)+f^{(-)}_{\rm eq}(-u\cdot p)\theta(-u\cdot p)\right\}. \tag*{(31)} V n ( p ) ≡ ( 2 π ) 3 2 ( 2 − δ n 0 ) δ [ ( u ⋅ p ) 2 − ( b ⋅ p ) 2 − m 2 − 2 n ℏ B 0 ] { f eq ( + ) ( u ⋅ p ) θ ( u ⋅ p ) + f eq ( − ) ( − u ⋅ p ) θ ( − u ⋅ p ) } . (31)
where f e q ( ± ) f_{\mathrm{eq}}^{(\pm)} f eq ( ± ) are Fermi-Dirac distributions at local temperature T = 1 / β T=1/\beta T = 1/ β and chemical potential μ = α T \mu=\alpha T μ = α T ,
f e q ( ± ) ( u ⋅ p ) = 1 1 + exp [ β ( u ⋅ p ) ∓ α ] . (32) f_{\mathrm{eq}}^{(\pm)}(u\cdot p)=\frac{1}{1+\exp\left[\beta(u\cdot p)\mp\alpha\right]}. \tag*{(32)} f eq ( ± ) ( u ⋅ p ) = 1 + exp [ β ( u ⋅ p ) ∓ α ] 1 . (32)
The space-time dependence of V e q μ \mathcal{V}_{\mathrm{eq}}^\mu V eq μ and A e q μ \mathcal{A}_{\mathrm{eq}}^\mu A eq μ is through β ( x ) \beta(x) β ( x ) , u μ ( x ) u^\mu(x) u μ ( x ) , and μ ( x ) \mu(x) μ ( x ) .
It can be easily checked that V e q μ \mathcal{V}_{\mathrm{eq}}^\mu V eq μ and A e q μ \mathcal{A}_{\mathrm{eq}}^\mu A eq μ in Eq. (29) fulfill the mass-shell conditions in (28) , and V e q μ \mathcal{V}_{\mathrm{eq}}^\mu V eq μ and A e q μ \mathcal{A}_{\mathrm{eq}}^\mu A eq μ should also satisfy Eqs. (21) , (22) , (25) , and (27) , leading to following constraints
∂ x μ { [ p μ − ( 1 − 2 n ℏ B 0 p T 2 ) Ξ μ ν p ν ] V n ( p ) Λ + ( n ) ( p T ) } = 0 , ϵ μ ν α β ∂ x α { [ ( u ⋅ p ) b β − ( b ⋅ p ) u β ] V n ( p ) Λ − ( n ) ( p T ) } = 0 , p ν ∂ x ν { [ p μ − ( 1 − 2 n ℏ B 0 p T 2 ) Ξ μ ρ p ρ ] V n ( p ) Λ + ( n ) ( p T ) } = 0. (33) \begin{aligned}
\partial_{x^\mu}\left\{\left[p^\mu-\left(1-\frac{2n\hbar B_0}{p_T^2}\right)\Xi^{\mu\nu}p_\nu\right]V_n(p)\Lambda_+^{(n)}(p_T)\right\}&=0,\\
\epsilon_{\mu\nu\alpha\beta}\partial_x^\alpha\left\{\left[(u\cdot p)b^\beta-(b\cdot p)u^\beta\right]V_n(p)\Lambda_-^{(n)}(p_T)\right\}&=0,\\
p^\nu\partial_{x^\nu}\left\{\left[p^\mu-\left(1-\frac{2n\hbar B_0}{p_T^2}\right)\Xi^{\mu\rho}p_\rho\right]V_n(p)\Lambda_+^{(n)}(p_T)\right\}&=0. \tag*{(33)}
\end{aligned} ∂ x μ { [ p μ − ( 1 − p T 2 2 n ℏ B 0 ) Ξ μν p ν ] V n ( p ) Λ + ( n ) ( p T ) } ϵ μν α β ∂ x α { [ ( u ⋅ p ) b β − ( b ⋅ p ) u β ] V n ( p ) Λ − ( n ) ( p T ) } p ν ∂ x ν { [ p μ − ( 1 − p T 2 2 n ℏ B 0 ) Ξ μ ρ p ρ ] V n ( p ) Λ + ( n ) ( p T ) } = 0 , = 0 , = 0. (33)
Since Λ ± ( n ) ( p T ) \Lambda_\pm^{(n)}(p_T) Λ ± ( n ) ( p T ) are independent of space-time coordinates, the constraints in Eq. (33) can be cast into those for equilibrium distributions
[ p μ − ( 1 − 2 n ℏ B 0 p T 2 ) Ξ μ ν p ν ] ∂ x μ f e q ( ± ) ( x ; u ⋅ p ) − ( 1 − 2 n ℏ B 0 p T 2 ) ( ∂ x μ Ξ μ ν p ν ) f e q ( ± ) ( x ; u ⋅ p ) = 0 , (34) \left[p^\mu-\left(1-\frac{2n\hbar B_0}{p_T^2}\right)\Xi^{\mu\nu}p_\nu\right]\partial_{x^\mu}f_{\mathrm{eq}}^{(\pm)}(x;u\cdot p)-\left(1-\frac{2n\hbar B_0}{p_T^2}\right)(\partial_{x^\mu}\Xi^{\mu\nu}p_\nu)f_{\mathrm{eq}}^{(\pm)}(x;u\cdot p)=0, \tag*{(34)} [ p μ − ( 1 − p T 2 2 n ℏ B 0 ) Ξ μν p ν ] ∂ x μ f eq ( ± ) ( x ; u ⋅ p ) − ( 1 − p T 2 2 n ℏ B 0 ) ( ∂ x μ Ξ μν p ν ) f eq ( ± ) ( x ; u ⋅ p ) = 0 , (34)
ϵ μ ν α β ∂ x α { [ ( u ⋅ p ) b β − ( b ⋅ p ) u β ] f e q ( ± ) ( x ; u ⋅ p ) } = 0 , (35) \epsilon_{\mu\nu\alpha\beta}\partial_x^\alpha\left\{\left[(u\cdot p)b^\beta-(b\cdot p)u^\beta\right]f_{\mathrm{eq}}^{(\pm)}(x;u\cdot p)\right\}=0, \tag*{(35)} ϵ μν α β ∂ x α { [ ( u ⋅ p ) b β − ( b ⋅ p ) u β ] f eq ( ± ) ( x ; u ⋅ p ) } = 0 , (35)
[ p μ − ( 1 − 2 n ℏ B 0 p T 2 ) Ξ μ ρ p ρ ] p ν ∂ x ν f e q ( ± ) ( x ; u ⋅ p ) − ( 1 − 2 n ℏ B 0 p T 2 ) p ν ( ∂ x ν Ξ μ ρ p ρ ) f e q ( ± ) ( x ; u ⋅ p ) = 0. (36) \left[p^\mu-\left(1-\frac{2n\hbar B_0}{p_T^2}\right)\Xi^{\mu\rho}p_\rho\right]p^\nu\partial_{x^\nu}f_{\mathrm{eq}}^{(\pm)}(x;u\cdot p)-\left(1-\frac{2n\hbar B_0}{p_T^2}\right)p^\nu(\partial_{x^\nu}\Xi^{\mu\rho}p_\rho)f_{\mathrm{eq}}^{(\pm)}(x;u\cdot p)=0. \tag*{(36)} [ p μ − ( 1 − p T 2 2 n ℏ B 0 ) Ξ μ ρ p ρ ] p ν ∂ x ν f eq ( ± ) ( x ; u ⋅ p ) − ( 1 − p T 2 2 n ℏ B 0 ) p ν ( ∂ x ν Ξ μ ρ p ρ ) f eq ( ± ) ( x ; u ⋅ p ) = 0. (36)
Substituting f e q ( ± ) f_{\mathrm{eq}}^{(\pm)} f eq ( ± ) in Eq. (32) into these equations, we find that α \alpha α should be a space-time constant, and β u μ \beta u_\mu β u μ should satisfy the Killing condition, namely,
∂ x μ α = 0 , \partial_x^\mu\alpha=0, ∂ x μ α = 0 ,
∂ μ x ( β u ν ) + ∂ ν x ( β u μ ) = 0. (37) \partial_\mu^x(\beta u_\nu)+\partial_\nu^x(\beta u_\mu)=0. \tag*{(37)} ∂ μ x ( β u ν ) + ∂ ν x ( β u μ ) = 0. (37)
These are also conditions for global equilibrium without EM fields. In addition to the conditions in (37) , we also obtain
∂ b u μ = 0 , ∂ ⊥ ν u μ = 0 , \partial_b u^\mu=0,\qquad\partial_\perp^\nu u^\mu=0, ∂ b u μ = 0 , ∂ ⊥ ν u μ = 0 ,
∂ b b μ = 0 , ∂ ⊥ ν b μ = 0 , \partial_b b^\mu=0,\qquad\partial_\perp^\nu b^\mu=0, ∂ b b μ = 0 , ∂ ⊥ ν b μ = 0 ,
∂ ⊥ μ β = 0 , Ξ μ ν ∂ t b ν = 0. (38) \partial_\perp^\mu\beta=0,\qquad\Xi^\mu{}_\nu\partial_t b^\nu=0. \tag*{(38)} ∂ ⊥ μ β = 0 , Ξ μ ν ∂ t b ν = 0. (38)
where ∂ t ≡ u μ ∂ x μ \partial_t \equiv u_\mu\partial_x^\mu ∂ t ≡ u μ ∂ x μ , ∂ b ≡ − b μ ∂ x μ \partial_b \equiv-b_\mu\partial_x^\mu ∂ b ≡ − b μ ∂ x μ , and ∂ ⊥ μ ≡ Ξ ν μ ∂ x ν \partial_\perp^\mu\equiv\Xi^\mu_\nu\partial_x^\nu ∂ ⊥ μ ≡ Ξ ν μ ∂ x ν . These constraints are satisfied by that β \beta β , b μ b^\mu b μ , and u μ u^\mu u μ are all space-time constants.
General equations of Wigner functions: off-equilibrium
The off-equilibrium part of the Wigner function δ W \delta W δ W satisfies Eq. (13) . Similar to Eq. (14) , we can decompose δ W \delta W δ W in terms of Γ i \Gamma_i Γ i , then from Eq. (13) we derive the equations for Clifford components,
p μ δ V μ − m δ F = 0 , (39) p_\mu\delta V^\mu-m\delta F=0, \tag*{(39)} p μ δ V μ − m δ F = 0 , (39)
ℏ 2 ( ∂ x μ − F 0 μ ν ∂ p ν + 1 τ u μ ) δ A μ + m δ P = ℏ 2 f μ ν ∂ p ν A μ e q , (40) \frac{\hbar}{2}\left(\partial_x^\mu-F_0^{\mu\nu}\partial_{p\nu}+\frac{1}{\tau}u^\mu\right)\delta A_\mu+m\delta P=\frac{\hbar}{2}f^{\mu\nu}\partial_{p\nu}A_\mu^{\mathrm{eq}}, \tag*{(40)} 2 ℏ ( ∂ x μ − F 0 μν ∂ p ν + τ 1 u μ ) δ A μ + m δ P = 2 ℏ f μν ∂ p ν A μ eq , (40)
p μ δ F − ℏ 2 ( ∂ x ν − F 0 ν ρ ∂ p ρ + 1 τ u ν ) δ S ν μ − m δ V μ = ℏ 2 f ν ρ ∂ p ρ S μ ν e q , (41) p_\mu\delta F-\frac{\hbar}{2}\left(\partial_x^\nu-F_0^{\nu\rho}\partial_{p\rho}+\frac{1}{\tau}u^\nu\right)\delta S_{\nu\mu}-m\delta V_\mu=\frac{\hbar}{2}f^{\nu\rho}\partial_{p\rho}S_{\mu\nu}^{\mathrm{eq}}, \tag*{(41)} p μ δ F − 2 ℏ ( ∂ x ν − F 0 ν ρ ∂ pρ + τ 1 u ν ) δ S ν μ − m δ V μ = 2 ℏ f ν ρ ∂ pρ S μν eq , (41)
− ℏ 2 ( ∂ x μ − F 0 , μ ν ∂ p ν + 1 τ u μ ) δ P + 1 2 ϵ μ ν α β p ν δ S α β + m δ A μ = − ℏ 2 f μ ν ∂ p ν P e q , (42) -\frac{\hbar}{2}\left(\partial_{x\mu}-F_{0,\mu\nu}\partial_p^\nu+\frac{1}{\tau}u_\mu\right)\delta P+\frac{1}{2}\epsilon_{\mu\nu\alpha\beta}p^\nu\delta S^{\alpha\beta}+m\delta A_\mu=-\frac{\hbar}{2}f_{\mu\nu}\partial_p^\nu P^{\mathrm{eq}}, \tag*{(42)} − 2 ℏ ( ∂ xμ − F 0 , μν ∂ p ν + τ 1 u μ ) δ P + 2 1 ϵ μν α β p ν δ S α β + m δ A μ = − 2 ℏ f μν ∂ p ν P eq , (42)
ℏ 2 ( ∂ x μ − F 0 , μ ρ ∂ p ρ + 1 τ u μ ) δ V ν − ℏ 2 ( ∂ x ν − F 0 , ν ρ ∂ p ρ + 1 τ u ν ) δ V μ − ϵ μ ν α β p α δ A β − m δ S μ ν = ℏ 2 f μ ρ ∂ p ρ V ν e q − ℏ 2 f ν ρ ∂ p ρ V μ e q , (43) \begin{aligned}
\frac{\hbar}{2}\left(\partial_{x\mu}-F_{0,\mu\rho}\partial_p^\rho+\frac{1}{\tau}u_\mu\right)\delta V_\nu\\
-\frac{\hbar}{2}\left(\partial_{x\nu}-F_{0,\nu\rho}\partial_p^\rho+\frac{1}{\tau}u_\nu\right)\delta V_\mu-\epsilon_{\mu\nu\alpha\beta}p^\alpha\delta A^\beta-m\delta S_{\mu\nu}\\
=\frac{\hbar}{2}f_{\mu\rho}\partial_p^\rho V_\nu^{\mathrm{eq}}-\frac{\hbar}{2}f_{\nu\rho}\partial_p^\rho V_\mu^{\mathrm{eq}}, \tag*{(43)}
\end{aligned} 2 ℏ ( ∂ xμ − F 0 , μ ρ ∂ p ρ + τ 1 u μ ) δ V ν − 2 ℏ ( ∂ xν − F 0 , ν ρ ∂ p ρ + τ 1 u ν ) δ V μ − ϵ μν α β p α δ A β − m δ S μν = 2 ℏ f μ ρ ∂ p ρ V ν eq − 2 ℏ f ν ρ ∂ p ρ V μ eq , (43)
ℏ ( ∂ x μ − F 0 μ ν ∂ p ν + 1 τ u μ ) δ V μ = ℏ f μ ν ∂ p ν V μ e q , (44) \hbar\left(\partial_x^\mu-F_0^{\mu\nu}\partial_{p\nu}+\frac{1}{\tau}u^\mu\right)\delta V_\mu=\hbar f^{\mu\nu}\partial_{p\nu}V_\mu^{\mathrm{eq}}, \tag*{(44)} ℏ ( ∂ x μ − F 0 μν ∂ p ν + τ 1 u μ ) δ V μ = ℏ f μν ∂ p ν V μ eq , (44)
p μ δ A μ = 0 , (45) p^\mu\delta A_\mu=0, \tag*{(45)} p μ δ A μ = 0 , (45)
ℏ 2 ( ∂ x μ − F 0 , μ ν ∂ p ν + 1 τ u μ ) δ F + p ν δ S ν μ = ℏ 2 f μ ρ ∂ p ρ F e q , (46) \frac{\hbar}{2}\left(\partial_{x\mu}-F_{0,\mu\nu}\partial_p^\nu+\frac{1}{\tau}u_\mu\right)\delta F+p^\nu\delta S_{\nu\mu}=\frac{\hbar}{2}f_{\mu\rho}\partial_p^\rho F_{\mathrm{eq}}, \tag*{(46)} 2 ℏ ( ∂ xμ − F 0 , μν ∂ p ν + τ 1 u μ ) δ F + p ν δ S ν μ = 2 ℏ f μ ρ ∂ p ρ F eq , (46)
p μ δ P + ℏ 4 ϵ μ ν α β ( ∂ x ν − F 0 ν ρ ∂ p ρ + 1 τ u ν ) δ S α β = ℏ 4 ϵ μ ν α β f ν ρ ∂ p ρ S e q α β , (47) p_\mu\delta P+\frac{\hbar}{4}\epsilon_{\mu\nu\alpha\beta}\left(\partial_x^\nu-F_0^{\nu\rho}\partial_{p\rho}+\frac{1}{\tau}u^\nu\right)\delta S^{\alpha\beta}=\frac{\hbar}{4}\epsilon_{\mu\nu\alpha\beta}f^{\nu\rho}\partial_{p\rho}S_{\mathrm{eq}}^{\alpha\beta}, \tag*{(47)} p μ δ P + 4 ℏ ϵ μν α β ( ∂ x ν − F 0 ν ρ ∂ pρ + τ 1 u ν ) δ S α β = 4 ℏ ϵ μν α β f ν ρ ∂ pρ S eq α β , (47)
p μ δ V ν − p ν δ V μ + ℏ 2 ϵ μ ν α β ( ∂ x α − F 0 α ρ ∂ p ρ + 1 τ u α ) δ A β = ℏ 2 ϵ μ ν α β f α ρ ∂ p ρ A e q β . (48) p_\mu\delta V_\nu-p_\nu\delta V_\mu+\frac{\hbar}{2}\epsilon_{\mu\nu\alpha\beta}\left(\partial_x^\alpha-F_0^{\alpha\rho}\partial_{p\rho}+\frac{1}{\tau}u^\alpha\right)\delta A^\beta=\frac{\hbar}{2}\epsilon_{\mu\nu\alpha\beta}f^{\alpha\rho}\partial_{p\rho}A_{\mathrm{eq}}^\beta. \tag*{(48)} p μ δ V ν − p ν δ V μ + 2 ℏ ϵ μν α β ( ∂ x α − F 0 α ρ ∂ pρ + τ 1 u α ) δ A β = 2 ℏ ϵ μν α β f α ρ ∂ pρ A eq β . (48)
Taking the vector and axial-vector components δ V μ \delta V_\mu δ V μ and δ A μ \delta A_\mu δ A μ as basic ones, the remaining components are then given by
δ F = 1 m p μ δ V μ , (49) \delta F=\frac{1}{m}p_\mu\delta V^\mu, \tag*{(49)} δ F = m 1 p μ δ V μ , (49)
δ P = ℏ 2 m f μ ν ∂ p ν A μ e q − ℏ 2 m ( ∂ x μ − F 0 μ ν ∂ p ν + 1 τ u μ ) δ A μ , (50) \delta P=\frac{\hbar}{2m}f^{\mu\nu}\partial_{p\nu}A_\mu^{\mathrm{eq}}-\frac{\hbar}{2m}\left(\partial_x^\mu-F_0^{\mu\nu}\partial_{p\nu}+\frac{1}{\tau}u^\mu\right)\delta A_\mu, \tag*{(50)} δ P = 2 m ℏ f μν ∂ p ν A μ eq − 2 m ℏ ( ∂ x μ − F 0 μν ∂ p ν + τ 1 u μ ) δ A μ , (50)
δ S μ ν = ℏ 2 m ( ∂ x μ − F 0 , μ ρ ∂ p ρ + 1 τ u μ ) δ V ν − ℏ 2 m ( ∂ x ν − F 0 , ν ρ ∂ p ρ + 1 τ u ν ) δ V μ − 1 m ϵ μ ν α β p α δ A β − ℏ 2 m f μ ρ ∂ p ρ V ν e q + ℏ 2 m f ν ρ ∂ p ρ V μ e q . (51) \begin{aligned}
\delta S_{\mu\nu}={}&\frac{\hbar}{2m}\left(\partial_{x\mu}-F_{0,\mu\rho}\partial_p^\rho+\frac{1}{\tau}u_\mu\right)\delta V_\nu-\frac{\hbar}{2m}\left(\partial_{x\nu}-F_{0,\nu\rho}\partial_p^\rho+\frac{1}{\tau}u_\nu\right)\delta V_\mu\\
&-\frac{1}{m}\epsilon_{\mu\nu\alpha\beta}p^\alpha\delta A^\beta-\frac{\hbar}{2m}f_{\mu\rho}\partial_p^\rho V_\nu^{\mathrm{eq}}+\frac{\hbar}{2m}f_{\nu\rho}\partial_p^\rho V_\mu^{\mathrm{eq}}. \tag*{(51)}
\end{aligned} δ S μν = 2 m ℏ ( ∂ xμ − F 0 , μ ρ ∂ p ρ + τ 1 u μ ) δ V ν − 2 m ℏ ( ∂ xν − F 0 , ν ρ ∂ p ρ + τ 1 u ν ) δ V μ − m 1 ϵ μν α β p α δ A β − 2 m ℏ f μ ρ ∂ p ρ V ν eq + 2 m ℏ f ν ρ ∂ p ρ V μ eq . (51)
Substituting δ F \delta\mathcal{F} δ F , δ P \delta\mathcal{P} δ P , and δ S μ ν \delta\mathcal{S}_{\mu\nu} δ S μν into Eqs. (41) and (42) , we obtain the mass-shell constraints for δ V μ \delta V_\mu δ V μ and δ A μ \delta A_\mu δ A μ ,
( p 2 − m 2 ) δ V μ + ℏ 2 ϵ μ ν α β F 0 ν α δ A β = − ℏ 2 ϵ μ ν α β f ν α A e q β + O ( ℏ 2 ) , (52) (p^2-m^2)\delta V_\mu+\frac{\hbar}{2}\epsilon_{\mu\nu\alpha\beta}F_0^{\nu\alpha}\delta A^\beta=-\frac{\hbar}{2}\epsilon_{\mu\nu\alpha\beta}f^{\nu\alpha}A_{\mathrm{eq}}^\beta+\mathcal{O}(\hbar^2), \tag*{(52)} ( p 2 − m 2 ) δ V μ + 2 ℏ ϵ μν α β F 0 ν α δ A β = − 2 ℏ ϵ μν α β f ν α A eq β + O ( ℏ 2 ) , (52)
( p 2 − m 2 ) δ A μ − ℏ 2 ϵ μ ν α β F 0 ν α δ V β = − ℏ 2 ϵ μ ν α β f ν α V e q β + O ( ℏ 2 ) . (53) (p^2-m^2)\delta A_\mu-\frac{\hbar}{2}\epsilon_{\mu\nu\alpha\beta}F_0^{\nu\alpha}\delta V^\beta=-\frac{\hbar}{2}\epsilon_{\mu\nu\alpha\beta}f^{\nu\alpha}V_{\mathrm{eq}}^\beta+\mathcal{O}(\hbar^2). \tag*{(53)} ( p 2 − m 2 ) δ A μ − 2 ℏ ϵ μν α β F 0 ν α δ V β = − 2 ℏ ϵ μν α β f ν α V eq β + O ( ℏ 2 ) . (53)
On the other hand, the kinetic or Boltzmann equations for δ V μ \delta V_\mu δ V μ and δ A μ \delta A_\mu δ A μ are derived from Eqs. (46) and (48)
ℏ p ν ( ∂ x ν − F 0 , ν ρ ∂ p ρ + 1 τ u ν ) δ V μ − ℏ F 0 , μ ν δ V ν = ℏ f μ ν V e q ν + ℏ p ν f ν ρ ∂ p ρ V μ e q + O ( ℏ 2 ) , (54) \hbar p^\nu\left(\partial_{x^\nu}-F_{0,\nu\rho}\partial_p^\rho+\frac{1}{\tau}u_\nu\right)\delta V_\mu-\hbar F_{0,\mu\nu}\delta V^\nu
=\hbar f_{\mu\nu}V_{\mathrm{eq}}^\nu+\hbar p^\nu f_{\nu\rho}\partial_p^\rho V_\mu^{\mathrm{eq}}+\mathcal{O}(\hbar^2), \tag*{(54)} ℏ p ν ( ∂ x ν − F 0 , ν ρ ∂ p ρ + τ 1 u ν ) δ V μ − ℏ F 0 , μν δ V ν = ℏ f μν V eq ν + ℏ p ν f ν ρ ∂ p ρ V μ eq + O ( ℏ 2 ) , (54)
ℏ p ν ( ∂ x ν − F 0 ν ρ ∂ p ρ + 1 τ u ν ) δ A μ − ℏ F 0 , μ ν δ A ν = ℏ f μ ν A e q ν + ℏ p ν f ν ρ ∂ p ρ A μ e q + O ( ℏ 2 ) . (55) \hbar p_\nu\left(\partial_x^\nu-F_0^{\nu\rho}\partial_{p^\rho}+\frac{1}{\tau}u^\nu\right)\delta A_\mu-\hbar F_{0,\mu\nu}\delta A^\nu
=\hbar f_{\mu\nu}A_{\mathrm{eq}}^\nu+\hbar p^\nu f_{\nu\rho}\partial_p^\rho A_\mu^{\mathrm{eq}}+\mathcal{O}(\hbar^2). \tag*{(55)} ℏ p ν ( ∂ x ν − F 0 ν ρ ∂ p ρ + τ 1 u ν ) δ A μ − ℏ F 0 , μν δ A ν = ℏ f μν A eq ν + ℏ p ν f ν ρ ∂ p ρ A μ eq + O ( ℏ 2 ) . (55)
where we have dropped terms in O ( ℏ 2 ) \mathcal{O}(\hbar^2) O ( ℏ 2 ) and higher orders. Since we considered the case of the strong background magnetic field B 0 B_0 B 0 , we treat ℏ B 0 \hbar B_0 ℏ B 0 as the zeroth order quantity in power counting of ℏ \hbar ℏ . As the result, the equilibrium solutions in Eq. (29) are the zeroth order contributions. Consequently, δ V μ \delta V_\mu δ V μ and δ A μ \delta A_\mu δ A μ obtained by solving Eq. (54) and (55) are also the zeroth order contributions. In the following discussions, we will set ℏ = 1 \hbar=1 ℏ = 1 for simplicity.
In the strong magnetic field, particle’s energy spectra are described by the Landau levels. However, the perturbation f μ ν f^{\mu\nu} f μν will modify the energy spectra, as shown in Eqs. (52) and (53) . Since the perturbation f μ ν f^{\mu\nu} f μν is assumed to be much weaker than F 0 μ ν F_0^{\mu\nu} F 0 μν , one can safely expect that the deviation from Landau levels is negligible. Therefore in the following discussion, we neglect the mass-shell constraints for δ V μ \delta V_\mu δ V μ and δ A μ \delta A_\mu δ A μ in Eqs. (52) and (53) , and focusing on the Boltzmann equations (54) and (55) . For simplicity, we set B 0 B_0 B 0 , β \beta β , b μ b^\mu b μ , and u μ u^\mu u μ as spacetime-independent quantities, and assume that f μ ν f^{\mu\nu} f μν only depend on the proper time t ≡ u ⋅ x t\equiv u\cdot x t ≡ u ⋅ x . We decompose the perturbation f μ ν f^{\mu\nu} f μν into the electric and magnetic part as
f μ ν = δ E μ u ν − δ E ν u μ + ϵ μ ν α β u α δ B β . (56) f^{\mu\nu}=\delta E^\mu u^\nu-\delta E^\nu u^\mu+\epsilon^{\mu\nu\alpha\beta}u_\alpha\delta B_\beta. \tag*{(56)} f μν = δ E μ u ν − δ E ν u μ + ϵ μν α β u α δ B β . (56)
In Eqs. (54) and (55) , f μ ν f^{\mu\nu} f μν plays the role of a source term for δ V μ \delta V_\mu δ V μ and δ A μ \delta A_\mu δ A μ . In general, the obtained δ V μ \delta V_\mu δ V μ and δ A μ \delta A_\mu δ A μ can be expressed as linear responses to δ E μ \delta E^\mu δ E μ and δ B μ \delta B^\mu δ B μ . The induced vector and axial vector currents, defined as momentum integrals of δ V μ \delta V^\mu δ V μ and δ A μ \delta A^\mu δ A μ , are given by
δ J μ ≡ ∫ d 4 p δ V μ = σ E μ ν δ E ν + σ B μ ν δ B ν , \delta J^\mu\equiv\int d^4p\,\delta V^\mu= \sigma_E^{\mu\nu}\delta E_\nu+\sigma_B^{\mu\nu}\delta B_\nu, δ J μ ≡ ∫ d 4 p δ V μ = σ E μν δ E ν + σ B μν δ B ν ,
δ J A μ ≡ ∫ d 4 p δ A μ = χ A , E μ ν δ E ν + χ A , B μ ν δ B ν , (57) \delta J_A^\mu\equiv\int d^4p\,\delta A^\mu= \chi_{A,E}^{\mu\nu}\delta E_\nu+\chi_{A,B}^{\mu\nu}\delta B_\nu, \tag*{(57)} δ J A μ ≡ ∫ d 4 p δ A μ = χ A , E μν δ E ν + χ A , B μν δ B ν , (57)
where the linear responses are described by conductivities σ E μ ν \sigma_E^{\mu\nu} σ E μν , σ B μ ν \sigma_B^{\mu\nu} σ B μν , χ A , E μ ν \chi_{A,E}^{\mu\nu} χ A , E μν , and χ A , B μ ν \chi_{A,B}^{\mu\nu} χ A , B μν . Since we assumed that δ E μ \delta E^\mu δ E μ and δ B μ \delta B^\mu δ B μ are functions of t t t , these conductivities are also proper time-dependent quantities. In the following two sections, we will show
σ E μ ν = Ξ μ ν σ ⊥ − b μ b ν σ ∥ + ϵ μ ν ρ σ u ρ b σ σ H , \sigma_E^{\mu\nu}=\Xi^{\mu\nu}\sigma_\perp-b^\mu b^\nu\sigma_\parallel+\epsilon^{\mu\nu\rho\sigma}u_\rho b_\sigma\sigma_H, σ E μν = Ξ μν σ ⊥ − b μ b ν σ ∥ + ϵ μν ρ σ u ρ b σ σ H ,
σ B μ ν = 0 , \sigma_B^{\mu\nu}=0, σ B μν = 0 ,
χ A , E μ ν = − u μ b ν χ A , \chi_{A,E}^{\mu\nu}=-u^\mu b^\nu\chi_A, χ A , E μν = − u μ b ν χ A ,
χ A , B μ ν = Ξ μ ν χ ∥ + ϵ μ ν ρ σ u ρ b σ χ ⊥ , (58) \chi_{A,B}^{\mu\nu}=\Xi^{\mu\nu}\chi_\parallel+\epsilon^{\mu\nu\rho\sigma}u_\rho b_\sigma\chi_\perp, \tag*{(58)} χ A , B μν = Ξ μν χ ∥ + ϵ μν ρ σ u ρ b σ χ ⊥ , (58)
where σ ⊥ \sigma_\perp σ ⊥ , σ ∥ \sigma_\parallel σ ∥ are Ohm conductivities in the transverse and longitudinal directions with respect to the direction of the background magnetic field, σ H \sigma_H σ H is the Hall conductivity, χ A \chi_A χ A is the coefficient for the axial-charge induced by parallel electric and background magnetic field, and χ ∥ \chi_\parallel χ ∥ and χ ⊥ \chi_\perp χ ⊥ describe the axial current induced by the perturbative magnetic field δ B μ \delta B^\mu δ B μ . The explicit expressions for these conductivities will be given in following sections.
Vector current induced by perturbation fields
Analytical solutions
In this section, we solve the Boltzmann equation (54) to obtain the vector current. With F 0 μ ν F_0^{\mu\nu} F 0 μν in Eq. (1) , the Boltzmann equation (54) can be written as
[ ( u ⋅ p ) ( d d t + 1 τ ) − B 0 ϵ ν λ ρ σ p ⊥ λ ∂ ν p u ρ b σ ] δ V μ − B 0 ϵ μ ν ρ σ u ν b σ δ V ρ = f μ ν V e q ν + p ν f ν ρ ∂ p ρ V μ e q . (59) \left[(u\cdot p)\left(\frac{d}{dt}+\frac{1}{\tau}\right)-B_0\epsilon^{\nu\lambda\rho\sigma}p_{\perp\lambda}^{\ } \partial_\nu^p u_\rho b_\sigma\right]\delta V_\mu-B_0\epsilon_{\mu\nu\rho\sigma}u^\nu b^\sigma\delta V^\rho=f_{\mu\nu}V_{\mathrm{eq}}^\nu+p^\nu f_{\nu\rho}\partial_p^\rho V_\mu^{\mathrm{eq}}. \tag*{(59)} [ ( u ⋅ p ) ( d t d + τ 1 ) − B 0 ϵ ν λ ρ σ p ⊥ λ ∂ ν p u ρ b σ ] δ V μ − B 0 ϵ μν ρ σ u ν b σ δ V ρ = f μν V eq ν + p ν f ν ρ ∂ p ρ V μ eq . (59)
where we made approximation that δ V μ \delta V_\mu δ V μ only depends on the proper time t ≡ u ⋅ x t\equiv u\cdot x t ≡ u ⋅ x because f μ ν f^{\mu\nu} f μν is assumed to be only t t t -dependent. The transverse momentum and the transverse momentum derivative are defined with the help of the projection operator Ξ μ ν \Xi^{\mu\nu} Ξ μν in Eq. (2) ,
p ⊥ μ ≡ Ξ μ ν p ν , ∂ p , ⊥ μ ≡ Ξ μ ν ∂ p ν , (60) p_\perp^\mu\equiv\Xi^{\mu\nu}p_\nu,\qquad\partial_{p,\perp}^\mu\equiv\Xi^{\mu\nu}\partial_p^\nu, \tag*{(60)} p ⊥ μ ≡ Ξ μν p ν , ∂ p , ⊥ μ ≡ Ξ μν ∂ p ν , (60)
while the longitudinal terms are
p ∥ μ ≡ p μ − p ⊥ μ , ∂ ∥ , p μ ≡ ∂ p μ − ∂ ⊥ , p μ . (61) p_{\parallel}^{\mu} \equiv p^{\mu}-p_{\perp}^{\mu},\qquad\partial_{\parallel,p}^{\mu}\equiv\partial_{p}^{\mu}-\partial_{\perp,p}^{\mu}. \tag*{(61)} p ∥ μ ≡ p μ − p ⊥ μ , ∂ ∥ , p μ ≡ ∂ p μ − ∂ ⊥ , p μ . (61)
The vectors u μ u^{\mu} u μ , b μ b^{\mu} b μ , p ⊥ μ p_{\perp}^{\mu} p ⊥ μ , and ϵ μ ν ρ σ p ⊥ ν u ρ b σ \epsilon^{\mu\nu\rho\sigma}p_{\perp\nu}u_{\rho}b_{\sigma} ϵ μν ρ σ p ⊥ ν u ρ b σ are orthogonal to each other when p ⊥ μ ≠ 0 p_{\perp}^{\mu}\ne0 p ⊥ μ = 0 and therefore they form a complete set of bases { e i μ , i = 1 , 2 , 3 , 4 } ≡ { u μ , b μ , p ⊥ μ , ϵ μ ν ρ σ p ⊥ ν u ρ b σ } \{e_{i}^{\mu},i=1,2,3,4\}\equiv\{u^{\mu},b^{\mu},p_{\perp}^{\mu},\epsilon^{\mu\nu\rho\sigma}p_{\perp\nu}u_{\rho}b_{\sigma}\} { e i μ , i = 1 , 2 , 3 , 4 } ≡ { u μ , b μ , p ⊥ μ , ϵ μν ρ σ p ⊥ ν u ρ b σ } for any four-vector. The vector δ V μ \delta V^{\mu} δ V μ can then be decomposed as
δ V μ = u μ δ V 1 − b μ δ V 2 − 1 p T 2 [ p ⊥ μ δ V 3 + ϵ μ ν ρ σ p ⊥ ν u ρ b σ δ V 4 ] , (62) \delta V^{\mu}=u^{\mu}\delta V_{1}-b^{\mu}\delta V_{2}-\frac{1}{p_{T}^{2}}\left[p_{\perp}^{\mu}\delta V_{3}+\epsilon^{\mu\nu\rho\sigma}p_{\perp\nu}u_{\rho}b_{\sigma}\delta V_{4}\right], \tag*{(62)} δ V μ = u μ δ V 1 − b μ δ V 2 − p T 2 1 [ p ⊥ μ δ V 3 + ϵ μν ρ σ p ⊥ ν u ρ b σ δ V 4 ] , (62)
where we have used p T = − p ⊥ μ p ⊥ μ p_{T}=\sqrt{-p_{\perp}^{\mu}p_{\perp\mu}} p T = − p ⊥ μ p ⊥ μ . One can extract all components as δ V i = e i μ δ V μ \delta V_{i}=e_{i}^{\mu}\delta V_{\mu} δ V i = e i μ δ V μ . Substituting δ V μ \delta V_{\mu} δ V μ in Eq. (62) into the Boltzmann equation (59) , we obtain an equation for each component
[ ( u ⋅ p ) ( d d t + 1 τ ) − B 0 ϵ ν λ ρ σ p ⊥ ν ∂ p ⊥ λ u ρ b σ ] δ V i = e i μ ( f μ ν V e q ν + p ν f ν ρ ∂ p ρ V μ e q ) . (63) \left[(u\cdot p)\left(\frac{d}{dt}+\frac{1}{\tau}\right)-B_{0}\epsilon^{\nu\lambda\rho\sigma}p_{\perp\nu}\partial_{p}^{\perp}\!{}_{\lambda}u_{\rho}b_{\sigma}\right]\delta V_{i}=e_{i}^{\mu}\left(f_{\mu\nu}\mathcal{V}_{\mathrm{eq}}^{\nu}+p^{\nu}f_{\nu\rho}\partial_{p}^{\rho}\mathcal{V}_{\mu}^{\mathrm{eq}}\right). \tag*{(63)} [ ( u ⋅ p ) ( d t d + τ 1 ) − B 0 ϵ ν λ ρ σ p ⊥ ν ∂ p ⊥ λ u ρ b σ ] δ V i = e i μ ( f μν V eq ν + p ν f ν ρ ∂ p ρ V μ eq ) . (63)
The source term on the right-hand-side can be put into the form
e i μ ( f μ ν V e q ν + p ν f ν ρ ∂ p ρ V μ e q ) = C i ( 1 ) ( t , p T , u ⋅ p , b ⋅ p ) + p ⊥ μ C i , μ ( 2 ) ( t , p T , u ⋅ p , b ⋅ p ) , (64) e_{i}^{\mu}\left(f_{\mu\nu}\mathcal{V}_{\mathrm{eq}}^{\nu}+p^{\nu}f_{\nu\rho}\partial_{p}^{\rho}\mathcal{V}_{\mu}^{\mathrm{eq}}\right)=C_{i}^{(1)}(t,p_{T},u\cdot p,b\cdot p)+p_{\perp}^{\mu}C_{i,\mu}^{(2)}(t,p_{T},u\cdot p,b\cdot p), \tag*{(64)} e i μ ( f μν V eq ν + p ν f ν ρ ∂ p ρ V μ eq ) = C i ( 1 ) ( t , p T , u ⋅ p , b ⋅ p ) + p ⊥ μ C i , μ ( 2 ) ( t , p T , u ⋅ p , b ⋅ p ) , (64)
where C i ( 1 ) C_{i}^{(1)} C i ( 1 ) and C i , μ ( 2 ) C_{i,\mu}^{(2)} C i , μ ( 2 ) are even functions of p ⊥ μ p_{\perp}^{\mu} p ⊥ μ . With Eqs. (29) and (64) , we obtain the source terms that correspond to f i = 1 , 2 ( 1 ) f_{i=1,2}^{(1)} f i = 1 , 2 ( 1 ) and f i = 3 , 4 ( 2 ) μ f_{i=3,4}^{(2)\mu} f i = 3 , 4 ( 2 ) μ as
C 1 ( 1 ) = ( u ⋅ p ) p ∥ ν f ν ρ ∑ n = 0 ∞ [ ∂ ∥ , p ρ V n ( p ) ] Λ + ( n ) ( p T ) , C 2 ( 1 ) = ( b ⋅ p ) p ∥ ν f ν ρ ∑ n = 0 ∞ [ ∂ ∥ , p ρ V n ( p ) ] Λ + ( n ) ( p T ) , C 3 ( 2 ) μ = ∑ n = 0 ∞ Ξ μ ν f ν ρ p ∥ ρ ( 1 + 2 n B 0 p T 2 ) V n ( p ) Λ + ( n ) ( p T ) − Ξ μ ν f ν ρ ∑ n = 0 ∞ 2 n B 0 [ ∂ ∥ , p ρ V n ( p ) ] Λ + ( n ) ( p T ) − 2 Ξ μ ν f ν ρ p ∥ ρ ∑ n = 0 ∞ 2 n B 0 V n ( p ) [ ∂ ∂ p T 2 Λ + ( n ) ( p T ) ] , C 4 ( 2 ) μ = − ∑ n = 0 ∞ ( 1 − 2 n B 0 p T 2 ) ϵ μ ν ρ σ u ρ b σ f ν λ p ∥ λ V n ( p ) Λ + ( n ) ( p T ) . (65) \begin{aligned}
C_{1}^{(1)}&=(u\cdot p)p_{\parallel}^{\nu}f_{\nu\rho}\sum_{n=0}^{\infty}\left[\partial_{\parallel,p}^{\rho}\mathcal{V}_{n}(p)\right]\Lambda_{+}^{(n)}(p_{T}),\\
C_{2}^{(1)}&=(b\cdot p)p_{\parallel}^{\nu}f_{\nu\rho}\sum_{n=0}^{\infty}\left[\partial_{\parallel,p}^{\rho}\mathcal{V}_{n}(p)\right]\Lambda_{+}^{(n)}(p_{T}),\\
C_{3}^{(2)\mu}&=\sum_{n=0}^{\infty}\Xi^{\mu\nu}f_{\nu\rho}p_{\parallel}^{\rho}\left(1+\frac{2nB_{0}}{p_{T}^{2}}\right)\mathcal{V}_{n}(p)\Lambda_{+}^{(n)}(p_{T})\\
&\quad-\Xi^{\mu\nu}f_{\nu\rho}\sum_{n=0}^{\infty}2nB_{0}\left[\partial_{\parallel,p}^{\rho}\mathcal{V}_{n}(p)\right]\Lambda_{+}^{(n)}(p_{T})\\
&\quad-2\Xi^{\mu\nu}f_{\nu\rho}p_{\parallel}^{\rho}\sum_{n=0}^{\infty}2nB_{0}\mathcal{V}_{n}(p)\left[\frac{\partial}{\partial p_{T}^{2}}\Lambda_{+}^{(n)}(p_{T})\right],\\
C_{4}^{(2)\mu}&=-\sum_{n=0}^{\infty}\left(1-\frac{2nB_{0}}{p_{T}^{2}}\right)\epsilon^{\mu\nu\rho\sigma}u_{\rho}b_{\sigma}f_{\nu\lambda}p_{\parallel}^{\lambda}\mathcal{V}_{n}(p)\Lambda_{+}^{(n)}(p_{T}). \tag*{(65)}
\end{aligned} C 1 ( 1 ) C 2 ( 1 ) C 3 ( 2 ) μ C 4 ( 2 ) μ = ( u ⋅ p ) p ∥ ν f ν ρ n = 0 ∑ ∞ [ ∂ ∥ , p ρ V n ( p ) ] Λ + ( n ) ( p T ) , = ( b ⋅ p ) p ∥ ν f ν ρ n = 0 ∑ ∞ [ ∂ ∥ , p ρ V n ( p ) ] Λ + ( n ) ( p T ) , = n = 0 ∑ ∞ Ξ μν f ν ρ p ∥ ρ ( 1 + p T 2 2 n B 0 ) V n ( p ) Λ + ( n ) ( p T ) − Ξ μν f ν ρ n = 0 ∑ ∞ 2 n B 0 [ ∂ ∥ , p ρ V n ( p ) ] Λ + ( n ) ( p T ) − 2 Ξ μν f ν ρ p ∥ ρ n = 0 ∑ ∞ 2 n B 0 V n ( p ) [ ∂ p T 2 ∂ Λ + ( n ) ( p T ) ] , = − n = 0 ∑ ∞ ( 1 − p T 2 2 n B 0 ) ϵ μν ρ σ u ρ b σ f ν λ p ∥ λ V n ( p ) Λ + ( n ) ( p T ) . (65)
Assuming δ V i ( t 0 ) = 0 \delta V_{i}(t_{0})=0 δ V i ( t 0 ) = 0 at a given initial moment t 0 t_{0} t 0 , the solution to Eq. (63) reads
δ V i ( t , p ) = f i ( 1 ) ( t , p T , u ⋅ p , b ⋅ p ) + p ⊥ μ f i , μ ( 2 ) ( t , p T , u ⋅ p , b ⋅ p ) , (66) \delta V_{i}(t,p)=f_{i}^{(1)}(t,p_{T},u\cdot p,b\cdot p)+p_{\perp}^{\mu}f_{i,\mu}^{(2)}(t,p_{T},u\cdot p,b\cdot p), \tag*{(66)} δ V i ( t , p ) = f i ( 1 ) ( t , p T , u ⋅ p , b ⋅ p ) + p ⊥ μ f i , μ ( 2 ) ( t , p T , u ⋅ p , b ⋅ p ) , (66)
where f i ( 1 ) f_i^{(1)} f i ( 1 ) and f i , μ ( 2 ) f_{i,\mu}^{(2)} f i , μ ( 2 ) are obtained from C i ( 1 ) C_i^{(1)} C i ( 1 ) and C i ( 2 ) μ C_i^{(2)\mu} C i ( 2 ) μ as
f i ( 1 ) ( t ) = 1 u ⋅ p ∫ t 0 t d t ′ e ( t ′ − t ) / τ C i ( 1 ) ( t ′ ) , f_i^{(1)}(t)=\frac{1}{u\cdot p}\int_{t_0}^{t}dt'e^{(t'-t)/\tau}C_i^{(1)}(t'), f i ( 1 ) ( t ) = u ⋅ p 1 ∫ t 0 t d t ′ e ( t ′ − t ) / τ C i ( 1 ) ( t ′ ) ,
f i , μ ( 2 ) ( t ) = − 1 B 0 ( u ⋅ p ) ∫ t 0 t d t ′ sin [ B 0 u ⋅ p ( t ′ − t ) ] e ( t ′ − t ) / τ [ ( u ⋅ p ) ( d d t ′ + 1 τ ) C i , μ ( 2 ) ( t ′ ) + B 0 ϵ μ ν α β u α b β C i ( 2 ) ν ( t ′ ) ] . (67) f_{i,\mu}^{(2)}(t)=-\frac{1}{B_0(u\cdot p)}\int_{t_0}^{t}dt'\sin\left[\frac{B_0}{u\cdot p}(t'-t)\right]e^{(t'-t)/\tau}
\left[(u\cdot p)\left(\frac{d}{dt'}+\frac{1}{\tau}\right)C_{i,\mu}^{(2)}(t')+B_0\epsilon_{\mu\nu\alpha\beta}u^\alpha b^\beta C_i^{(2)\nu}(t')\right]. \tag*{(67)} f i , μ ( 2 ) ( t ) = − B 0 ( u ⋅ p ) 1 ∫ t 0 t d t ′ sin [ u ⋅ p B 0 ( t ′ − t ) ] e ( t ′ − t ) / τ [ ( u ⋅ p ) ( d t ′ d + τ 1 ) C i , μ ( 2 ) ( t ′ ) + B 0 ϵ μν α β u α b β C i ( 2 ) ν ( t ′ ) ] . (67)
Here we have suppressed the dependence of all functions on p T p_T p T , u ⋅ p u\cdot p u ⋅ p , and b ⋅ p b\cdot p b ⋅ p for notational simplicity. Substituting Eq. (66) into Eq. (62) , we obtain
δ V μ = u μ f 1 ( 1 ) − b μ f 2 ( 1 ) + [ u μ f 1 , ν ( 2 ) − b μ f 2 , ν ( 2 ) ] p ⊥ ν + 1 p T 2 [ p ⊥ μ p ⊥ ν f 3 , ν ( 1 ) + ϵ μ ν ρ σ p ⊥ ν u ρ b σ f 4 ( 1 ) ] − 1 p T 2 [ p ⊥ μ p ⊥ ν f 3 , ν ( 2 ) μ + ϵ μ ν ρ σ p ⊥ ν u ρ b σ p ⊥ λ f 4 , λ ( 2 ) ] . (68) \begin{aligned}
\delta V^\mu={}&u^\mu f_1^{(1)}-b^\mu f_2^{(1)}
+\left[u^\mu f_{1,\nu}^{(2)}-b^\mu f_{2,\nu}^{(2)}\right]p_\perp^\nu\\
&+\frac{1}{p_T^2}\left[p_\perp^\mu p_\perp^\nu f_{3,\nu}^{(1)}
+\epsilon^{\mu\nu\rho\sigma}p_{\perp\nu}u_\rho b_\sigma f_{4}^{(1)}\right]\\
&-\frac{1}{p_T^2}\left[p_\perp^\mu p_\perp^\nu f_{3,\nu}^{(2)\mu}
+\epsilon^{\mu\nu\rho\sigma}p_{\perp\nu}u_\rho b_\sigma p_\perp^\lambda f_{4,\lambda}^{(2)}\right]. \tag*{(68)}
\end{aligned} δ V μ = u μ f 1 ( 1 ) − b μ f 2 ( 1 ) + [ u μ f 1 , ν ( 2 ) − b μ f 2 , ν ( 2 ) ] p ⊥ ν + p T 2 1 [ p ⊥ μ p ⊥ ν f 3 , ν ( 1 ) + ϵ μν ρ σ p ⊥ ν u ρ b σ f 4 ( 1 ) ] − p T 2 1 [ p ⊥ μ p ⊥ ν f 3 , ν ( 2 ) μ + ϵ μν ρ σ p ⊥ ν u ρ b σ p ⊥ λ f 4 , λ ( 2 ) ] . (68)
Note that f i ( 1 ) f_i^{(1)} f i ( 1 ) and f i ( 2 ) μ f_i^{(2)\mu} f i ( 2 ) μ are even functions of p ⊥ μ p_\perp^\mu p ⊥ μ , thus when calculating the vector current J μ ( x ) J^\mu(x) J μ ( x ) through integration δ V μ \delta V_\mu δ V μ over four-momentum, only terms with f 1 ( 1 ) f_1^{(1)} f 1 ( 1 ) , f 2 ( 1 ) f_2^{(1)} f 2 ( 1 ) , f 3 ( 2 ) μ f_3^{(2)\mu} f 3 ( 2 ) μ and f 4 ( 2 ) μ f_4^{(2)\mu} f 4 ( 2 ) μ contribute but other terms do not. So we have
J μ ( x ) = u μ ∫ d 4 p f 1 ( 1 ) − b μ ∫ d 4 p f 2 ( 1 ) + 1 2 ∫ d 4 p [ f 3 ( 2 ) μ + ϵ μ ν ρ σ u ρ b σ f 4 , ν ( 2 ) ] . (69) J^\mu(x)=u^\mu\int d^4p\,f_1^{(1)}-b^\mu\int d^4p\,f_2^{(1)}
+\frac{1}{2}\int d^4p\left[f_3^{(2)\mu}+\epsilon^{\mu\nu\rho\sigma}u_\rho b_\sigma f_{4,\nu}^{(2)}\right]. \tag*{(69)} J μ ( x ) = u μ ∫ d 4 p f 1 ( 1 ) − b μ ∫ d 4 p f 2 ( 1 ) + 2 1 ∫ d 4 p [ f 3 ( 2 ) μ + ϵ μν ρ σ u ρ b σ f 4 , ν ( 2 ) ] . (69)
where we have used the relation
∫ d 4 p p ⊥ μ p ⊥ ν p T 2 f i , ν ( 2 ) = − 1 2 ∫ d 4 p f i ( 2 ) μ . (70) \int d^4p\frac{p_\perp^\mu p_\perp^\nu}{p_T^2}f_{i,\nu}^{(2)}=-\frac{1}{2}\int d^4p\,f_i^{(2)\mu}. \tag*{(70)} ∫ d 4 p p T 2 p ⊥ μ p ⊥ ν f i , ν ( 2 ) = − 2 1 ∫ d 4 p f i ( 2 ) μ . (70)
The vector current density J μ ( x ) J_\mu(x) J μ ( x ) can be calculated using Eqs. (67) , (69) , and (65) and the decomposition of f μ ν f^{\mu\nu} f μν in Eq. (56) ,
J μ ( x ) = − 1 ( 2 π ) 2 ∑ n = 0 ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) ∫ t 0 t d t ′ e ( t ′ − t ) / τ lim ε → ε b ⋅ p ( n ) [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] × { b μ b ν δ E ν ( t ′ ) B 0 ( m 2 + 2 n B 0 ) ε 3 + [ Ξ μ ν ( d d t ′ + 1 τ ) δ E ν ( t ′ ) ] ( 1 + n B 0 ε ∂ ∂ ε ) sin [ B 0 ε ( t ′ − t ) ] } − 1 ( 2 π ) 2 ∑ n = 0 ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) ∫ t 0 t d t ′ e ( t ′ − t ) / τ lim ε → ε b ⋅ p ( n ) [ f e q ( + ) ( ε ) − f e q ( − ) ( ε ) ] × B 0 ε ϵ μ ν ρ σ u ρ b σ δ E ν ( t ′ ) ( 1 + n B 0 ∂ ∂ ε 1 ε ) sin [ B 0 ε ( t ′ − t ) ] . (71) \begin{aligned}
J^\mu(x)={}&-\frac{1}{(2\pi)^2}\sum_{n=0}^{\infty}(2-\delta_{n0})\int d(b\cdot p)\int_{t_0}^{t}dt'e^{(t'-t)/\tau}\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(n)}}\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)+f_{\mathrm{eq}}^{(-)}(\varepsilon)\right]\\
&\times\left\{b^\mu b^\nu\delta E_\nu(t')\frac{B_0(m^2+2nB_0)}{\varepsilon^3}
+\left[\Xi^{\mu\nu}\left(\frac{d}{dt'}+\frac{1}{\tau}\right)\delta E_\nu(t')\right]\left(1+\frac{nB_0}{\varepsilon}\frac{\partial}{\partial\varepsilon}\right)\sin\left[\frac{B_0}{\varepsilon}(t'-t)\right]\right\}\\
&-\frac{1}{(2\pi)^2}\sum_{n=0}^{\infty}(2-\delta_{n0})\int d(b\cdot p)\int_{t_0}^{t}dt'e^{(t'-t)/\tau}\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(n)}}\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)-f_{\mathrm{eq}}^{(-)}(\varepsilon)\right]\\
&\times\frac{B_0}{\varepsilon}\epsilon^{\mu\nu\rho\sigma}u_\rho b_\sigma\delta E_\nu(t')\left(1+nB_0\frac{\partial}{\partial\varepsilon}\frac{1}{\varepsilon}\right)\sin\left[\frac{B_0}{\varepsilon}(t'-t)\right]. \tag*{(71)}
\end{aligned} J μ ( x ) = − ( 2 π ) 2 1 n = 0 ∑ ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) ∫ t 0 t d t ′ e ( t ′ − t ) / τ ε → ε b ⋅ p ( n ) lim [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] × { b μ b ν δ E ν ( t ′ ) ε 3 B 0 ( m 2 + 2 n B 0 ) + [ Ξ μν ( d t ′ d + τ 1 ) δ E ν ( t ′ ) ] ( 1 + ε n B 0 ∂ ε ∂ ) sin [ ε B 0 ( t ′ − t ) ] } − ( 2 π ) 2 1 n = 0 ∑ ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) ∫ t 0 t d t ′ e ( t ′ − t ) / τ ε → ε b ⋅ p ( n ) lim [ f eq ( + ) ( ε ) − f eq ( − ) ( ε ) ] × ε B 0 ϵ μν ρ σ u ρ b σ δ E ν ( t ′ ) ( 1 + n B 0 ∂ ε ∂ ε 1 ) sin [ ε B 0 ( t ′ − t ) ] . (71)
where f e q ( ± ) ( ε ) ≡ f e q ( ± ) ( u ⋅ p = ε ) f_{\mathrm{eq}}^{(\pm)}(\varepsilon)\equiv f_{\mathrm{eq}}^{(\pm)}(u\cdot p=\varepsilon) f eq ( ± ) ( ε ) ≡ f eq ( ± ) ( u ⋅ p = ε ) are given in Eq. (32) , ε b ⋅ p ( n ) ≡ m 2 + ( b ⋅ p ) 2 + 2 n B 0 \varepsilon_{b\cdot p}^{(n)}\equiv\sqrt{m^2+(b\cdot p)^2+2nB_0} ε b ⋅ p ( n ) ≡ m 2 + ( b ⋅ p ) 2 + 2 n B 0 is the on-shell energy at n n n -th Landau level, and we have completed the integrals over u ⋅ p u\cdot p u ⋅ p and p ⊥ μ p_\perp^\mu p ⊥ μ using explicit expressions for V e q ( n ) V_{\mathrm{eq}}^{(n)} V eq ( n ) , Λ ± ( n ) ( p T ) \Lambda_\pm^{(n)}(p_T) Λ ± ( n ) ( p T ) and V n ( p ) V_n(p) V n ( p ) in Eqs. (67) , (30) , and (31) .
Assuming that the electric fields are suddenly switched on at t 0 = 0 t_0=0 t 0 = 0 , δ E μ ∝ θ ( t ) \delta E^\mu\propto\theta(t) δ E μ ∝ θ ( t ) , and δ E μ \delta E^\mu δ E μ is a constant for t > 0 t>0 t > 0 . Then we can write the vector current as
J μ ( x ) = σ ⊥ Ξ μ ν δ E ν − σ ∥ b μ b ν δ E ν + σ H ϵ μ ν ρ σ u ρ b σ δ E ν , (72) J^\mu(x)=\sigma_\perp\Xi^{\mu\nu}\delta E_\nu-\sigma_\parallel b^\mu b^\nu\delta E_\nu+\sigma_H\epsilon^{\mu\nu\rho\sigma}u_\rho b_\sigma\delta E_\nu, \tag*{(72)} J μ ( x ) = σ ⊥ Ξ μν δ E ν − σ ∥ b μ b ν δ E ν + σ H ϵ μν ρ σ u ρ b σ δ E ν , (72)
where we still used δ E ν \delta E_\nu δ E ν but to represent the constant magnitude of the electric field perturbation, σ ∥ \sigma_\parallel σ ∥ and σ ⊥ \sigma_\perp σ ⊥ are the electric conductivities in the longitudinal and transverse direction respectively, and σ H \sigma_H σ H is the Hall conductivity. These conductivities are given by
σ ∥ = B 0 ( 2 π ) 2 R 1 ( t ) ∑ n = 0 ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) × lim ε → ε b ⋅ p ( n ) m 2 + 2 n B 0 ε 3 [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] , (73) \begin{aligned}
\sigma_\parallel={}&\frac{B_0}{(2\pi)^2}\mathcal{R}_1(t)\sum_{n=0}^{\infty}(2-\delta_{n0})\int d(b\cdot p)\\
&\times\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(n)}}\frac{m^2+2nB_0}{\varepsilon^3}\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)+f_{\mathrm{eq}}^{(-)}(\varepsilon)\right], \tag*{(73)}
\end{aligned} σ ∥ = ( 2 π ) 2 B 0 R 1 ( t ) n = 0 ∑ ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) × ε → ε b ⋅ p ( n ) lim ε 3 m 2 + 2 n B 0 [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] , (73)
σ ⊥ = − 1 ( 2 π ) 2 τ ∑ n = 0 ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) × lim ε → ε b ⋅ p ( n ) [ ( 1 + n B 0 ε ∂ ∂ ε ) R 2 ( ε , B 0 , t ) ] [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] , (74) \begin{aligned}
\sigma_\perp={}&-\frac{1}{(2\pi)^2\tau}\sum_{n=0}^{\infty}(2-\delta_{n0})\int d(b\cdot p)\\
&\times\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(n)}}\left[\left(1+\frac{nB_0}{\varepsilon}\frac{\partial}{\partial\varepsilon}\right)\mathcal{R}_2(\varepsilon,B_0,t)\right]\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)+f_{\mathrm{eq}}^{(-)}(\varepsilon)\right], \tag*{(74)}
\end{aligned} σ ⊥ = − ( 2 π ) 2 τ 1 n = 0 ∑ ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) × ε → ε b ⋅ p ( n ) lim [ ( 1 + ε n B 0 ∂ ε ∂ ) R 2 ( ε , B 0 , t ) ] [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] , (74)
σ H = − B 0 ( 2 π ) 2 ∑ n = 0 ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) × lim ε → ε b ⋅ p ( n ) 1 ε [ ( 1 + n B 0 ∂ ∂ ε 1 ε ) R 2 ( ε , B 0 , t ) ] [ f e q ( + ) ( ε ) − f e q ( − ) ( ε ) ] . (75) \begin{aligned}
\sigma_H={}&-\frac{B_0}{(2\pi)^2}\sum_{n=0}^{\infty}(2-\delta_{n0})\int d(b\cdot p)\\
&\times\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(n)}}\frac{1}{\varepsilon}\left[\left(1+nB_0\frac{\partial}{\partial\varepsilon}\frac{1}{\varepsilon}\right)\mathcal{R}_2(\varepsilon,B_0,t)\right]\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)-f_{\mathrm{eq}}^{(-)}(\varepsilon)\right]. \tag*{(75)}
\end{aligned} σ H = − ( 2 π ) 2 B 0 n = 0 ∑ ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) × ε → ε b ⋅ p ( n ) lim ε 1 [ ( 1 + n B 0 ∂ ε ∂ ε 1 ) R 2 ( ε , B 0 , t ) ] [ f eq ( + ) ( ε ) − f eq ( − ) ( ε ) ] . (75)
The auxiliary functions R 1 ( t ) \mathcal{R}_1(t) R 1 ( t ) and R 2 ( ε , B 0 , t ) \mathcal{R}_2(\varepsilon,B_0,t) R 2 ( ε , B 0 , t ) are defined as
R 1 ( t ) ≡ τ ( 1 − e − t / τ ) , \mathcal{R}_1(t)\equiv\tau\left(1-e^{-t/\tau}\right), R 1 ( t ) ≡ τ ( 1 − e − t / τ ) ,
R 2 ( ε , B 0 , t ) = − ε τ ε 2 + B 0 2 τ 2 { B 0 τ − e − t / τ [ B 0 τ cos ( B 0 ε t ) + ε sin ( B 0 ε t ) ] } , (76) \mathcal{R}_2(\varepsilon,B_0,t)=-\frac{\varepsilon\tau}{\varepsilon^2+B_0^2\tau^2}\left\{B_0\tau-e^{-t/\tau}\left[B_0\tau\cos\left(\frac{B_0}{\varepsilon}t\right)+\varepsilon\sin\left(\frac{B_0}{\varepsilon}t\right)\right]\right\}, \tag*{(76)} R 2 ( ε , B 0 , t ) = − ε 2 + B 0 2 τ 2 ε τ { B 0 τ − e − t / τ [ B 0 τ cos ( ε B 0 t ) + ε sin ( ε B 0 t ) ] } , (76)
which controls the time evolution. Since f e q ( ± ) → 0 f_{\mathrm{eq}}^{(\pm)}\to0 f eq ( ± ) → 0 at ε → ∞ \varepsilon\to\infty ε → ∞ , we find that the sum over Landau levels converges. This ensures that all these conductivities have finite values.
At the long time limit t → ∞ t\to\infty t → ∞ , the auxiliary functions R 1 ( t ) \mathcal{R}_1(t) R 1 ( t ) and R 2 ( ε , B 0 , t ) \mathcal{R}_2(\varepsilon,B_0,t) R 2 ( ε , B 0 , t ) in Eq. (76) are reduced to τ \tau τ and − ε B 0 τ 2 / ( ε 2 + B 0 2 τ 2 ) -\varepsilon B_0\tau^2/(\varepsilon^2+B_0^2\tau^2) − ε B 0 τ 2 / ( ε 2 + B 0 2 τ 2 ) , respectively. The asymptotic values of conductivities are
lim t → ∞ σ ∥ = B 0 τ ( 2 π ) 2 ∑ n = 0 ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) lim ε → ε b ⋅ p ( n ) m 2 + 2 n B 0 ε 3 [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] , (77) \lim_{t\to\infty}\sigma_\parallel=\frac{B_0\tau}{(2\pi)^2}\sum_{n=0}^{\infty}(2-\delta_{n0})\int d(b\cdot p)\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(n)}}\frac{m^2+2nB_0}{\varepsilon^3}\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)+f_{\mathrm{eq}}^{(-)}(\varepsilon)\right], \tag*{(77)} t → ∞ lim σ ∥ = ( 2 π ) 2 B 0 τ n = 0 ∑ ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) ε → ε b ⋅ p ( n ) lim ε 3 m 2 + 2 n B 0 [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] , (77)
lim t → ∞ σ ⊥ = B 0 τ ( 2 π ) 2 ∑ n = 0 ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) lim ε → ε b ⋅ p ( n ) [ ( 1 + n B 0 ε ∂ ∂ ε ) ε ε 2 + B 0 2 τ 2 ] × [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] . (78) \lim_{t\to\infty}\sigma_{\perp}=\frac{B_{0}\tau}{(2\pi)^{2}}\sum_{n=0}^{\infty}(2-\delta_{n0})\int d(b\cdot p)\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(n)}}\left[\left(1+\frac{nB_{0}}{\varepsilon}\frac{\partial}{\partial\varepsilon}\right)\frac{\varepsilon}{\varepsilon^{2}+B_{0}^{2}\tau^{2}}\right]\times\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)+f_{\mathrm{eq}}^{(-)}(\varepsilon)\right]. \tag*{(78)} t → ∞ lim σ ⊥ = ( 2 π ) 2 B 0 τ n = 0 ∑ ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) ε → ε b ⋅ p ( n ) lim [ ( 1 + ε n B 0 ∂ ε ∂ ) ε 2 + B 0 2 τ 2 ε ] × [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] . (78)
lim t → ∞ σ H = B 0 2 τ 2 ( 2 π ) 2 ∑ n = 0 ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) lim ε → ε b ⋅ p ( n ) 1 ε 2 + B 0 2 τ 2 ( 1 − 2 n B 0 ε 2 + B 0 2 τ 2 ) × [ f e q ( + ) ( ε ) − f e q ( − ) ( ε ) ] . (79) \lim_{t\to\infty}\sigma_{H}=\frac{B_{0}^{2}\tau^{2}}{(2\pi)^{2}}\sum_{n=0}^{\infty}(2-\delta_{n0})\int d(b\cdot p)\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(n)}}\frac{1}{\varepsilon^{2}+B_{0}^{2}\tau^{2}}\left(1-\frac{2nB_{0}}{\varepsilon^{2}+B_{0}^{2}\tau^{2}}\right)\times\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)-f_{\mathrm{eq}}^{(-)}(\varepsilon)\right]. \tag*{(79)} t → ∞ lim σ H = ( 2 π ) 2 B 0 2 τ 2 n = 0 ∑ ∞ ( 2 − δ n 0 ) ∫ d ( b ⋅ p ) ε → ε b ⋅ p ( n ) lim ε 2 + B 0 2 τ 2 1 ( 1 − ε 2 + B 0 2 τ 2 2 n B 0 ) × [ f eq ( + ) ( ε ) − f eq ( − ) ( ε ) ] . (79)
We observe that σ ∥ ( t → ∞ ) \sigma_{\parallel}(t\to\infty) σ ∥ ( t → ∞ ) is always proportional to the relaxation time τ \tau τ , which is consistent with the result in materials [95 ] . For weak magnetic fields, the transverse conductivity σ ⊥ ( t → ∞ ) \sigma_{\perp}(t\to\infty) σ ⊥ ( t → ∞ ) is also proportional to τ \tau τ , while the Hall conductivity σ H \sigma_{H} σ H is proportional to τ 2 \tau^{2} τ 2 . Strong magnetic fields break the linear proportionality between σ ⊥ ( t → ∞ ) \sigma_{\perp}(t\to\infty) σ ⊥ ( t → ∞ ) and τ \tau τ , and that between σ H \sigma_{H} σ H and τ 2 \tau^{2} τ 2 due to the term B 0 2 τ 2 B_{0}^{2}\tau^{2} B 0 2 τ 2 in the denominators in Eqs. (78) and (79) .
Numerical results
Now we present numerical results for conductivities in Eqs. (73) -(75) . In displaying numerical results, we regard the temperature T T T as the energy scale so that all quantities and parameters can be made dimensionless by multiplying powers of T T T . We choose values of dimensionless parameters in numerical calculation.
In Figure 1 (a), we show dimensionless conductivities σ ∥ / ( T 2 τ ) \sigma_{\parallel}/(T^{2}\tau) σ ∥ / ( T 2 τ ) and σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) as functions of the time t / τ t/\tau t / τ for T τ = 0.1 T\tau=0.1 T τ = 0.1 , 1.0 1.0 1.0 and 2.0 2.0 2.0 . As an example, we set μ / T = 0 \mu/T=0 μ / T = 0 , B 0 / T 2 = 1 B_{0}/T^{2}=1 B 0 / T 2 = 1 , and m / T = 1 m/T=1 m / T = 1 . We observe that σ ∥ / ( T 2 τ ) \sigma_{\parallel}/(T^{2}\tau) σ ∥ / ( T 2 τ ) does not depend on T τ T\tau T τ , as expected, while σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) is larger for smaller T τ T\tau T τ . As t t t increases, both σ ∥ \sigma_{\parallel} σ ∥ and σ ⊥ \sigma_{\perp} σ ⊥ increase and eventually reach saturated or asymptotic values when t t t is sufficiently large. We have verified that these behaviors hold for any set of parameters { μ / T , B 0 / T 2 , m / T } \{\mu/T,B_{0}/T^{2},m/T\} { μ / T , B 0 / T 2 , m / T } . In Figure 1 (b), we plot σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) as functions of t / τ t/\tau t / τ for T τ = 0.1 T\tau=0.1 T τ = 0.1 , 1.0 1.0 1.0 and 2.0 2.0 2.0 . We set μ / T = 0.1 \mu/T=0.1 μ / T = 0.1 because the Hall conductivity vanishes at μ = 0 \mu=0 μ = 0 . We observe that σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) increases with t / τ t/\tau t / τ and reaches saturated or asymptotic values when t t t is large enough. For any set of { μ / T , B 0 / T 2 , m / T } \{\mu/T,B_{0}/T^{2},m/T\} { μ / T , B 0 / T 2 , m / T } , the Hall conductivity is larger for smaller T τ T\tau T τ , which is due to the presence of B 0 2 τ 2 B_{0}^{2}\tau^{2} B 0 2 τ 2 in the denominator of Eq. (76) .
Figure 1. (a) The Ohm conductivities σ ∥ / ( T 2 τ ) \sigma_{\parallel}/(T^{2}\tau) σ ∥ / ( T 2 τ ) and σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) as functions of t / τ t/\tau t / τ for different values of T τ T\tau T τ . The parameters are set to μ / T = 0 \mu/T=0 μ / T = 0 , B 0 / T 2 = 1.0 B_{0}/T^{2}=1.0 B 0 / T 2 = 1.0 and m / T = 1.0 m/T=1.0 m / T = 1.0 . Note that σ ∥ / ( T 2 τ ) \sigma_{\parallel}/(T^{2}\tau) σ ∥ / ( T 2 τ ) does not depend on T τ T\tau T τ and is shown in black solid line. The red, green and blue dotted lines are σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) for T τ = 0.1 T\tau=0.1 T τ = 0.1 , 1.0 1.0 1.0 and 2.0 2.0 2.0 , respectively. (b) The Hall conductivity σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) as functions of t / τ t/\tau t / τ for different values of T τ T\tau T τ . The parameters are set to μ / T = 0.1 \mu/T=0.1 μ / T = 0.1 , B 0 / T 2 = 1.0 B_{0}/T^{2}=1.0 B 0 / T 2 = 1.0 and m / T = 1.0 m/T=1.0 m / T = 1.0 .
The long-time asymptotic values of σ ∥ \sigma_{\parallel} σ ∥ , σ ⊥ \sigma_{\perp} σ ⊥ , and σ H \sigma_{H} σ H are presented as functions of B 0 / T 2 B_{0}/T^{2} B 0 / T 2 in Figure 2 and Figure 3 . The results for σ ∥ / ( T 2 τ ) \sigma_{\parallel}/(T^{2}\tau) σ ∥ / ( T 2 τ ) and σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) are shown in solid and dotted lines respectively. In Figure 2 (a), we set μ / T = 0 \mu/T=0 μ / T = 0 , m / T = 1 m/T=1 m / T = 1 , the black solid line represents σ ∥ / ( T 2 τ ) \sigma_{\parallel}/(T^{2}\tau) σ ∥ / ( T 2 τ ) which does not depend on T τ T\tau T τ , as can be seen in Eq. (77) , and the red, green and blue dotted lines represent σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) for T τ = 0.1 T\tau=0.1 T τ = 0.1 , 1.0 1.0 1.0 and 2.0 2.0 2.0 respectively. As B 0 / T 2 ≈ 0 B_{0}/T^{2}\approx0 B 0 / T 2 ≈ 0 , we observe that σ ∥ \sigma_{\parallel} σ ∥ and σ ⊥ \sigma_{\perp} σ ⊥ are equal. At finite values of B 0 / T 2 B_{0}/T^{2} B 0 / T 2 , there is a difference between σ ∥ \sigma_{\parallel} σ ∥ and σ ⊥ \sigma_{\perp} σ ⊥ . We find that σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) increases with B 0 / T 2 B_{0}/T^{2} B 0 / T 2 for small T τ T\tau T τ , while it decreases with B 0 / T 2 B_{0}/T^{2} B 0 / T 2 and finally approaches to zero at T τ → ∞ T\tau\rightarrow\infty T τ → ∞ . Figure 2 (b) shows the particle mass effects. We set m / T = 0.1 m/T=0.1 m / T = 0.1 , 1.0 1.0 1.0 and 2.0 2.0 2.0 represented by red, green and blue lines respectively. For all three cases, we find that σ ∥ / ( T 2 τ ) \sigma_{\parallel}/(T^{2}\tau) σ ∥ / ( T 2 τ ) increases while σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) decreases with B 0 / T 2 B_{0}/T^{2} B 0 / T 2 . Such results are qualitatively consistent with those by the lattice QCD calculation [72 ] and the holographic model [76 ] .
Figure 2. Asymptotic values at long time limit for σ ∥ / ( T 2 τ ) \sigma_{\parallel}/(T^{2}\tau) σ ∥ / ( T 2 τ ) (solid lines) and σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) (dotted lines) as functions of B 0 / T 2 B_{0}/T^{2} B 0 / T 2 . The chemical potential is set zero. (a) m / T = 1 m/T=1 m / T = 1 , black solid line is for σ ∥ / ( T 2 τ ) \sigma_{\parallel}/(T^{2}\tau) σ ∥ / ( T 2 τ ) which does not depend on T τ T\tau T τ , and the red, green and blue dotted lines are σ ⊥ / ( T 2 τ ) \sigma_{\perp}/(T^{2}\tau) σ ⊥ / ( T 2 τ ) for T τ = 0.1 T\tau=0.1 T τ = 0.1 , 1.0 1.0 1.0 and 2.0 2.0 2.0 , respectively. (b) T τ = 1 T\tau=1 T τ = 1 and m / T = 0.1 m/T=0.1 m / T = 0.1 (red), m / T = 1.0 m/T=1.0 m / T = 1.0 (green), and m / T = 2.0 m/T=2.0 m / T = 2.0 (blue).
Figure 3. The long-time asymptotic values of σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) (79) as functions of B 0 / T 2 B_{0}/T^{2} B 0 / T 2 with different set of parameter values. (a) μ / T = 0.1 \mu/T=0.1 μ / T = 0.1 , m / T = 1 m/T=1 m / T = 1 , T τ = 0.1 T\tau=0.1 T τ = 0.1 (red), 1.0 1.0 1.0 (green), 2.0 2.0 2.0 (blue); (b) μ / T = 0.1 \mu/T=0.1 μ / T = 0.1 , T τ = 1.0 T\tau=1.0 T τ = 1.0 , m / T = 0.1 m/T=0.1 m / T = 0.1 (red), 1.0 1.0 1.0 (green), 2.0 2.0 2.0 (blue).
In Figure 3 , we show the long-time asymptotic values of σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) in Eq. (79) as functions of B 0 / T 2 B_{0}/T^{2} B 0 / T 2 . In Figure 3 (a), we set μ / T = 0.1 \mu/T=0.1 μ / T = 0.1 , m / T = 1 m/T=1 m / T = 1 , and T τ = 0.1 T\tau=0.1 T τ = 0.1 , 1.0 1.0 1.0 , 2.0 2.0 2.0 . In the weak-field limit we have σ H ∝ τ 2 \sigma_{H}\propto\tau^{2} σ H ∝ τ 2 , and then σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) is independent of T τ T\tau T τ . The behaviors of σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) versus B 0 / T 2 B_{0}/T^{2} B 0 / T 2 are sensitive to the values of T τ T\tau T τ . When T τ = 2.0 T\tau=2.0 T τ = 2.0 , σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) increases and then decreases with B 0 / T 2 B_{0}/T^{2} B 0 / T 2 . When T τ = 1.0 T\tau=1.0 T τ = 1.0 , σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) increases and then approaches a constant value with B 0 / T 2 B_{0}/T^{2} B 0 / T 2 . When T τ = 0.1 T\tau=0.1 T τ = 0.1 , σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) always increases with B 0 / T 2 B_{0}/T^{2} B 0 / T 2 . Figure 3 (b) shows σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) as functions of B 0 / T 2 B_{0}/T^{2} B 0 / T 2 at μ / T = 0.1 \mu/T=0.1 μ / T = 0.1 , T τ = 1.0 T\tau=1.0 T τ = 1.0 , and m / T = 0.1 m/T=0.1 m / T = 0.1 , 1.0 1.0 1.0 , 2.0 2.0 2.0 . For all cases of Figure 3 (a,b), σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) is proportional to B 0 / T 2 B_{0}/T^{2} B 0 / T 2 in the weak field limit. We also find that a large particle mass gives a small σ H / ( T 3 τ 2 ) \sigma_{H}/(T^{3}\tau^{2}) σ H / ( T 3 τ 2 ) .
Axial-vector current induced by perturbation fields
In this section, we solve the Boltzmann equation (55) for δ A μ \delta A^\mu δ A μ and calculate the axial-current defined as
J A μ ( x ) ≡ ∫ d 4 p δ A μ ( x , p ) . (80) J^\mu_A(x) \equiv\int d^4p\,\delta A^\mu(x,p). \tag*{(80)} J A μ ( x ) ≡ ∫ d 4 p δ A μ ( x , p ) . (80)
Similar to the decomposition for δ V μ \delta V^\mu δ V μ in Eq. (62) on the bases e i μ ≡ { u μ , b μ , p ⊥ μ , ϵ μ ν ρ σ p ν u ρ b σ } e^\mu_i \equiv\{u^\mu,b^\mu,p^\mu_\perp,\epsilon^{\mu\nu\rho\sigma}p_{\nu}u_{\rho}b_{\sigma}\} e i μ ≡ { u μ , b μ , p ⊥ μ , ϵ μν ρ σ p ν u ρ b σ } , we can also decompose δ A μ \delta A^\mu δ A μ as
δ A μ = u μ δ A 1 − b μ δ A 2 − 1 p T 2 [ p ⊥ μ δ A 3 ⊥ + ϵ μ ν ρ σ p ⊥ ν u ρ b σ δ A 4 ⊥ ] . (81) \delta A^\mu=u_\mu\delta A_1-b_\mu\delta A_2-\frac{1}{p_T^2}\left[p^\mu_\perp\delta A_3^\perp+\epsilon^{\mu\nu\rho\sigma}p_{\perp\nu}u_\rho b_\sigma\delta A_4^\perp\right]. \tag*{(81)} δ A μ = u μ δ A 1 − b μ δ A 2 − p T 2 1 [ p ⊥ μ δ A 3 ⊥ + ϵ μν ρ σ p ⊥ ν u ρ b σ δ A 4 ⊥ ] . (81)
The dynamical equation for each component reads
[ ( u ⋅ p ) ( d d t + 1 τ ) − B 0 ϵ ν ρ α β p ν ∂ p ρ u α b β ] δ A i = e i μ ( f μ ρ A e q ρ + p ν f ν ρ ∂ p ρ A μ e q ) . (82) \left[(u\cdot p)\left(\frac{d}{dt}+\frac{1}{\tau}\right)-B_0\epsilon^{\nu\rho\alpha\beta}p_\nu\partial_{p^\rho}u_\alpha b_\beta\right]\delta A_i=e^\mu_i\left(f_{\mu\rho}A^\rho_{\rm eq}+p^\nu f_{\nu\rho}\partial_p^\rho A^{\rm eq}_\mu\right). \tag*{(82)} [ ( u ⋅ p ) ( d t d + τ 1 ) − B 0 ϵ ν ρ α β p ν ∂ p ρ u α b β ] δ A i = e i μ ( f μ ρ A eq ρ + p ν f ν ρ ∂ p ρ A μ eq ) . (82)
where the source term on the right-hand-side can be expressed as
e i μ ( f μ ρ A e q ρ + p ν f ν ρ ∂ p ρ A μ e q ) = D i ( 1 ) ( t , p T , u ⋅ p , b ⋅ p ) + p ⊥ μ D i , μ ( 2 ) ( t , p T , u ⋅ p , b ⋅ p ) . (83) e^\mu_i\left(f_{\mu\rho}A^\rho_{\rm eq}+p^\nu f_{\nu\rho}\partial_p^\rho A^{\rm eq}_\mu\right)=D_i^{(1)}(t,p_T,u\cdot p,b\cdot p)+p^\mu_\perp D^{(2)}_{i,\mu}(t,p_T,u\cdot p,b\cdot p). \tag*{(83)} e i μ ( f μ ρ A eq ρ + p ν f ν ρ ∂ p ρ A μ eq ) = D i ( 1 ) ( t , p T , u ⋅ p , b ⋅ p ) + p ⊥ μ D i , μ ( 2 ) ( t , p T , u ⋅ p , b ⋅ p ) . (83)
Similar to J μ ( x ) J^\mu(x) J μ ( x ) in Eq. (69) , the axial-vector current is given by
J A μ ( x ) = u μ ∫ d 4 p g 1 ( 1 ) − b μ ∫ d 4 p g 2 ( 1 ) + 1 2 ∫ d 4 p [ g 3 ( 2 ) μ + ϵ μ ν ρ σ u ρ b σ g 4 , ν ( 2 ) ] . (84) J^\mu_A(x)=u^\mu\int d^4p\,g_1^{(1)}-b^\mu\int d^4p\,g_2^{(1)}
+\frac{1}{2}\int d^4p\left[g_3^{(2)\mu}+\epsilon^{\mu\nu\rho\sigma}u_\rho b_\sigma g^{(2)}_{4,\nu}\right]. \tag*{(84)} J A μ ( x ) = u μ ∫ d 4 p g 1 ( 1 ) − b μ ∫ d 4 p g 2 ( 1 ) + 2 1 ∫ d 4 p [ g 3 ( 2 ) μ + ϵ μν ρ σ u ρ b σ g 4 , ν ( 2 ) ] . (84)
where the functions g i ( 1 ) g_i^{(1)} g i ( 1 ) and g i , μ ( 2 ) g_{i,\mu}^{(2)} g i , μ ( 2 ) are derived from D i ( 1 ) D_i^{(1)} D i ( 1 ) and D i , μ ( 2 ) D_{i,\mu}^{(2)} D i , μ ( 2 ) in a similar way to Eq. (67) . Using A e q μ A^\mu_{\rm eq} A eq μ in Eq. (29) , we obtain explicit expressions for D 1 ( 1 ) D_1^{(1)} D 1 ( 1 ) , D 2 ( 1 ) D_2^{(1)} D 2 ( 1 ) , D 3 ( 2 ) μ D_3^{(2)\mu} D 3 ( 2 ) μ , and D 4 ( 2 ) μ D_4^{(2)\mu} D 4 ( 2 ) μ as
D 1 ( 1 ) = − ( b ⋅ p ) p ∥ ν f ν ρ ∑ n = 0 ∞ [ ∂ p ρ V n ( p ) ] Λ − ( n ) ( p T ) , D_1^{(1)}=-(b\cdot p)p_\parallel^\nu f_{\nu\rho}\sum_{n=0}^{\infty}\left[\partial_p^\rho V_n(p)\right]\Lambda_-^{(n)}(p_T), D 1 ( 1 ) = − ( b ⋅ p ) p ∥ ν f ν ρ n = 0 ∑ ∞ [ ∂ p ρ V n ( p ) ] Λ − ( n ) ( p T ) ,
D 2 ( 1 ) = − ( u ⋅ p ) p ∥ ν f ν ρ ∑ n = 0 ∞ [ ∂ p ρ V n ( p ) ] Λ − ( n ) ( p T ) , D_2^{(1)}=-(u\cdot p)p_\parallel^\nu f_{\nu\rho}\sum_{n=0}^{\infty}\left[\partial_p^\rho V_n(p)\right]\Lambda_-^{(n)}(p_T), D 2 ( 1 ) = − ( u ⋅ p ) p ∥ ν f ν ρ n = 0 ∑ ∞ [ ∂ p ρ V n ( p ) ] Λ − ( n ) ( p T ) ,
D 3 ( 2 ) μ = Ξ μ ν f ν ρ [ ( u ⋅ p ) b ρ − ( b ⋅ p ) u ρ ] ∑ n = 0 ∞ V n ( p ) Λ − ( n ) ( p T ) , D_3^{(2)\mu}=\Xi^{\mu\nu}f_{\nu\rho}\left[(u\cdot p)b^\rho-(b\cdot p)u^\rho\right]\sum_{n=0}^{\infty}V_n(p)\Lambda_-^{(n)}(p_T), D 3 ( 2 ) μ = Ξ μν f ν ρ [ ( u ⋅ p ) b ρ − ( b ⋅ p ) u ρ ] n = 0 ∑ ∞ V n ( p ) Λ − ( n ) ( p T ) ,
D 4 ( 2 ) μ = − 1 2 Ξ λ μ ϵ λ ν ρ σ p ∥ ν f ρ σ ∑ n = 0 ∞ V n ( p ) Λ − ( n ) ( p T ) . (85) D_4^{(2)\mu}=-\frac{1}{2}\Xi^\mu_{\ \lambda}\epsilon^{\lambda\nu\rho\sigma}p_{\parallel\nu}f_{\rho\sigma}\sum_{n=0}^{\infty}V_n(p)\Lambda_-^{(n)}(p_T). \tag*{(85)} D 4 ( 2 ) μ = − 2 1 Ξ λ μ ϵ λ ν ρ σ p ∥ ν f ρ σ n = 0 ∑ ∞ V n ( p ) Λ − ( n ) ( p T ) . (85)
Then by making replacements C i ( 1 ) → D i ( 1 ) C_i^{(1)}\to D_i^{(1)} C i ( 1 ) → D i ( 1 ) , C i , μ ( 2 ) → D i , μ ( 2 ) C_{i,\mu}^{(2)}\to D_{i,\mu}^{(2)} C i , μ ( 2 ) → D i , μ ( 2 ) , f i ( 1 ) → g i ( 1 ) f_i^{(1)}\to g_i^{(1)} f i ( 1 ) → g i ( 1 ) and f i , μ ( 2 ) → g i , μ ( 2 ) f_{i,\mu}^{(2)}\to g_{i,\mu}^{(2)} f i , μ ( 2 ) → g i , μ ( 2 ) in Eq. (67) , we are able to derive g 1 ( 1 ) g_1^{(1)} g 1 ( 1 ) , g 2 ( 1 ) g_2^{(1)} g 2 ( 1 ) , g 3 , μ ( 2 ) g_{3,\mu}^{(2)} g 3 , μ ( 2 ) and g 4 , μ ( 2 ) g_{4,\mu}^{(2)} g 4 , μ ( 2 ) . Substituting them into Eq. (84) , we obtain
J A μ ( x ) = − u μ b ν B 0 ( 2 π ) 2 ∫ t 0 t d t ′ e ( t ′ − t ) / τ δ E ν ∫ d ( b ⋅ p ) × lim ε → ε b ⋅ p ( 0 ) m 2 ε 3 [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] − B 0 ( 2 π ) 2 Ξ μ ν ∫ t 0 t d t ′ e ( t ′ − t ) / τ δ B ν ∫ d ( b ⋅ p ) × lim ε → ε b ⋅ p ( 0 ) 1 ε sin [ B 0 ε ( t ′ − t ) ] [ f e q ( + ) ( ε ) − f e q ( − ) ( ε ) ] + 1 ( 2 π ) 2 τ ϵ μ ν ρ σ u ρ b σ ∫ t 0 t d t ′ e ( t ′ − t ) / τ [ ( d d t ′ + 1 τ ) δ B ν ] ∫ d ( b ⋅ p ) × lim ε → ε b ⋅ p ( 0 ) sin [ B 0 ε ( t ′ − t ) ] [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] . (86) \begin{aligned}
J_A^\mu(x)={}&-u^\mu b^\nu\frac{B_0}{(2\pi)^2}\int_{t_0}^{t}dt'e^{(t'-t)/\tau}\delta E_\nu\int d(b\cdot p)\\
&\times\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(0)}}\frac{m^2}{\varepsilon^3}\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)+f_{\mathrm{eq}}^{(-)}(\varepsilon)\right]\\
&-\frac{B_0}{(2\pi)^2}\Xi^{\mu\nu}\int_{t_0}^{t}dt'e^{(t'-t)/\tau}\delta B_\nu\int d(b\cdot p)\\
&\times\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(0)}}\frac{1}{\varepsilon}\sin\left[\frac{B_0}{\varepsilon}(t'-t)\right]\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)-f_{\mathrm{eq}}^{(-)}(\varepsilon)\right]\\
&+\frac{1}{(2\pi)^2\tau}\epsilon^{\mu\nu\rho\sigma}u_\rho b_\sigma\int_{t_0}^{t}dt'e^{(t'-t)/\tau}\left[\left(\frac{d}{dt'}+\frac{1}{\tau}\right)\delta B_\nu\right]\int d(b\cdot p)\\
&\times\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(0)}}\sin\left[\frac{B_0}{\varepsilon}(t'-t)\right]\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)+f_{\mathrm{eq}}^{(-)}(\varepsilon)\right]. \tag*{(86)}
\end{aligned} J A μ ( x ) = − u μ b ν ( 2 π ) 2 B 0 ∫ t 0 t d t ′ e ( t ′ − t ) / τ δ E ν ∫ d ( b ⋅ p ) × ε → ε b ⋅ p ( 0 ) lim ε 3 m 2 [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] − ( 2 π ) 2 B 0 Ξ μν ∫ t 0 t d t ′ e ( t ′ − t ) / τ δ B ν ∫ d ( b ⋅ p ) × ε → ε b ⋅ p ( 0 ) lim ε 1 sin [ ε B 0 ( t ′ − t ) ] [ f eq ( + ) ( ε ) − f eq ( − ) ( ε ) ] + ( 2 π ) 2 τ 1 ϵ μν ρ σ u ρ b σ ∫ t 0 t d t ′ e ( t ′ − t ) / τ [ ( d t ′ d + τ 1 ) δ B ν ] ∫ d ( b ⋅ p ) × ε → ε b ⋅ p ( 0 ) lim sin [ ε B 0 ( t ′ − t ) ] [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] . (86)
Assuming that δ E μ \delta E^\mu δ E μ and δ B μ \delta B^\mu δ B μ are proportional to θ ( t ) \theta(t) θ ( t ) and they maintain constants at t > 0 t>0 t > 0 , we can express J A μ J_A^\mu J A μ as
J A μ = − u μ b ν δ E ν χ A + Ξ μ ν δ B ν χ ∥ + ϵ μ ν ρ σ u ρ b σ δ B ν χ ⊥ , (87) J_A^\mu=-u^\mu b^\nu\delta E_\nu\chi_A+\Xi^{\mu\nu}\delta B_\nu\chi_\parallel+\epsilon^{\mu\nu\rho\sigma}u_\rho b_\sigma\delta B_\nu\chi_\perp, \tag*{(87)} J A μ = − u μ b ν δ E ν χ A + Ξ μν δ B ν χ ∥ + ϵ μν ρ σ u ρ b σ δ B ν χ ⊥ , (87)
where we still used δ E ν \delta E_\nu δ E ν and δ B ν \delta B_\nu δ B ν but to represent the constant magnitudes of the electric and magentic field perturbation respectively. The conductivity χ A \chi_A χ A denotes the axial charge induced by the longitudinal electric field b ν δ E ν b^\nu\delta E_\nu b ν δ E ν , and conductivities χ ∥ \chi_\parallel χ ∥ and χ ⊥ \chi_\perp χ ⊥ correspond to the axial current induced by δ B ν \delta B_\nu δ B ν in the direction Ξ μ ν δ B ν \Xi^{\mu\nu}\delta B_\nu Ξ μν δ B ν and ϵ μ ν ρ σ u ρ b σ δ B ν \epsilon^{\mu\nu\rho\sigma}u_\rho b_\sigma\delta B_\nu ϵ μν ρ σ u ρ b σ δ B ν , respectively, which are given by
χ A = B 0 R 1 ( t ) ( 2 π ) 2 ∫ d ( b ⋅ p ) lim ε → ε b ⋅ p ( 0 ) m 2 ε 3 [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] , (88) \chi_A=\frac{B_0\mathcal{R}_1(t)}{(2\pi)^2}\int d(b\cdot p)\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(0)}}\frac{m^2}{\varepsilon^3}\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)+f_{\mathrm{eq}}^{(-)}(\varepsilon)\right], \tag*{(88)} χ A = ( 2 π ) 2 B 0 R 1 ( t ) ∫ d ( b ⋅ p ) ε → ε b ⋅ p ( 0 ) lim ε 3 m 2 [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] , (88)
χ ∥ = − B 0 ( 2 π ) 2 ∫ d ( b ⋅ p ) lim ε → ε b ⋅ p ( 0 ) 1 ε R 2 ( ε , B 0 , t ) [ f e q ( + ) ( ε ) − f e q ( − ) ( ε ) ] , (89) \chi_\parallel=-\frac{B_0}{(2\pi)^2}\int d(b\cdot p)\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(0)}}\frac{1}{\varepsilon}\mathcal{R}_2(\varepsilon,B_0,t)\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)-f_{\mathrm{eq}}^{(-)}(\varepsilon)\right], \tag*{(89)} χ ∥ = − ( 2 π ) 2 B 0 ∫ d ( b ⋅ p ) ε → ε b ⋅ p ( 0 ) lim ε 1 R 2 ( ε , B 0 , t ) [ f eq ( + ) ( ε ) − f eq ( − ) ( ε ) ] , (89)
χ ⊥ = 1 ( 2 π ) 2 τ ∫ d ( b ⋅ p ) lim ε → ε b ⋅ p ( 0 ) R 2 ( ε , B 0 , t ) [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] , (90) \chi_\perp=\frac{1}{(2\pi)^2\tau}\int d(b\cdot p)\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(0)}}\mathcal{R}_2(\varepsilon,B_0,t)\left[f_{\mathrm{eq}}^{(+)}(\varepsilon)+f_{\mathrm{eq}}^{(-)}(\varepsilon)\right], \tag*{(90)} χ ⊥ = ( 2 π ) 2 τ 1 ∫ d ( b ⋅ p ) ε → ε b ⋅ p ( 0 ) lim R 2 ( ε , B 0 , t ) [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] , (90)
where ε b ⋅ p ( 0 ) = m 2 + ( b ⋅ p ) 2 \varepsilon_{b\cdot p}^{(0)}=\sqrt{m^2+(b\cdot p)^2} ε b ⋅ p ( 0 ) = m 2 + ( b ⋅ p ) 2 is the on-shell energy at the lowest Landau level and R 1 ( t ) \mathcal{R}_1(t) R 1 ( t ) , R 2 ( ε , B 0 , t ) \mathcal{R}_2(\varepsilon,B_0,t) R 2 ( ε , B 0 , t ) are defined in Eq. (76) . We note that only the lowest Landau level contributes to these quantities, because higher Landau levels are degenerated with respect to spin and thus do not contribute to the axial current.
Similar to σ ∥ \sigma_\parallel σ ∥ , σ ⊥ \sigma_\perp σ ⊥ , and σ H \sigma_H σ H in Eqs. (77) -(79) , the conductivities χ A \chi_A χ A , χ ∥ \chi_\parallel χ ∥ , and χ ⊥ \chi_\perp χ ⊥ also approach to their long-time asymptotic values that are given by
lim t → ∞ χ A = B 0 τ ( 2 π ) 2 ∫ d ( b ⋅ p ) lim ε → ε b ⋅ p ( 0 ) m 2 ε 3 [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] , (91) \lim_{t\to\infty}\chi_A=\frac{B_0\tau}{(2\pi)^2}\int d(b\cdot p)\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(0)}}\frac{m^2}{\varepsilon^3}\left[f_{\rm eq}^{(+)}(\varepsilon)+f_{\rm eq}^{(-)}(\varepsilon)\right], \tag*{(91)} t → ∞ lim χ A = ( 2 π ) 2 B 0 τ ∫ d ( b ⋅ p ) ε → ε b ⋅ p ( 0 ) lim ε 3 m 2 [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] , (91)
lim t → ∞ χ ∥ = B 0 2 τ 2 ( 2 π ) 2 ∫ d ( b ⋅ p ) lim ε → ε b ⋅ p ( 0 ) 1 ε 2 + B 0 2 τ 2 [ f e q ( + ) ( ε ) − f e q ( − ) ( ε ) ] , (92) \lim_{t\to\infty}\chi_{\parallel}=\frac{B_0^2\tau^2}{(2\pi)^2}\int d(b\cdot p)\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(0)}}\frac{1}{\varepsilon^2+B_0^2\tau^2}\left[f_{\rm eq}^{(+)}(\varepsilon)-f_{\rm eq}^{(-)}(\varepsilon)\right], \tag*{(92)} t → ∞ lim χ ∥ = ( 2 π ) 2 B 0 2 τ 2 ∫ d ( b ⋅ p ) ε → ε b ⋅ p ( 0 ) lim ε 2 + B 0 2 τ 2 1 [ f eq ( + ) ( ε ) − f eq ( − ) ( ε ) ] , (92)
lim t → ∞ χ ⊥ = − B 0 τ ( 2 π ) 2 ∫ d ( b ⋅ p ) lim ε → ε b ⋅ p ( 0 ) ε ε 2 + B 0 2 τ 2 [ f e q ( + ) ( ε ) + f e q ( − ) ( ε ) ] . (93) \lim_{t\to\infty}\chi_{\perp}=-\frac{B_0\tau}{(2\pi)^2}\int d(b\cdot p)\lim_{\varepsilon\to\varepsilon_{b\cdot p}^{(0)}}\frac{\varepsilon}{\varepsilon^2+B_0^2\tau^2}\left[f_{\rm eq}^{(+)}(\varepsilon)+f_{\rm eq}^{(-)}(\varepsilon)\right]. \tag*{(93)} t → ∞ lim χ ⊥ = − ( 2 π ) 2 B 0 τ ∫ d ( b ⋅ p ) ε → ε b ⋅ p ( 0 ) lim ε 2 + B 0 2 τ 2 ε [ f eq ( + ) ( ε ) + f eq ( − ) ( ε ) ] . (93)
Note that χ A \chi_A χ A is exactly the same as the lowest Landau level contribution to σ ∥ \sigma_{\parallel} σ ∥ . This is because particles at the lowest Landau level always have spin along the direction of b μ b^\mu b μ , thus they are right-handed when their momenta are parallel to b μ b^\mu b μ , and left-handed when their momenta are anti-parallel to b μ b^\mu b μ . Therefore these particles have the same contribution to the longitudinal vector current and the axial charge.
Conclusion
We study the linear response of a fermion system to the perturbative electromagnetic field in a constant magnetic background field based on the Wigner function in relaxation time approximation. The Wigner function is separated into an equilibrium and an off-equilibrium part. The equilibrium part is the solution to the Wigner function in the constant magnetic background field, which depends on the thermodynamical variables.
The off-equilibrium part of the vector and axial vector components induced by the field perturbation can be determined by solving kinetic or Boltzmann equations. They also satisfy the mass-shell constraints. In the case that the field perturbation only depends on the proper time, the kinetic equations can be solved analytically. The vector current only depends on the perturbative electric field, with the linear response are described by longitudinal and transverse Ohm conductivities as well as the Hall conductivity. As an example, we assume that the perturbative electric field is turned on at the initial proper time and maintains constant at later time, we numerically calculate these conductivities. In the long time limit, these conductivities approach to asymptotic values which depend on the relaxation time, the particle mass and the background magnetic field. The contributions from all Landau levels are included and therefore the results in this paper can be applied to any strength of the magnetic field. The results for Ohm conductivities qualitatively agree with those from the lattice QCD [72 ] and the holographic model [76 ] . We also obtain the analytical solution to the axial vector current as the linear response to the perturbative electric and magnetic fields.
Acknowledgments
The authors thank Lixin Yang for enlightening discussions. H.H.P, S.P. and Q.W. are supported by National Natural Science Foundation of China (NSFC) under Grants No. 12135011, 12075235 and 11890713 (a subgrant of 11890710). X.L.S. is supported by the Project funded by China Postdoctoral Science Foundation under grant No. 2021M701369.
References