betaFeedback

WorksPhysics and Astronomy

The Connection between the Stochastic Schrödinger Equation and Boltzmann Equation

Abstract

The heavy quarks present in the quark-gluon plasma (QGP) can act as a probe of relativistic heavy ion collisions as they retain the memory of their interaction history. In a previous study, a stochastic Schrödinger equation (SSE) has been applied to describe the transport process of heavy quarks, where an external field with random phases is used to simulate the thermal medium. In this work, we study the connection between the SSE and the Boltzmann equation (BE) approach in the Keldysh Green's function formalism. By comparing the Green's function of the heavy quark from the SSE and the Keldysh Green's functions leading to the Boltzmann equation, we demonstrate that the SSE is consistent with the Boltzmann equation in the weak coupling limit. We subsequently confirm their consistency through numerical calculations.

Introduction

In ultra-relativistic heavy-ion collisions (URHICs), a significant portion of the kinetic energy from the colliding nuclei is transformed into thermal energy, leading to the creation of quark-gluon plasma (QGP) from the strongly interacting matter. At high temperatures, quarks and gluons behave like free particles comprising the components of QGP. The expansion and cooling of QGP are driven by pressure, leading to the formation of the hadron gas. The thermalization time of QGP is typically within the range of 5-10 fm/c [1, 2]. While light quarks like uu and dd quarks thermalize at around 0.5 fm/c, which is much shorter than the lifetime of QGP, heavy quarks have a comparable lifetime due to their larger mass; thus, they retain the memory of their interaction history. As a result, they act as a “probe” of relativistic heavy ion collisions [3, 4, 5, 6]. In recent years, various descriptions of heavy quarks’ transport process in QGP (or other thermal media) have been developed, including the Boltzmann equation (Fokker-Planck equation) [1, 7, 8], Schrödinger-Langevin equation [9, 10], and the Lindblad equation [11, 12, 13]. Additionally, the Schwinger-Keldysh formalism has been used to describe the evolution of non-equilibrium systems [14, 15]. In this formalism, the Keldysh Green’s functions are introduced. The Boltzmann equation can be derived by performing perturbative expansion of the Keldysh Green’s functions. The Schwinger-Keldysh formalism serves as a bridge connecting quantum field theory and transport theory.

Although the Boltzmann equation can describe the evolution of heavy quarks in a thermal medium, it only takes into account a partial set of quantum effects through the scattering matrix element and the Pauli blocking factor. In order to overcome these limitations, the authors in [16] constructed a new approach to describe a heavy quark’s evolution in a thermal medium using a stochastic Schrödinger Equation (SSE). In that work, the authors introduced a classical gluon field to model the thermal medium and random phases for the gluon field to simulate the thermal fluctuation of the medium on the amplitude level. The interaction between this gluon field and the heavy quark was described using the SSE. On this basis, the authors in [16] further introduced a time correlation between random phases of the gluon field at different times, which characterizes the time scale for the evolution of the random phases of the gluon field. Based on numerically comparing the time evolution of the average momentum squared and average displacement squared with those from the Langevin equation approach, the authors suggested that the SSE approach can be used to describe the evolution of a heavy quark in the thermal medium. Because the SSE approach describes the evolution on the amplitude level, it can potentially capture more complete quantum effects compared to the Boltzmann equation (BE) approach. The explicit connection between the SSE and BE approach was however not shown in [16].

In this paper, we demonstrate the connection between the SSE and the Boltzmann equation using the Schwinger-Keldysh formalism and subsequently verify their consistency through numerical calculations. We first briefly review the Schwinger-Keldysh approach in Sec.2. Then we demonstrate the connection between the Boltzmann equation and the stochastic Schrödinger equation (SSE) in the Schwinger-Keldysh formalism in Sec.3. By considering the time correlation of random phases, we find that the SSE describes the process of an on-shell heavy quark absorbing (or emitting) an off-shell gluon and forming a final-state on-shell heavy quark. Next, we numerically calculate the evolution of the heavy quark momentum distributions in the Boltzmann equation and compare with those from the SSE in Sec.4. Finally, we conclude in Sec.5.

The Keldysh Green's function and the Boltzmann equation

The Green’s function (or propagator) is a mathematical tool in quantum field theory to calculate the probability of particle scattering and spectral distributions in various systems, at both zero-temperature [17] and finite temperature [18]. To extend the Green’s functions to non-equilibrium systems, Schwinger and Keldysh introduced the Keldysh Green’s functions, and the Keldysh time contour [14, 15], as illustrated in Figure 1. The Keldysh Green’s functions play a central role in illustrating the connection between the SSE and the Boltzmann equation approach. Before introducing this connection, we briefly review the properties of the Keldysh Green’s functions and the derivation of the Boltzmann equation using the Keldysh Green’s functions [19, 20].

Keldysh time contour

Figure 1. The Keldysh time contour. τ\tau and τ′\tau' denote the time coordinates of the Keldysh Green’s functions, see text for the detail.

The Keldysh Green’s functions for spin-0 bosons are [19]:

GB−−(x,y)=−i⟨T[ϕ(x)ϕ†(y)]⟩,GB−+(x,y)=−i⟨T~[ϕ(x)ϕ†(y)]⟩,GB+−(x,y)=−i⟨ϕ(x)ϕ†(y)⟩,GB++(x,y)=−i⟨ϕ†(y)ϕ(x)⟩.(1)\begin{aligned} G_B^{--}(x,y)&=-i\langle T[\phi(x)\phi^\dagger(y)]\rangle,\\ G_B^{-+}(x,y)&=-i\langle\widetilde{T}[\phi(x)\phi^\dagger(y)]\rangle,\\ G_B^{+-}(x,y)&=-i\langle\phi(x)\phi^\dagger(y)\rangle,\\ G_B^{++}(x,y)&=-i\langle\phi^\dagger(y)\phi(x)\rangle. \tag*{(1)} \end{aligned}

Here xx and yy denote the positions in space-time, ⟨...⟩\langle...\rangle denotes the ensemble average and TT (T~\widetilde{T}) denotes the (anti-)time-ordered product. For spin-12\frac{1}{2} fermions, the Green’s functions are:

GF−−(x,y)=−i⟨T[ψ(x)ψˉ(y)]⟩,GF−+(x,y)=−i⟨T~[ψ(x)ψˉ(y)]⟩,GF+−(x,y)=−i⟨ψ(x)ψˉ(y)⟩,GF++(x,y)=i⟨ψˉ(y)ψ(x)⟩.(2)\begin{aligned} G_F^{--}(x,y)&=-i\langle T[\psi(x)\bar{\psi}(y)]\rangle,\\ G_F^{-+}(x,y)&=-i\langle\widetilde{T}[\psi(x)\bar{\psi}(y)]\rangle,\\ G_F^{+-}(x,y)&=-i\langle\psi(x)\bar{\psi}(y)\rangle,\\ G_F^{++}(x,y)&=i\langle\bar{\psi}(y)\psi(x)\rangle. \tag*{(2)} \end{aligned}

In the above definitions of Green’s functions, the boson field ϕ(x)=ϕ(x⃗,t)\phi(x)=\phi(\vec{x},t) and its complex conjugation ϕ†(x)\phi^\dagger(x) can be expanded in momentum space as

ϕ(x⃗,t)=∫d3p⃗(2π)32Ep⃗(a^p⃗eip⃗⋅x⃗−iEp⃗t+a^p⃗†e−ip⃗⋅x⃗+iEp⃗t),\phi(\vec{x},t)=\int\frac{d^3\vec{p}}{(2\pi)^3\sqrt{2E_{\vec{p}}}}\left(\hat{a}_{\vec{p}}e^{i\vec{p}\cdot\vec{x}-iE_{\vec{p}}t}+\hat{a}^\dagger_{\vec{p}}e^{-i\vec{p}\cdot\vec{x}+iE_{\vec{p}}t}\right),
ϕ†(x⃗,t)=∫d3p⃗(2π)32Ep⃗(a^p⃗†e−ip⃗⋅x⃗+iEp⃗t+a^p⃗eip⃗⋅x⃗−iEp⃗t).(3)\phi^\dagger(\vec{x},t)=\int\frac{d^3\vec{p}}{(2\pi)^3\sqrt{2E_{\vec{p}}}}\left(\hat{a}^\dagger_{\vec{p}}e^{-i\vec{p}\cdot\vec{x}+iE_{\vec{p}}t}+\hat{a}_{\vec{p}}e^{i\vec{p}\cdot\vec{x}-iE_{\vec{p}}t}\right). \tag*{(3)}

Here Ep⃗=p⃗ 2+m2E_{\vec{p}}=\sqrt{\vec{p}^{\,2}+m^2} is the kinetic energy of the momentum mode p⃗\vec{p}, mm is the mass of the boson, p=(Ep⃗,p⃗)p=(E_{\vec{p}},\vec{p}) is the 4-momentum of the mode p⃗\vec{p}. a^p⃗\hat{a}_{\vec{p}} (a^p⃗†\hat{a}^\dagger_{\vec{p}}) is the annihilation (creation) operator of the boson and satisfies the following commutation relation

[a^p⃗,a^p⃗ ′†]=(2π)3δ3(p⃗−p⃗ ′).(4)[\hat{a}_{\vec{p}},\hat{a}^\dagger_{\vec{p}^{\,\prime}}]=(2\pi)^3\delta^3(\vec{p}-\vec{p}^{\,\prime}). \tag*{(4)}

The fermion fields ψ(x)\psi(x) and ψˉ(x)\bar{\psi}(x) can be similarly expanded as

ψ(x⃗,t)=∫d3p⃗(2π)32Ep⃗∑s(b^p⃗sus(p)eip⃗⋅x⃗−iEp⃗t+d^p⃗s†vs(p)e−ip⃗⋅x⃗+iEp⃗t),\psi(\vec{x},t)=\int\frac{d^3\vec{p}}{(2\pi)^3\sqrt{2E_{\vec{p}}}}\sum_s\left(\hat{b}^s_{\vec{p}}u^s(p)e^{i\vec{p}\cdot\vec{x}-iE_{\vec{p}}t}+\hat{d}^{s\dagger}_{\vec{p}}v^s(p)e^{-i\vec{p}\cdot\vec{x}+iE_{\vec{p}}t}\right),
ψˉ(x⃗,t)=∫d3p⃗(2π)32Ep⃗∑s(b^p⃗s†uˉs(p)e−ip⃗⋅x⃗+iEp⃗t+d^p⃗svˉs(p)eip⃗⋅x⃗−iEp⃗t).(5)\bar{\psi}(\vec{x},t)=\int\frac{d^3\vec{p}}{(2\pi)^3\sqrt{2E_{\vec{p}}}}\sum_s\left(\hat{b}^{s\dagger}_{\vec{p}}\bar{u}^s(p)e^{-i\vec{p}\cdot\vec{x}+iE_{\vec{p}}t}+\hat{d}^s_{\vec{p}}\bar{v}^s(p)e^{i\vec{p}\cdot\vec{x}-iE_{\vec{p}}t}\right). \tag*{(5)}

Here ss is the spin index. b^p⃗s\hat{b}^s_{\vec{p}} (b^p⃗s†\hat{b}^{s\dagger}_{\vec{p}}) is the annihilation (creation) operator of the spin-12\frac{1}{2} fermion with momentum p⃗\vec{p}, and d^p⃗s\hat{d}^s_{\vec{p}} (d^p⃗s†\hat{d}^{s\dagger}_{\vec{p}}) is the annihilation (creation) operator of the spin-12\frac{1}{2} antifermion. These operators satisfy the following anti-commutation relation

{b^p⃗r,b^p⃗ ′s†}={d^p⃗r,d^p⃗ ′s†}=(2π)3δ3(p⃗−p⃗ ′)δrs.(6)\{\hat{b}^r_{\vec{p}},\hat{b}^{s\dagger}_{\vec{p}^{\,\prime}}\}=\{\hat{d}^r_{\vec{p}},\hat{d}^{s\dagger}_{\vec{p}^{\,\prime}}\}=(2\pi)^3\delta^3(\vec{p}-\vec{p}^{\,\prime})\delta^{rs}. \tag*{(6)}

Here us(p)u^s(p) and vs(p)v^s(p), s=1,2s=1,2 are two linearly independent spinors for the fermion and antifermion, respectively. us(p)e−ip⋅xu^s(p)e^{-ip\cdot x} and vs(p)eip⋅xv^s(p)e^{ip\cdot x} are the corresponding plane-wave solutions of the Dirac equation.

In the definition of the Keldysh Green’s functions in Eq. (1) and Eq. (2), the superscripts “++” and “−-” signify the respective branch of the time contour C\mathbb{C} where the time coordinates τ\tau and τ′\tau' of the Green’s function Gαβ(τ,x⃗;τ′,x⃗′)G^{\alpha\beta}(\tau,\vec{x};\tau',\vec{x}') reside: “++” represents the branch from −∞-\infty to −∞-\infty (C+\mathbb{C}_+), and “−-” represents the branch from −∞-\infty to ∞\infty (C−\mathbb{C}_-), see Figure 1. For bosons, we have [19, 21, 22]

⟨T[ϕ(x)ϕ†(y)]⟩={⟨ϕ(x)ϕ†(y)⟩,tx>ty,⟨ϕ†(y)ϕ(x)⟩,tx<ty,(7)\langle T[\phi(x)\phi^\dagger(y)]\rangle= \begin{cases} \langle\phi(x)\phi^\dagger(y)\rangle, & t_x>t_y,\\ \langle\phi^\dagger(y)\phi(x)\rangle, & t_x<t_y, \end{cases} \tag*{(7)}

where tx,ty∈C−t_x,t_y\in\mathbb{C}_-, and

⟨T~[ϕ(x)ϕ†(y)]⟩={⟨ϕ†(y)ϕ(x)⟩,tx>ty,⟨ϕ(x)ϕ†(y)⟩,tx<ty.(8)\langle\widetilde{T}[\phi(x)\phi^\dagger(y)]\rangle= \begin{cases} \langle\phi^\dagger(y)\phi(x)\rangle, & t_x>t_y,\\ \langle\phi(x)\phi^\dagger(y)\rangle, & t_x<t_y. \end{cases} \tag*{(8)}

where tx,ty∈C+t_x,t_y \in\mathbb{C}_+. For fermions, we have

⟨T[ψ(x)ψˉ(y)]⟩={⟨ψ(x)ψˉ(y)⟩,tx>ty−⟨ψˉ(y)ψ(x)⟩,tx<ty,(9)\langle T[\psi(x)\bar{\psi}(y)]\rangle= \begin{cases} \langle\psi(x)\bar{\psi}(y)\rangle, & t_x>t_y \\ -\langle\bar{\psi}(y)\psi(x)\rangle, & t_x<t_y, \end{cases} \tag*{(9)}
⟨T~[ψ(x)ψˉ(y)]⟩={−⟨ψˉ(y)ψ(x)⟩,tx>ty⟨ψ(x)ψˉ(y)⟩,tx<ty.(10)\langle\widetilde{T}[\psi(x)\bar{\psi}(y)]\rangle= \begin{cases} -\langle\bar{\psi}(y)\psi(x)\rangle, & t_x>t_y \\ \langle\psi(x)\bar{\psi}(y)\rangle, & t_x<t_y. \end{cases} \tag*{(10)}

By substituting the expansions of ϕ\phi and ϕ†\phi^\dagger into the definition of Green’s functions, we can obtain the non-interacting boson Green’s functions GB+−G_B^{+-} and GB−+G_B^{-+} [22]:

GB+−(x,y)=−2πi∫d3p⃗dE(2π)4{nB(p⃗)eip⃗⋅Δx⃗−iEΔtθ(E)δ(E2−p⃗ 2−m2)+[nB(p⃗)+1]e−ip⃗⋅Δx⃗+iEΔtθ(E)δ(E2−p⃗ 2−m2)},G_B^{+-}(x,y)=-2\pi i \int\frac{d^3\vec{p}dE}{(2\pi)^4} \left\{ n_B(\vec{p})e^{i\vec{p}\cdot\Delta\vec{x}-iE\Delta t}\theta(E)\delta(E^2-\vec{p}^{\,2}-m^2) +[n_B(\vec{p})+1]e^{-i\vec{p}\cdot\Delta\vec{x}+iE\Delta t}\theta(E)\delta(E^2-\vec{p}^{\,2}-m^2) \right\},
GB−+(x,y)=−2πi∫d3p⃗dE(2π)4{nB(p⃗)eiEΔt−ip⃗⋅Δx⃗θ(E)δ(E2−p⃗ 2−m2)+[nB(p⃗)+1]e−iEΔt+ip⃗⋅Δx⃗θ(E)δ(E2−p⃗ 2−m2)}.(11)G_B^{-+}(x,y)=-2\pi i \int\frac{d^3\vec{p}dE}{(2\pi)^4} \left\{ n_B(\vec{p})e^{iE\Delta t-i\vec{p}\cdot\Delta\vec{x}}\theta(E)\delta(E^2-\vec{p}^{\,2}-m^2) +[n_B(\vec{p})+1]e^{-iE\Delta t+i\vec{p}\cdot\Delta\vec{x}}\theta(E)\delta(E^2-\vec{p}^{\,2}-m^2) \right\}. \tag*{(11)}

In Eq. (11), nB(p⃗)=⟨Ω∣a^p⃗†a^p⃗∣Ω⟩n_B(\vec{p})=\langle\Omega|\hat{a}^{\dagger}_{\vec{p}}\hat{a}_{\vec{p}}|\Omega\rangle is the momentum distribution of the bosons in the system ∣Ω⟩|\Omega\rangle. The term nB(p⃗)+1n_B(\vec{p})+1 follows from the commutation relation in Eq. (4). Here x=(tx,x⃗)x=(t_x,\vec{x}) and y=(ty,y⃗)y=(t_y,\vec{y}) denote the positions in 4-dimensional space-time. Δx⃗=x⃗−y⃗\Delta\vec{x}=\vec{x}-\vec{y} and Δt=tx−ty\Delta t=t_x-t_y are their differences.

The non-interacting fermion Keldysh Green’s functions can be similarly derived by expanding the quark field. Since in [16] only a single fermion is considered in the SSE, in this study we accordingly neglect the contribution from the antifermions. According to the definitions of the fermion Keldysh Green’s function, we then have

GF,ξξ′+−(x,y)=2πi(iγ⋅∂x+m)ξξ′GF,0+−(x,y)=2πi(iγ⋅∂x+m)ξξ′∫d3p⃗dE(2π)4nF(p⃗)eip⃗⋅Δx⃗−iEΔtθ(E)δ(E2−p⃗ 2−m2),G_{F,\xi\xi'}^{+-}(x,y)=2\pi i(i\gamma\cdot\partial_x+m)_{\xi\xi'}G_{F,0}^{+-}(x,y) =2\pi i(i\gamma\cdot\partial_x+m)_{\xi\xi'} \int\frac{d^3\vec{p}dE}{(2\pi)^4} n_F(\vec{p})e^{i\vec{p}\cdot\Delta\vec{x}-iE\Delta t}\theta(E)\delta(E^2-\vec{p}^{\,2}-m^2),
GF,ξξ′−+(x,y)=−2πi(iγ⋅∂x+m)ξξ′GF,0−+(x,y)=−2πi(iγ⋅∂x+m)ξξ′∫d3p⃗dE(2π)4{[1−nF(p⃗)]eip⃗⋅Δx⃗−iEΔt×θ(E)δ(E2−p⃗ 2−m2)}.(12)G_{F,\xi\xi'}^{-+}(x,y)=-2\pi i(i\gamma\cdot\partial_x+m)_{\xi\xi'}G_{F,0}^{-+}(x,y) =-2\pi i(i\gamma\cdot\partial_x+m)_{\xi\xi'} \int\frac{d^3\vec{p}dE}{(2\pi)^4} \left\{[1-n_F(\vec{p})]e^{i\vec{p}\cdot\Delta\vec{x}-iE\Delta t} \times\theta(E)\delta(E^2-\vec{p}^{\,2}-m^2)\right\}. \tag*{(12)}

Here GF,0+−G_{F,0}^{+-} and GF,0−+G_{F,0}^{-+} denote the fermion Keldysh Green’s function without the spinor part. nF(p⃗)=∑s=1,2⟨Ω∣b^p⃗†b^p⃗∣Ω⟩n_F(\vec{p})=\sum_{s=1,2}\langle\Omega|\hat{b}^{\dagger}_{\vec{p}}\hat{b}_{\vec{p}}|\Omega\rangle is the momentum distribution of the fermions in the system ∣Ω⟩|\Omega\rangle. ξ\xi and ξ′\xi' are the spinor indices. γ\gamma in Eq. (12) represents the Dirac matrices γμ\gamma^\mu, and μ=0,1,2,3\mu=0,1,2,3. ∂x=∂x,μ\partial_x=\partial_{x,\mu} denotes the derivative with respect to the space-time coordinate xx, and mm is the mass of the fermion.

The interacting Keldysh Green’s functions can be expanded in perturbation theory in terms of the Feynman diagrams. In Figure 2, we provide the Feynman rules, which are adopted from [19]: the solid line represents the fermions, and the wavy line represents the bosons. α,β=±\alpha,\beta=\pm at the end and start points of the lines are the labels for the doublet fermion or boson field at finite temperature. At the vertices, the signs of the labels can be either ++ or −-. gg is the coupling constant between the fermion and boson field, see Sec. 3 for the details.

Figure 2. Feynman rules for the Keldysh Green’s functions [19].

For the fermion field, the perturbative expansion of the interacting two-point Keldysh Green’s functions is illustrated in Figure 3 [19, 23]. We ignore the term represented in Figure 4 because in this work we assume the thermal medium to be symmetric with respect to particles and antiparticles for simplicity.

Figure 3. The perturbative expansion of the interacting Keldysh Green’s function GF+−G_F^{+-} in a medium. The solid lines represent the noninteracting Keldysh Green’s functions for the fermion. The wavy lines represent the noninteracting Keldysh Green’s functions for the boson. α,β,γ,δ=+,−\alpha,\beta,\gamma,\delta=+,- are the labels of different kinds of Keldysh Green’s functions in Eq. (2).

Figure 4. The tadpole contribution to the Keldysh Green’s function GF+−G_F^{+-}. This contribution vanishes in a symmetric medium with respect to fermions and antifermions. The solid line represents the Green’s function for the fermion, and the wavy line represents the Green’s function for the boson.

Now we are ready to derive the Boltzmann equation for the fermions using the Keldysh Green’s functions. The following derivation follows from [19]. We repeat it here for the convenience of the readers.

To obtain the Boltzmann equation, we first apply the differential operator iγ⋅∂x−mi\gamma\cdot\partial_x-m from the Dirac equation to the fermion Keldysh Green’s function GF+−G_F^{+-}, and obtain its equation of motion in 4-dimensional space-time. To simplify the derivation, we introduce the differential operator Gx,ξξ′−1=(∂x2+m2)Iξξ′G_{x,\xi\xi'}^{-1}=(\partial_x^2+m^2)I_{\xi\xi'} (II is a 4×44\times4 identity matrix in the spinor space), which is the product of iγ⋅∂x−mi\gamma\cdot\partial_x-m and the operator iγ⋅∂x+mi\gamma\cdot\partial_x+m. Applying the operator Gx,ξξ′−1G_{x,\xi\xi'}^{-1} to the Keldysh Green’s function GF+−G_F^{+-} and making use of the equation of motion for the fermion operator ψ(x)\psi(x), we obtain the following equation of motion [19]

Gx,ξξ′−1GF,0+−(x,y)=∑α=+,−∑ξ′′∫d4x′ΣF,ξξ′′−α(x,x′)GF,ξ′′ξ′α+(x′,y).(13)G_{x,\xi\xi'}^{-1}G_{F,0}^{+-}(x,y) = \sum_{\alpha=+,-}\sum_{\xi''}\int d^4x'\Sigma_{F,\xi\xi''}^{-\alpha}(x,x')G_{F,\xi''\xi'}^{\alpha+}(x',y). \tag*{(13)}

Here x,y,x′x,y,x' are the space-time coordinates. ΣF,ξξ′−α\Sigma_{F,\xi\xi'}^{-\alpha} is the fermion self-energy multiplied by the imaginary unit −i-i, which is necessary for the compatibility between the definition of the self-energy and its Feynman diagram representation. The fermion’s self-energy Σαβ\Sigma^{\alpha\beta} consists of the sum of one-particle-irreducible diagrams, see Figure 5 [23]. For convenience, we omit the spinor indices ξ\xi and ξ′\xi' in the following part of this paper.

Figure 5. Feynman diagrams of the one-particle-irreducible diagrams of the self-energy Σαβ\Sigma^{\alpha\beta} for the fermion.

Now, we perform the Wigner transformation on both sides of the following equation, which follows from Eq. (13) [19],

(Gy−1−Gx−1)GF,0+−(x,y)=−∑α=+,−∫d4x′[Σ−α(x,x′)GFα+(x′,y)−GF−α(x,x′)Σα+(x′,y)].(14)\left(G_y^{-1}-G_x^{-1}\right)G_{F,0}^{+-}(x,y)=-\sum_{\alpha=+,-}\int d^4x'\left[\Sigma^{-\alpha}(x,x')G_F^{\alpha+}(x',y)-G_F^{-\alpha}(x,x')\Sigma^{\alpha+}(x',y)\right]. \tag*{(14)}

In this paper, we adopt the following definition for the Wigner transformation in 4-dimensional space-time for the fermion Green’s functions [19, 24]

W[GF+−(x,y)]=W[GF+−(X+12x′,X−12x′)]=∫d4x′(2π)4eiP⋅x′GF+−(X+12x′,X−12x′).(15)\mathcal{W}\left[G_F^{+-}(x,y)\right]=\mathcal{W}\left[G_F^{+-}\left(X+\frac{1}{2}x',X-\frac{1}{2}x'\right)\right] =\int\frac{d^4x'}{(2\pi)^4}e^{iP\cdot x'}G_F^{+-}\left(X+\frac{1}{2}x',X-\frac{1}{2}x'\right). \tag*{(15)}

Here W\mathcal{W} denotes the Wigner transformation and X=x+y2X=\frac{x+y}{2}, x′=x−yx'=x-y. P=(P0,P⃗)P=(P^0,\vec{P}) is the 4-momentum of the fermion. Next we will show that after the Wigner transformation, the left-hand side of Eq. (14) is identified as the drift term of the Boltzmann equation, and the right-hand side becomes the scattering term.

If we assume that the fermion Green’s function GF+−(X+x′2,X−x′2)G_F^{+-}\left(X+\frac{x'}{2},X-\frac{x'}{2}\right) weakly depends on the position XX [19], we have

∫d4x′(2π)4eiP⋅x′GF+−(X+12x′,X−12x′)≈∫d4x′(2π)4eiP⋅x′GF+−(12x′,−12x′)=∫d4x′(2π)4∫d3p⃗ dE(2π)4(iγ⋅∂x′+m)nF(p⃗)eiP⋅x′eip⃗⋅x⃗′−iEt′×(2πi)θ(E)δ(E2−p⃗ 2−m2)=1(2π)4∫d3p⃗ dE(2π)4(iγ⋅p+m)nF(p⃗)×(2π)5iθ(E)δ(E2−p⃗ 2−m2)δ3(P⃗−p⃗)δ(P0−E)=i(iγ⋅P+m)(2π)3nF(P⃗)θ(P0)δ[(P0)2−P⃗ 2−m2].(16)\begin{aligned} \int\frac{d^4x'}{(2\pi)^4}e^{iP\cdot x'}G_F^{+-}\left(X+\frac{1}{2}x',X-\frac{1}{2}x'\right) \approx\int\frac{d^4x'}{(2\pi)^4}e^{iP\cdot x'}G_F^{+-}\left(\frac{1}{2}x',-\frac{1}{2}x'\right) \\ &=\int\frac{d^4x'}{(2\pi)^4}\int\frac{d^3\vec{p}\,dE}{(2\pi)^4}(i\gamma\cdot\partial_{x'}+m)n_F(\vec{p})e^{iP\cdot x'}e^{i\vec{p}\cdot\vec{x}'-iEt'}\times(2\pi i)\theta(E)\delta(E^2-\vec{p}^{\,2}-m^2) \\ &=\frac{1}{(2\pi)^4}\int\frac{d^3\vec{p}\,dE}{(2\pi)^4}(i\gamma\cdot p+m)n_F(\vec{p})\times(2\pi)^5i\theta(E)\delta(E^2-\vec{p}^{\,2}-m^2)\delta^3(\vec{P}-\vec{p})\delta(P^0-E) \\ &=\frac{i(i\gamma\cdot P+m)}{(2\pi)^3}n_F(\vec{P})\theta(P^0)\delta\left[(P^0)^2-\vec{P}^{\,2}-m^2\right]. \tag*{(16)} \end{aligned}

Here p=(E,p⃗)p=(E,\vec{p}) is the 4-momentum in the fermion Green’s function. In the second equality, we have used the Fourier expansion of the fermion Green’s function in Eq. (12). From this derivation, we find that the Wigner transformation of GF+−(x,y)G_F^{+-}(x,y) is approximately equal to the 4-dimensional Fourier transformation of GF+−(12x′,−12x′)G_F^{+-}\left(\frac{1}{2}x',-\frac{1}{2}x'\right) multiplied by (2π)−4(2\pi)^{-4}, where the weak dependence on XX is carried by nF(P⃗)n_F(\vec{P}). For the fermion systems weakly depending on the position XX, its phase space distribution f(X,P)f(X,P) approximately satisfies

f(X,P)≈nF(P⃗)2EP⃗θ(P0)δ[(P0)2−P⃗ 2−m2].(17)f(X,P) \approx n_F(\vec{P})2E_{\vec{P}}\theta(P^0)\delta\left[(P^0)^2-\vec{P}^{\,2}-m^2\right]. \tag*{(17)}

Here nF(P⃗)n_F(\vec{P}) implicitly carries a weak dependence on the position XX of the fermion. Comparing Eq. (16) and Eq. (17), we identify the relation between the fermion’s phase space distribution and the Wigner transformation of GF−+G_F^{-+} as

W[GF−+(x,y)]=i(iγ⋅P+m)(2π)32EP⃗f(X,P).(18)\mathcal{W}\left[G_F^{-+}(x,y)\right]=\frac{i(\mathrm{i}\gamma\cdot P+m)}{(2\pi)^3 2E_{\vec{P}}}f(X,P). \tag*{(18)}

The Wigner transformation of the Green’s function without the spinor part GF,0−+G_{F,0}^{-+} is accordingly

W[GF,0−+(x,y)]≈i(2π)3nF(P⃗)θ(P0)δ[(P0)2−P⃗ 2−m2]≈i(2π)32EP⃗f(X,P).(19)\begin{aligned} \mathcal{W}\left[G_{F,0}^{-+}(x,y)\right] \approx\frac{i}{(2\pi)^3}n_F(\vec{P})\theta(P^0)\delta\left[(P^0)^2-\vec{P}^{\,2}-m^2\right] \\ &\approx\frac{i}{(2\pi)^3 2E_{\vec{P}}}f(X,P). \tag*{(19)} \end{aligned}

Similarly, the Wigner transformation of GF,0+−G_{F,0}^{+-} is

W[GF,0+−(x,y)]≈i(2π)32EP⃗[1−f(X,P)].(20)\mathcal{W}\left[G_{F,0}^{+-}(x,y)\right]\approx\frac{i}{(2\pi)^3 2E_{\vec{P}}}\left[1-f(X,P)\right]. \tag*{(20)}

Now, we are ready to perform the Wigner transformation on the left-hand side of Eq. (14). In position space, any two-variable function ρ(xi,yi)\rho(x^i,y^i) satisfies the following relation [24, 25, 26] for the Wigner transformation (where i=1,2,3i=1,2,3 is the label for the spatial directions):

W[i(∂x,i2−∂y,i2)ρ(xi,yi)]=−2Pi∂∂XiS(Xi,Pi),\mathcal{W}\left[\mathrm{i}\left(\partial_{x,i}^2-\partial_{y,i}^2\right)\rho(x^i,y^i)\right]=-2P^i\frac{\partial}{\partial X^i}S(X^i,P^i),
S(Xi,Pi)=W[ρ(xi,yi)].(21)S(X^i,P^i)=\mathcal{W}\left[\rho(x^i,y^i)\right]. \tag*{(21)}

Here W\mathcal{W} denotes the Wigner transformation and S(Xi,Pi)S(X^i,P^i) is the Wigner transformation of ρ(xi,yi)\rho(x^i,y^i) and

Xi=xi+yi2.X^i=\frac{x^i+y^i}{2}.

xix^i and yiy^i are the iith spatial components of the 4-position xx and yy, respectively. PiP^i is the iith spatial component of the 4-momentum. For the time component of xx and yy, the Wigner transformation is

W[i(∂x,02−∂y,02)ρ(x0,y0)]=∫d(tx−ty)2πeiP0(tx−ty)[i(∂x,02−∂y,02)ρ(x0,y0)]=2P0∂∂X0S(X0,P0).(22)\begin{aligned} \mathcal{W}\left[\mathrm{i}\left(\partial_{x,0}^2-\partial_{y,0}^2\right)\rho(x^0,y^0)\right] =\int\frac{d(t_x-t_y)}{2\pi}e^{iP^0(t_x-t_y)} \left[\mathrm{i}\left(\partial_{x,0}^2-\partial_{y,0}^2\right)\rho(x^0,y^0)\right] \\ &=2P^0\frac{\partial}{\partial X^0}S(X^0,P^0). \tag*{(22)} \end{aligned}

Here we have used the relation eiP⋅x′=eiP0t′−iP⃗⋅x⃗′e^{iP\cdot x'}=e^{iP^0t'-i\vec{P}\cdot\vec{x}'} for the 4-momentum PP and 4-position x′x'.

By using Eq. (21) and Eq. (22), the Wigner transformation on the left-hand side of Eq. (14) is

W[(Gy−1−Gx−1)GF,0−+(x,y)]=W[I((∂xμ∂x,μ+m2)−(∂yμ∂y,μ+m2))GF,0−+(x,y)]=2(P0∂∂X0+∑j=13Pj∂∂Xj)W[−iGF,0−+(x,y)]I,(23)\begin{aligned} \mathcal{W}\left[\left(G_y^{-1}-G_x^{-1}\right)G_{F,0}^{-+}(x,y)\right] =\mathcal{W}\left[\mathbf{I}\left((\partial_x^\mu\partial_{x,\mu}+m^2)-(\partial_y^\mu\partial_{y,\mu}+m^2)\right)G_{F,0}^{-+}(x,y)\right] \\ &=2\left(P^0\frac{\partial}{\partial X^0}+\sum_{j=1}^{3}P^j\frac{\partial}{\partial X^j}\right)\mathcal{W}\left[-\mathrm{i}G_{F,0}^{-+}(x,y)\right]\mathbf{I}, \tag*{(23)} \end{aligned}

where I\mathbf{I} is the 4×44\times4 identity matrix in the spinor space. After integrating P0P^0, the result is

W[(Gy−1−Gx−1)GF,0−+(x,y)]=I(2π)3EP⃗Pν∂f(X,P)∂Xν.(24)\mathcal{W}\left[\left(G_y^{-1}-G_x^{-1}\right)G_{F,0}^{-+}(x,y)\right] =\frac{\mathbf{I}}{(2\pi)^3E_{\vec{P}}}P^\nu\frac{\partial f(X,P)}{\partial X^\nu}. \tag*{(24)}

Here we have used the result of the Wigner transformation of GF,0−+G_{F,0}^{-+} in Eq. (19).

Next, we perform a Wigner transformation on the right-hand side of Eq. (14). We first simplify the right-hand side using the following relationship between the Keldysh Green’s functions and fermion self-energies [19],

GF−−(x,y)+GF++(x,y)−GF−+(x,y)−GF+−(x,y)=0,G_F^{--}(x,y)+G_F^{++}(x,y)-G_F^{-+}(x,y)-G_F^{+-}(x,y)=0,
Σ−−(x,y)+Σ++(x,y)+Σ−+(x,y)+Σ+−(x,y)=0.(25)\Sigma^{--}(x,y)+\Sigma^{++}(x,y)+\Sigma^{-+}(x,y)+\Sigma^{+-}(x,y)=0. \tag*{(25)}

As shown above, the Wigner transformation of a function is approximately equal to its Fourier transformation up to the integration over 4-position and a constant factor of (2π)−4(2\pi)^{-4}. Therefore, the Wigner transform of the convolution of the two functions is approximately equal to (2π)4(2\pi)^4 times the product of each Wigner transformation.

Finally, after substituting Eq. (25) into the right side of Eq. (14), we obtain the Wigner transformation result of Eq. (14) as [19]

IPμ∂f(X,P)∂Xμ=(2π)3EP⃗(2π)4×[−Σw−+(X,P)GF,w+−(X,P)+Σw+−(X,P)GF,w−+(X,P)].(26)\begin{aligned} \mathbf{I}P^\mu\frac{\partial f(X,P)}{\partial X^\mu} =(2\pi)^3E_{\vec{P}}(2\pi)^4 \\ &\times\left[-\Sigma_w^{-+}(X,P)G_{F,w}^{+-}(X,P)+\Sigma_w^{+-}(X,P)G_{F,w}^{-+}(X,P)\right]. \tag*{(26)} \end{aligned}

Here GF,w−+(X,P)G_{F,w}^{-+}(X,P), GF,w+−(X,P)G_{F,w}^{+-}(X,P), Σw+−(X,P)\Sigma_w^{+-}(X,P), Σw−+(X,P)\Sigma_w^{-+}(X,P) are the Wigner transformation of GF−+(x,y)G_F^{-+}(x,y), GF+−(x,y)G_F^{+-}(x,y), Σ+−(x,y)\Sigma^{+-}(x,y), Σ−+(x,y)\Sigma^{-+}(x,y) respectively.

By comparing with the standard form of the relativistic Boltzmann equation [27]

Pμ∂f(X,P)∂Xμ=12(2π)3∫d3P⃗1EP⃗1∫d3P⃗′EP⃗′∫d3P⃗1′EP⃗1′×W(P′,P1′,P,P1)(f′f1′−ff1),(27)\begin{aligned} P^\mu\frac{\partial f(X,P)}{\partial X^\mu} =\frac{1}{2(2\pi)^3}\int\frac{d^3\vec{P}_1}{E_{\vec{P}_1}}\int\frac{d^3\vec{P}'}{E_{\vec{P}'}}\int\frac{d^3\vec{P}'_1}{E_{\vec{P}'_1}} \\ &\times W(P',P'_1,P,P_1)\left(f'f'_1-ff_1\right), \tag*{(27)} \end{aligned}

we identify the right-hand side of Eq. (26) as the scattering term in the absence of the 4-force. In Eq. (27), ff, f1f_1, f′f', f1′f'_1 are the phase space distribution functions of the incoming and outgoing particles in the scattering process at the position XX with momentum P⃗\vec{P}, P⃗1\vec{P}_1, P⃗′\vec{P}', P⃗1′\vec{P}'_1, respectively. EP⃗′E_{\vec{P}'}, EP⃗1E_{\vec{P}_1}, EP⃗1′E_{\vec{P}'_1} are the corresponding energy of the particles participating in the scattering. The function WW is the scattering probability, which is encoded in the self-energies Σw+−\Sigma^{+-}_w and Σw−+\Sigma^{-+}_w in Eq. (26) [19, 20, 23]. The exact relation between the right-hand side of Eq. (26) and (27) will be investigated in the next section.

Stochastic Schrödinger equation

In reference [16], the authors introduced a stochastic Schrödinger equation to describe the time evolution of a heavy quark in a thermal medium, which takes the following form:

i∂∂tφ(x⃗,t)=(H0+HI)φ(x⃗,t),i\frac{\partial}{\partial t}\varphi(\vec{x},t)=(H_0+H_I)\varphi(\vec{x},t),
H0=−∇2+mq2,H_0=\sqrt{-\nabla^2+m_q^2},
HI=∫d3x⃗ gψˉγμψAμ.(28)H_I=\int\mathrm{d}^3\vec{x}\,g\bar{\psi}\gamma^\mu\psi A^\mu. \tag*{(28)}

Here φ(x⃗,t)\varphi(\vec{x},t) is the wave function of the heavy quark, H0=−∇2+mq2H_0=\sqrt{-\nabla^2+m_q^2} is the heavy quark’s kinetic energy, mqm_q is the mass of the heavy quark, and HIH_I describes the interaction between the heavy quark and the thermal medium with the color degrees of freedom being ignored for simplicity. To account for the interaction between the heavy quark and the thermal medium, the authors modeled the medium as an external gluon field Aμ(x⃗,t)A^\mu(\vec{x},t). In HIH_I, gg is the coupling constant between the quark field and the gluon field. Since the mass of the heavy quark is much larger than the typical temperature of the medium, the authors employed a non-relativistic approximation and considered only the coupling between the heavy quark and the zeroth component of the background field A0(x⃗,t)A^0(\vec{x},t) [16, 17]. For conciseness, we denote A0A^0 as AA in the following part of this paper. By expanding this field on the plane-wave basis, the authors obtained

A(x⃗,t)=∫d3p⃗(2π)3A(p⃗,t)eip⃗⋅x⃗=∫d3p⃗(2π)32dgEp⃗a(p⃗,t)eip⃗⋅x⃗,A(\vec{x},t)=\int\frac{\mathrm{d}^3\vec{p}}{(2\pi)^3}A(\vec{p},t)e^{i\vec{p}\cdot\vec{x}}=\int\frac{\mathrm{d}^3\vec{p}}{(2\pi)^3}\sqrt{\frac{2d_g}{E_{\vec{p}}}}a(\vec{p},t)e^{i\vec{p}\cdot\vec{x}},
a(p⃗,t)=n(p⃗)eiθ(p⃗,t),a(\vec{p},t)=\sqrt{n(\vec{p})}e^{i\theta(\vec{p},t)},
n(p⃗)=exp⁡(−Ep⃗T)=exp⁡(−p⃗ 2+mg2T).(29)n(\vec{p})=\exp\left(-\frac{E_{\vec{p}}}{T}\right)=\exp\left(-\frac{\sqrt{\vec{p}^{\,2}+m_g^2}}{T}\right). \tag*{(29)}

Here A(p⃗,t)A(\vec{p},t) denotes the AA field in momentum space. The gluon momentum distribution n(p⃗)n(\vec{p}) is assumed to (approximately) follow the Boltzmann distribution. Ep⃗E_{\vec{p}} represents the on-shell gluon kinetic energy. The gluon thermal mass is denoted as mg=g01+Nf6Tm_g=g_0\sqrt{1+\frac{N_f}{6}}T. In this work we take g0g_0 as a strong coupling constant independent of the coupling constant describing the interaction between the heavy quark and the medium, that is, gg, in Eq. (28). The number of flavor for light quarks is Nf=3N_f=3, and TT denotes the temperature of the medium. In Eq. (29), p⃗\vec{p} is the momentum of the heavy quark, dg=16d_g=16 is the color-spin degeneracy of the gluon. The authors [16] introduced a time-dependent random phase factor eiθ(p⃗,t)e^{i\theta(\vec{p},t)} for each momentum mode of the gluon field. These random phase factors parameterize the thermal fluctuation of the gluon field on the amplitude level.

The matrix element of the interaction Hamiltonian in momentum basis takes the following form:

⟨p⃗,s∣HI∣p⃗′,s′⟩=∫d3x⃗ gA(x⃗)2Eq,p⃗2Eq,p⃗′ uˉs(p)γ0us′(p′)e−i(p⃗′−p⃗)⋅x⃗eiθ(p⃗′−p⃗,t)≈g2dgEga(p⃗′−p⃗)δs′s,\begin{aligned} \langle\vec{p},s|H_I|\vec{p}',s'\rangle=\int\frac{\mathrm{d}^3\vec{x}\,gA(\vec{x})}{\sqrt{2E_{q,\vec{p}}\sqrt{2E_{q,\vec{p}'}}}\,}\bar{u}^{s}(p)\gamma^0u^{s'}(p')e^{-i(\vec{p}'-\vec{p})\cdot\vec{x}}e^{i\theta(\vec{p}'-\vec{p},t)} \\ &\approx g\sqrt{\frac{2d_g}{E_g}}a(\vec{p}'-\vec{p})\delta^{s's}, \end{aligned}
Eg=(p⃗′−p⃗)2+mg2.(30)E_g=\sqrt{(\vec{p}'-\vec{p})^2+m_g^2}. \tag*{(30)}

Here Eq,p⃗=p⃗ 2+mq2E_{q,\vec{p}}=\sqrt{\vec{p}^{\,2}+m_q^2} is the kinetic energy of the heavy quark. p=(Eq,p⃗,p⃗)p=(E_{q,\vec{p}},\vec{p}) is its 4-momentum. In this derivation, the following non-relativistic approximations are adopted

uˉs(p)γ0us′(p′)≈2Eq,p⃗δs′s,\bar{u}^{s}(p)\gamma^0u^{s'}(p')\approx2E_{q,\vec{p}}\delta^{s's},
p⃗≈p⃗′,\vec{p}\approx\vec{p}',
Eq,p⃗≈mq.(31)E_{q,\vec{p}}\approx m_q. \tag*{(31)}

Since the authors in [16] ignored the spin degrees of freedom, the δs′s\delta^{s's} factor is omitted in the stochastic Schrödinger equation. After substituting Eq. (30) into Eq. (28), we obtain the relativistic Schrödinger equation in momentum space describing the interaction between the heavy quark and the thermal gluon field,

i∂∂tφ(p⃗,t)=p⃗ 2+mq2φ(p⃗,t)+g∫d3p⃗′2dgEga(p⃗′−p⃗)φ(p⃗′,t).(32)i\frac{\partial}{\partial t}\varphi(\vec{p},t)=\sqrt{\vec{p}^{\,2}+m_q^2}\varphi(\vec{p},t)+g\int\mathrm{d}^3\vec{p}'\sqrt{\frac{2d_g}{E_g}}a(\vec{p}'-\vec{p})\varphi(\vec{p}',t). \tag*{(32)}

Here φ(p⃗,t)\varphi(\vec{p},t) is the wave function of the heavy quark in momentum space, and p⃗ 2+mq2\sqrt{\vec{p}^{\,2}+m_q^2} is the heavy quark’s kinetic energy. Since the main goal of this paper is to illustrate the connection between the SSE and Boltzmann equation, for simplicity, we neglect the dissipation term in the interaction term of Eq. (32), and as a result, this equation will lead to a uniform distribution in momentum space as the equilibrium limit for the heavy quark [16]. For the SSE with the dissipation term, we refer to [16], where the Boltzmann distribution is reached as the equilibrium distribution for the heavy quark.

Alternatively, Eq. (32) can be rewritten in the form of the Lippmann-Schwinger equation [28]:

∣φ⟩=∣φ0⟩+HIG~F0∣φ⟩.(33)|\varphi\rangle= |\varphi_0\rangle+ H_I \widetilde{G}_F^0 |\varphi\rangle. \tag*{(33)}

Here G~F0=1E−H0+iϵ\widetilde{G}_F^0 = \frac{1}{E-H_0+i\epsilon} is the noninteracting Green’s function for the heavy quark at zero temperature. ∣φ0⟩|\varphi_0\rangle denotes the wave function for the free heavy quark. H0H_0 (HIH_I) is the kinetic energy (interaction) operator of the SSE, see Eq. (28). EE is the energy of the heavy quark. The corresponding interacting Green’s function satisfies the following relation,

G~F=G~F0+G~F0HIG~F0+G~F0HIG~F0HIG~F0+…,(34)\widetilde{G}_F = \widetilde{G}_F^0 + \widetilde{G}_F^0 H_I \widetilde{G}_F^0 + \widetilde{G}_F^0 H_I \widetilde{G}_F^0 H_I \widetilde{G}_F^0 + \ldots, \tag*{(34)}

where G~F\widetilde{G}_F denotes the interacting Green’s function for the heavy quark at zero-temperature.

Here, we assume the ensemble average of the zero-temperature Green’s function corresponding to the SSE is equal to the Keldysh Green’s function, that is,

⟨G~SSE⟩=GKeldysh.(35)\langle\widetilde{G}_{\mathrm{SSE}}\rangle= G_{\mathrm{Keldysh}}. \tag*{(35)}

Here ⟨O^⟩\langle\hat{O}\rangle denotes the ensemble average of the operator O^\hat{O} [16], which is taken for all observables evaluated from the SSE. When the coupling constant is small enough, the contributions from higher-order terms in Eq. (34) are expected to be small, so ⟨G~SSE⟩\langle\widetilde{G}_{\mathrm{SSE}}\rangle can be expanded perturbatively, as illustrated in Figure 6 [19]. The Feynman rules for the external field are illustrated in Figure 7. The dashed lines represent the AA field and are labeled with ±\pm at the interaction vertices.

Figure 6. The perturbative expansion of the heavy quark Green’s function GFρσG_F^{\rho\sigma} with an external classical gluon field AA coupled to the heavy quark. The dashed lines represent the external AA field in Eq. (29). The solid lines represent the heavy quark Green’s functions. α,β,γ,ρ,σ=+,−\alpha,\beta,\gamma,\rho,\sigma=+,- are the labels of different kinds of the Keldysh Green’s functions in Eq. (2).

Figure 7. Feynman rules for the external classical AA field, which is denoted by the cross [19].

In order to derive the Boltzmann equation from the stochastic Schrödinger equation, let us first work out the corresponding gluon distribution from Eq. (32).

The random phase factor in the gluon field satisfies eiθ(−p⃗,t)=e−iθ(p⃗,t)e^{i\theta(-\vec{p},t)}=e^{-i\theta(\vec{p},t)} [16], which leads to a real A(x⃗,t)A(\vec{x},t) field,

A(p⃗,t)=2dgEp⃗a(p⃗,t)=2dgEp⃗n(p⃗)eiθ(p⃗,t),A(−p⃗,t)=2dgE−p⃗a(−p⃗,t)=2dgE−p⃗n(−p⃗)eiθ(−p⃗,t)=2dgEp⃗n(p⃗)e−iθ(p⃗,t)=A∗(p⃗,t).(36)\begin{aligned} A(\vec{p},t) &= \sqrt{\frac{2d_g}{E_{\vec{p}}}}a(\vec{p},t) = \sqrt{\frac{2d_g}{E_{\vec{p}}}}\sqrt{n(\vec{p})}e^{i\theta(\vec{p},t)}, \\ A(-\vec{p},t) &= \sqrt{\frac{2d_g}{E_{-\vec{p}}}}a(-\vec{p},t) = \sqrt{\frac{2d_g}{E_{-\vec{p}}}}\sqrt{n(-\vec{p})}e^{i\theta(-\vec{p},t)} \\ &= \sqrt{\frac{2d_g}{E_{\vec{p}}}}\sqrt{n(\vec{p})}e^{-i\theta(\vec{p},t)} \\ &= A^*(\vec{p},t). \tag*{(36)} \end{aligned}

Here A∗(p⃗,t)A^*(\vec{p},t) is the complex conjugate of A(p⃗,t)A(\vec{p},t). Therefore, A(x⃗,t)A(\vec{x},t) is real.

In [16], the random phases are assumed to be uncorrelated for different momentum p⃗\vec{p}, and correlated within the time interval of tcorrt_{\mathrm{corr}}, that is, the random phases are updated every period of the correlation time tcorrt_{\mathrm{corr}}. tcorrt_{\mathrm{corr}} is assumed to be independent of the momentum p⃗\vec{p}. The exact time when the phases get updated for different momentum is also randomized such that the random phases for different momenta p⃗\vec{p} are updated asynchronously, see [16] for the details of the numerical implementation of tcorrt_{\mathrm{corr}}. Later in this section, we will show that the time correlation of the random phases is connected to the energy distribution of the gluons in the medium.

Following Eq. (35), we now consider the ensemble average on the AA field and its correlations, which will be needed for constructing the scattering term of the Boltzmann equation from the SSE. In Figure 6, the one-vertex term vanishes due to ⟨A(p⃗,t)⟩∝⟨eiθ(p⃗,t)⟩=0\langle A(\vec{p},t)\rangle\propto\langle e^{i\theta(\vec{p},t)}\rangle=0 since the phase θ(p⃗,t)\theta(\vec{p},t) is uniformly distributed over [−π,π][-\pi,\pi]. Because when p⃗1≠p⃗2\vec{p}_1\ne\vec{p}_2, θ(p⃗1,t)\theta(\vec{p}_1,t) is independent of θ(p⃗2,t′)\theta(\vec{p}_2,t'), the functions of two phases of different momenta are also independent of each other. Therefore, the AA field satisfies the following relation,

⟨A∗(p⃗1,t1)A(p⃗2,t2)⟩=2dgEp⃗12dgEp⃗2⟨a∗(p⃗1,t1)a(p⃗2,t2)⟩=2dgEp⃗12dgEp⃗2n(p⃗1)n(p⃗2)⟨e−iθ(p⃗1,t1)eiθ(p⃗2,t2)⟩=2dgEp⃗12dgEp⃗2n(p⃗1)n(p⃗2)δ(p⃗1−p⃗2)×⟨e−iθ(p⃗1,t1)eiθ(p⃗2,t2)⟩.(37)\begin{aligned} \langle A^*(\vec{p}_1,t_1)A(\vec{p}_2,t_2)\rangle &= \sqrt{\frac{2d_g}{E_{\vec{p}_1}}}\sqrt{\frac{2d_g}{E_{\vec{p}_2}}}\langle a^*(\vec{p}_1,t_1)a(\vec{p}_2,t_2)\rangle\\ &= \sqrt{\frac{2d_g}{E_{\vec{p}_1}}}\sqrt{\frac{2d_g}{E_{\vec{p}_2}}}\sqrt{n(\vec{p}_1)}\sqrt{n(\vec{p}_2)} \langle e^{-i\theta(\vec{p}_1,t_1)}e^{i\theta(\vec{p}_2,t_2)}\rangle\\ &= \sqrt{\frac{2d_g}{E_{\vec{p}_1}}}\sqrt{\frac{2d_g}{E_{\vec{p}_2}}}\sqrt{n(\vec{p}_1)n(\vec{p}_2)} \delta(\vec{p}_1-\vec{p}_2) \\ &\quad\times\langle e^{-i\theta(\vec{p}_1,t_1)}e^{i\theta(\vec{p}_2,t_2)}\rangle. \tag*{(37)} \end{aligned}

When ∣t1−t2∣≤tcorr|t_1-t_2|\le t_{\mathrm{corr}}, the correlation of the AA fields at two different times but the same momentum is nonzero. Therefore, the two-vertex terms are the lowest-order non-vanishing terms in the expansion of ⟨GSSE⟩\langle G_{\mathrm{SSE}}\rangle in Eq. (34). Here, we note that the correlation of the AA field with different momenta vanishes [16].

Because the correlation function of the random phases is determined only by the difference in time and momentum, the correlation of eiθ(p⃗,t)e^{i\theta(\vec{p},t)} is also determined only by these differences. In order to study the time correlation of the AA field, we define the following correlation function of the random phase factor eiθ(p⃗,t)e^{i\theta(\vec{p},t)} as,

g(p⃗,Δt)=1Tmax⁡∫0Tmax⁡⟨e−iθ(p⃗,t+Δt)eiθ(p⃗,t)⟩ dt,(38)g(\vec{p},\Delta t)=\frac{1}{T_{\max}}\int_0^{T_{\max}}\langle e^{-i\theta(\vec{p},t+\Delta t)}e^{i\theta(\vec{p},t)}\rangle\,dt, \tag*{(38)}

where Tmax⁡T_{\max} is the duration of the time evolution. In [16] the SSE was solved numerically through stepwise evolution using the MSD algorithm [29]. If the phase θ(p⃗,ti)\theta(\vec{p},t_i) is known at each time step tit_i, we could use the following discretized formula to obtain g(p⃗,Δt)g(\vec{p},\Delta t) numerically [30]

g(p⃗,Δt)=1n−h⟨∑i=1n−he−iθ(p⃗,ti+h)eiθ(p⃗,ti)⟩.(39)g(\vec{p},\Delta t)=\frac{1}{n-h}\left\langle\sum_{i=1}^{n-h}e^{-i\theta(\vec{p},t_{i+h})}e^{i\theta(\vec{p},t_i)}\right\rangle. \tag*{(39)}

Here nn represents the total number of steps in an event, and h=Δttstep=ti+h−titsteph=\frac{\Delta t}{t_{\mathrm{step}}}=\frac{t_{i+h}-t_i}{t_{\mathrm{step}}}, where tstep=ti+1−tit_{\mathrm{step}}=t_{i+1}-t_i denotes the step length in time.

Taking the complex conjugation of the correlation function Eq. (38), we can obtain g∗(p⃗,Δt)=g(p⃗,−Δt)g^{*}(\vec{p},\Delta t)=g(\vec{p},-\Delta t) as,

g∗(p⃗,Δt)=⟨e−iθ(p⃗,t+Δt)eiθ(p⃗,t)⟩∗=⟨e−iθ(p⃗,t)eiθ(p⃗,t+Δt)⟩=g(p⃗,−Δt).(40)\begin{aligned} g^{*}(\vec{p},\Delta t)&=\left\langle e^{-i\theta(\vec{p},t+\Delta t)}e^{i\theta(\vec{p},t)}\right\rangle^{*}\\ &=\left\langle e^{-i\theta(\vec{p},t)}e^{i\theta(\vec{p},t+\Delta t)}\right\rangle=g(\vec{p},-\Delta t). \tag*{(40)} \end{aligned}

This property will later be used to obtain the energy distribution of the gluon field.

Since the random phases θ(p⃗,t)\theta(\vec{p},t) are independently sampled for each different momentum p⃗\vec{p}, its time correlation g(p⃗,Δt)g(\vec{p},\Delta t) is independent of p⃗\vec{p}. So from now on, we will write g(p⃗,Δt)g(\vec{p},\Delta t) as g(Δt)g(\Delta t) for simplicity.

In terms of g(Δt)g(\Delta t), the correlation of the AA field can be written as

⟨A∗(p⃗1,t1)A(p⃗2,t2)⟩=2dgEp⃗1n(p⃗1)δ(p⃗1−p⃗2)g(Δt).(41)\left\langle A^{*}(\vec{p}_{1},t_{1})A(\vec{p}_{2},t_{2})\right\rangle=\frac{2d_g}{E_{\vec{p}_{1}}}n(\vec{p}_{1})\delta(\vec{p}_{1}-\vec{p}_{2})g(\Delta t). \tag*{(41)}

To obtain the corresponding Boltzmann equation from the SSE, we need to know the Wigner distribution of the gluons in the thermal medium, which can be derived from the Wigner transformation of the correlation of the

A field. The resulting WAW_A is

WA(X⃗,P⃗,t,E)=∫d3p⃗ ′(2π)4dt′⟨A∗(P⃗−p⃗ ′2,t+t′2)A(P⃗+p⃗ ′2,t−t′2)⟩eiX⃗⋅p⃗ ′eiEt′=∫d3p⃗ ′(2π)42dgEP⃗−p⃗ ′2n(P⃗−p⃗ ′2)δ3(p⃗ ′)g(t′)eiX⃗⋅p⃗ ′eiEt′=∫dt′(2π)42dgEP⃗n(P⃗)g(t′)eiEt′=2dg(2π)4EP⃗n(P⃗)G(E).(42)\begin{aligned} W_A(\vec{X},\vec{P},t,E)&=\int\frac{d^3\vec{p}^{\,\prime}}{(2\pi)^4}dt'\left\langle A^{*}\left(\vec{P}-\frac{\vec{p}^{\,\prime}}{2},t+\frac{t'}{2}\right)A\left(\vec{P}+\frac{\vec{p}^{\,\prime}}{2},t-\frac{t'}{2}\right)\right\rangle e^{i\vec{X}\cdot\vec{p}^{\,\prime}}e^{iEt'}\\ &=\int\frac{d^3\vec{p}^{\,\prime}}{(2\pi)^4}\frac{2d_g}{E_{\vec{P}-\frac{\vec{p}^{\,\prime}}{2}}}n\left(\vec{P}-\frac{\vec{p}^{\,\prime}}{2}\right)\delta^3(\vec{p}^{\,\prime})g(t')e^{i\vec{X}\cdot\vec{p}^{\,\prime}}e^{iEt'}\\ &=\int\frac{dt'}{(2\pi)^4}\frac{2d_g}{E_{\vec{P}}}n(\vec{P})g(t')e^{iEt'}\\ &=\frac{2d_g}{(2\pi)^4E_{\vec{P}}}n(\vec{P})G(E). \tag*{(42)} \end{aligned}

Here we assume that the thermal field follows a uniform distribution in position X⃗\vec{X} and time tt, therefore g(t′)≈⟨e−ip⃗⋅X⃗eip⃗⋅X⃗⟩g(t')\approx\left\langle e^{-i\vec{p}\cdot\vec{X}}e^{i\vec{p}\cdot\vec{X}}\right\rangle. In the last equality, G(E)G(E) is the Fourier transformation of g(Δt)g(\Delta t),

G(E)=∫dΔt g(Δt)eiΔtE.(43)G(E)=\int d\Delta t\,g(\Delta t)e^{i\Delta t E}. \tag*{(43)}

By comparing the expression of the Keldysh Green’s function for the boson in Eq. (11), and Eq. (42), we find that the correlation function of the AA field resembles the Keldysh Green’s function GB−+G_B^{-+}, and that WA(X⃗,P⃗,t,E)W_A(\vec{X},\vec{P},t,E) resembles the Wigner transformation of GB−+G_B^{-+}. The only difference is on the energy dependence: in Eq. (11), the boson’s energy is on-shell; whereas in WA(X⃗,P⃗,t,E)W_A(\vec{X},\vec{P},t,E), its energy dependence is given by G(E)G(E). If we assume that the Keldysh Green’s functions for the AA field can be constructed in terms of its correlation functions GA−+(x,y)=−i⟨A∗(y)A(x)⟩G_A^{-+}(x,y)=-i\langle A^{*}(y)A(x)\rangle and GA+−(x,y)=−i⟨A(x)A∗(y)⟩G_A^{+-}(x,y)=-i\langle A(x)A^{*}(y)\rangle, we can obtain the Wigner transformation on iGA−+iG_A^{-+} and iGA+−iG_A^{+-} as,

W[iGA−+]=iGA,W−+(X,P)=WA(X⃗,P⃗,t,E),W[iGA+−]=iGA,W+−(X,P)=WA(X⃗,P⃗,t,E).(44)\begin{aligned} \mathcal{W}\left[iG_A^{-+}\right]&=iG_{A,W}^{-+}(X,P)=W_A(\vec{X},\vec{P},t,E),\\ \mathcal{W}\left[iG_A^{+-}\right]&=iG_{A,W}^{+-}(X,P)=W_A(\vec{X},\vec{P},t,E). \tag*{(44)} \end{aligned}

Here X=(t,X⃗),P=(E,P⃗)X=(t,\vec{X}),P=(E,\vec{P}). We note that in the approximation of treating the gluon field as a classical field, the Wigner transformation of iGA−+iG_A^{-+} and iGA+−iG_A^{+-} is identical. By comparing the AA field’s correlation function WA(X⃗,P⃗,t,E)W_A(\vec{X},\vec{P},t,E) and the Fourier expansion of the boson’s Keldysh Green’s function GBG_B, one can identify G(E)G(E) as, the off-shell gluons’ energy distribution. In order to ensure that the energy distribution function is real and nonnegative, g(Δt)g(\Delta t) needs to be real and even. Now, we show that g(Δt)g(\Delta t) indeed satisfies these properties.

We first show that g(Δt)g(\Delta t) is a real function through the time correlation function of the random phases. Let us first consider the case when ∣t′−t∣≤tcorr|t^{\prime}-t| \le t_{\mathrm{corr}}. Since θ(p⃗,t)\theta(\vec{p},t) and θ(p⃗,t′)\theta(\vec{p},t^{\prime}) are equal to each other when ∣t′−t∣≤tcorr|t^{\prime}-t| \le t_{\mathrm{corr}}, and according to Eq. (39), we have

g(ti+h−ti)=g(p⃗,Δt)=1n−h∑i=1n−h⟨e−iθ(p⃗,ti+h)eiθ(p⃗,ti)⟩=1n−h∑i=1n−h⟨e−iθ(p⃗,ti)eiθ(p⃗,ti)⟩.(45)\begin{aligned} g(t_{i+h}-t_i) &= g(\vec{p},\Delta t) \\ &= \frac{1}{n-h}\sum_{i=1}^{n-h}\left\langle e^{-i\theta(\vec{p},t_{i+h})}e^{i\theta(\vec{p},t_i)}\right\rangle\\ &= \frac{1}{n-h}\sum_{i=1}^{n-h}\left\langle e^{-i\theta(\vec{p},t_i)}e^{i\theta(\vec{p},t_i)}\right\rangle. \tag*{(45)} \end{aligned}

Here Δt=ti+h−ti\Delta t=t_{i+h}-t_i, and note that in the last equality, the two phases are taken to be at the same time. Because ⟨e−iθ(p⃗,ti)eiθ(p⃗,ti)⟩\left\langle e^{-i\theta(\vec{p},t_i)}e^{i\theta(\vec{p},t_i)}\right\rangle is a real function, g(Δt)g(\Delta t) is also a real function. When ∣t′−t∣>tcorr|t^{\prime}-t|>t_{\mathrm{corr}}, these two phases are independent of each other. The ensemble average of their product is

g(t−t′)=g(p⃗,Δt)=⟨e−iθ(p⃗,t)eiθ(p⃗,t′)⟩=⟨e−iθ(p⃗,t)⟩⟨eiθ(p⃗,t′)⟩=0,(46)\begin{aligned} g(t-t^{\prime})=g(\vec{p},\Delta t) &=\left\langle e^{-i\theta(\vec{p},t)}e^{i\theta(\vec{p},t^{\prime})}\right\rangle\\ &=\left\langle e^{-i\theta(\vec{p},t)}\right\rangle\left\langle e^{i\theta(\vec{p},t^{\prime})}\right\rangle=0, \tag*{(46)} \end{aligned}

which is also real. Therefore, regardless of the value of Δt\Delta t, g(Δt)g(\Delta t) is a real function. Thus, according to Eq. (40), g(Δt)g(\Delta t) is an even function and therefore G(E)G(E) is also an even function.

To establish the connection between the SSE and BE, we need to calculate the scattering term of BE from the SSE. Since we aim to derive BE from the SSE in the weak coupling limit, the dominant contribution comes from the term containing two HIH_I’s on the right-hand side of Eq. (34), which is represented by the two-vertex term in Figure 3. In BE, this term corresponds to the one-loop term in the series expansion of the heavy quark self-energy Σw+−\Sigma_{\mathrm{w}}^{+-}, which is represented by the first term in Figure 5. Keeping only the one-loop term in Σw+−\Sigma_{\mathrm{w}}^{+-}, the second term of the right-hand side of Eq. (26) becomes:

Σw+−(X,P)GF,w+−(X,P)=−iΣw+−(X,P)[iGF,w−+(X,P)]=g2∫d3k⃗dk0(2π)4[iGA,w+−(X,−k)][iGF,w+−(X,P−k)]×[iGF,w−+(X,P)]=∫d3k⃗dk0(2π)42g2daEk⃗n(−k⃗)G(−k0)×1(2π)3(2Eq,P−k⃗)[γμ(P−k)μ+mq][1−f(X,P−k)]×1(2π)3(2Eq,P⃗)(γνPν+mq)[−f(X,P)]≈−g2∫d3k⃗dk02(2π)10Ek⃗Eq,P−k⃗Eq,P⃗[γμ(P−k)μ+mq](γνPν+mq)×dan(k⃗)G(k0)f(X,P).(47)\begin{aligned} \Sigma_{\mathrm{w}}^{+-}(X,P)G_{\mathrm{F,w}}^{+-}(X,P) &=-i\Sigma_{\mathrm{w}}^{+-}(X,P)\left[iG_{\mathrm{F,w}}^{-+}(X,P)\right]\\ &=g^2\int\frac{d^3\vec{k}dk^0}{(2\pi)^4}\left[iG_{\mathrm{A,w}}^{+-}(X,-k)\right]\left[iG_{\mathrm{F,w}}^{+-}(X,P-k)\right]\\ &\qquad\times\left[iG_{\mathrm{F,w}}^{-+}(X,P)\right]\\ &=\int\frac{d^3\vec{k}dk^0}{(2\pi)^4}\frac{2g^2d_a}{E_{\vec{k}}}n(-\vec{k})G(-k^0)\\ &\qquad\times\frac{1}{(2\pi)^3(2E_{q,P-\vec{k}})}\left[\gamma^\mu(P-k)_\mu+m_q\right]\left[1-f(X,P-k)\right]\\ &\qquad\times\frac{1}{(2\pi)^3(2E_{q,\vec{P}})}\left(\gamma^\nu P_\nu+m_q\right)\left[-f(X,P)\right]\\ &\approx-g^2\int\frac{d^3\vec{k}dk^0}{2(2\pi)^{10}E_{\vec{k}}E_{q,P-\vec{k}}E_{q,\vec{P}}}\left[\gamma^\mu(P-k)_\mu+m_q\right]\left(\gamma^\nu P_\nu+m_q\right)\\ &\qquad\times d_a n(\vec{k})G(k^0)f(X,P). \tag*{(47)} \end{aligned}

Here gg is the coupling constant, Pμ=(Eq,P⃗,P⃗)P^\mu=(E_{q,\vec{P}},\vec{P}) and kμ=(k0,k⃗)k^\mu=(k^0,\vec{k}) are 4-momentum of the heavy quark and the gluon respectively. Eq,P⃗E_{q,\vec{P}} is the kinetic energy of the heavy quark with the momentum P⃗\vec{P}, and satisfies the mass-shell relation which comes from the Wigner transformation of G−+G^{-+} in Eq. (16). In the third equality, we substitute GF,wG_{\mathrm{F,w}} with the Wigner transformation of the Green’s functions in Eq. (20) and the gluon Keldysh Green’s function with the transformation of the correlation of a AA field in Eq. (42). In the final step, we ignore the Pauli blocking factor 1−f(X,P−k)1-f(X,P-k) because the system we consider contains only one heavy quark. By analogy with Eq. (47), we obtain the result of Σw+−(X,P)GF,w+−(X,P)\Sigma_{\mathrm{w}}^{+-}(X,P)G_{\mathrm{F,w}}^{+-}(X,P) as,

Σw+−(X,P)GF,w+−(X,P)≈−g2∫d3k⃗dk02(2π)10Ek⃗Eq,P−k⃗Eq,P⃗(γνPν+mq)[γμ(P−k)μ+mq]×dan(k⃗)G(k0)f(X,P−k).(48)\begin{aligned} \Sigma_{\mathrm{w}}^{+-}(X,P)G_{\mathrm{F,w}}^{+-}(X,P) &\approx-g^2\int\frac{d^3\vec{k}dk^0}{2(2\pi)^{10}E_{\vec{k}}E_{q,P-\vec{k}}E_{q,\vec{P}}}\left(\gamma^\nu P_\nu+m_q\right)\left[\gamma^\mu(P-k)_\mu+m_q\right]\\ &\qquad\times d_a n(\vec{k})G(k^0)f(X,P-k). \tag*{(48)} \end{aligned}

Next we sum over the spin degrees of freedom of the heavy quark by tracing out the spinor space on both sides of Eq. (26). The trace of the gamma matrices on the right-hand side of Eq. (48) leads to the squared scattering amplitude

∣M∣2=14g2tr⁡{(γμPμ+mq)[γν(P−k)ν+mq]}.(49)|\mathcal{M}|^2=\frac{1}{4}g^2\operatorname{tr}\left\{(\gamma^\mu P_\mu+m_q)\left[\gamma^\nu(P-k)_\nu+m_q\right]\right\}. \tag*{(49)}

Here the factor 14\frac{1}{4} arises from tracing the drift term. Therefore, the scattering process in the SSE can be interpreted as an on-shell heavy quark absorbing (or emitting) an off-shell gluon and forming a new on-shell heavy quark. Since the gluons are off-shell, there is no contradiction between momentum conservation and the heavy quark’s mass-shell relation. Therefore the two-vertex term in Figure 6 does not vanish.

Due to the non-relativistic approximation adopted in the SSE, only the zeroth component of uˉ(P)γμu(P′)\bar{u}(P)\gamma^\mu u(P') enters the interaction Hamiltonian [17]. Therefore, the squared scattering amplitude can be further simplified as,

∣M∣2≈14g2tr⁡{[uˉ(P)γ0u(P)][uˉ(P)γ0u(P)]}=14g2tr⁡(4mq2I)=4g2mq2.(50)\begin{aligned} |M|^2 &\approx\frac{1}{4}g^2\operatorname{tr}\left\{\left[\bar{u}(P)\gamma^0u(P)\right]\left[\bar{u}(P)\gamma^0u(P)\right]\right\}\\ &=\frac{1}{4}g^2\operatorname{tr}\left(4m_q^2\mathbf{I}\right)\\ &=4g^2m_q^2. \tag*{(50)} \end{aligned}

Here I\mathbf{I} is the 4×44 \times4 identity matrix in the spinor space.

In BE the energy and momentum in the scattering process are conserved, and they satisfy

P⃗−k⃗=P⃗′,\vec{P}-\vec{k}=\vec{P}',
P⃗ 2+mq2−k0=P⃗′ 2+mq2.(51)\sqrt{\vec{P}^{\,2}+m_q^2}-k^0=\sqrt{\vec{P}'^{\,2}+m_q^2}. \tag*{(51)}

where P⃗\vec{P} and P⃗′\vec{P}' are incoming and outgoing momentum of the heavy quark, respectively, and their corresponding kinetic energy are P⃗ 2+mq2\sqrt{\vec{P}^{\,2}+m_q^2} and P⃗′ 2+mq2\sqrt{\vec{P}'^{\,2}+m_q^2} respectively. k⃗\vec{k} and k0k^0 are the momentum and energy of the off-shell gluon participating in the scattering process, respectively.

Because Eq,P⃗E_{q,\vec{P}} satisfies the mass-shell relation and only depends on the 3-momentum P⃗\vec{P}, the phase space distribution of heavy quarks depends only on the 3-momentum P⃗\vec{P}, 3-position X⃗\vec{X}, and time tt. In the following part of this paper, we use f~(X⃗,P⃗,t)≡f(X,P)∣P0=Eq,P⃗\tilde{f}(\vec{X},\vec{P},t)\equiv f(X,P)|_{P^0=E_{q,\vec{P}}} to denote the heavy quark’s phase space distribution. For conciseness, we will drop the tilde on f~(X⃗,P⃗,t)\tilde{f}(\vec{X},\vec{P},t) from now on. By substituting the product of the self-energies and Green’s functions into Eq. (26), performing the trace over the spinor space, setting P⃗′=P⃗−k⃗\vec{P}'=\vec{P}-\vec{k}, E=k0E=k^0 and integrating over P0P^0, we obtain the Boltzmann equation in the following form

∂f(X⃗,P⃗,t)∂t+P⃗Eq,P⃗⋅∂f(X⃗,P⃗,t)∂X⃗=dg2(2π)4Eq,P⃗∫d3P⃗′Eq,P⃗′∫d3k⃗Ek⃗∫dE W′(P⃗,P⃗′,k⃗,Eq,P⃗,Eq,P⃗′,E)×[f(X⃗,P⃗′,t)n(k⃗)G(E)−f(X⃗,P⃗,t)n(k⃗)G(E)].(52)\begin{aligned} \frac{\partial f(\vec{X},\vec{P},t)}{\partial t}+\frac{\vec{P}}{E_{q,\vec{P}}}\cdot\frac{\partial f(\vec{X},\vec{P},t)}{\partial\vec{X}}={}&\frac{d_g}{2(2\pi)^4E_{q,\vec{P}}}\int\frac{d^3\vec{P}'}{E_{q,\vec{P}'}}\int\frac{d^3\vec{k}}{E_{\vec{k}}}\int dE\,W'(\vec{P},\vec{P}',\vec{k},E_{q,\vec{P}},E_{q,\vec{P}'},E)\\ &\times\left[f(\vec{X},\vec{P}',t)n(\vec{k})G(E)-f(\vec{X},\vec{P},t)n(\vec{k})G(E)\right]. \tag*{(52)} \end{aligned}

Here X⃗\vec{X} is the 3-position, tt is the time, P⃗\vec{P} is the 3-momentum, and k⃗\vec{k} denotes the momentum of the gluon absorbed or emitted by the heavy quark. Eq,P⃗E_{q,\vec{P}} is the kinetic energy of the quark, and in the non-relativistic approximation, Eq,P⃗≈mqE_{q,\vec{P}}\approx m_q.

The function W′W' is related to the squared scattering amplitude ∣M∣2|M|^2 [27, 31] as,

W′(P⃗,P⃗′,k⃗,Eq,P⃗,Eq,P⃗′,E)=∣M∣2(2π)3(2π)4δ3(P⃗−P⃗′−k⃗)δ(Eq,P⃗−Eq,P⃗′−E).(53)W'(\vec{P},\vec{P}',\vec{k},E_{q,\vec{P}},E_{q,\vec{P}'},E)=\frac{|M|^2}{(2\pi)^3}(2\pi)^4\delta^3(\vec{P}-\vec{P}'-\vec{k})\delta(E_{q,\vec{P}}-E_{q,\vec{P}'}-E). \tag*{(53)}

In the above derivation, the gluon energy distribution G(E)G(E) serves as a bridge connecting the SSE with the Boltzmann equation. It provides the energy distribution of the gluons in the Boltzmann equation following from the time dependence of the random phases of the gluon field in the stochastic Schrödinger equation.

Let us now consider the equilibrium limit of the BE and SSE. Applying the Boltzmann’s HH-theorem [19] to the BE in Eq. (52), we find that the equilibrium distribution of the heavy quark in this system satisfies

feq(X⃗,P⃗′,t)=feq(X⃗,P⃗,t).(54)f_{\mathrm{eq}}(\vec{X},\vec{P}',t)=f_{\mathrm{eq}}(\vec{X},\vec{P},t). \tag*{(54)}

This equation implies that the equilibrium distribution is uniform in momentum space (assuming a weak dependence on the position XX), which arises from the absence of the dissipation term in the SSE considered in this paper, see the discussion below Eq. (32). As shown in [16], the SSE with the dissipation term included indeed leads to the Boltzmann distribution in the equilibrium limit [16] for the heavy quark.

Based on the above derivation, we demonstrate that in the weak coupling constant condition, the SSE in Eq. (32) is consistent with the Boltzmann equation in Eq. (52). In the next section, we will verify this consistency through numerical calculations.

Numerical calculation

In this section, we compare the evolution of the heavy quark distributions obtained from the SSE and the Boltzmann equation through numerical calculations. For demonstration purposes, we perform numerical calculations in 1+1 dimensional space-time.

In this section, we adopt the following values for the parameters: Our calculations are performed in discretized momentum space with the step size Δp=0.0196 GeV\Delta p=0.0196\ \mathrm{GeV}, where the momentum ranges in p⃗∈[−π GeV,π GeV]\vec{p}\in[-\pi\ \mathrm{GeV},\pi\ \mathrm{GeV}]. The coupling constant for the thermal mass of the gluon is g0=2.22g_0=2.22, and αs0=g024π=0.4\alpha_{s0}=\frac{g_0^2}{4\pi}=0.4, the medium temperature is T=0.15 GeVT=0.15\ \mathrm{GeV}, the heavy quark mass is mq=1.0 GeVm_q=1.0\ \mathrm{GeV}, the gluon mass is mg=g01+Nf6T=0.41 GeVm_g=g_0\sqrt{1+\frac{N_f}{6}}T=0.41\ \mathrm{GeV} with the number of quark flavor Nf=3N_f=3. Unless otherwise specified, the coupling constant between the heavy quark and the gluon field is αs=g24π=0.4\alpha_s=\frac{g^2}{4\pi}=0.4, the correlation time for the gluon field is tcorr=0.067 GeV−1t_{\mathrm{corr}}=0.067\ \mathrm{GeV}^{-1}, and we take the ensemble average over Nevent=100N_{\mathrm{event}}=100 events for the numerical results shown in this section.

Let us begin with analyzing the time correlation of the random phases. For each discretized momentum p⃗i\vec{p}_i, the random sequences of the gluon phase θj(p⃗i,t)\theta_j(\vec{p}_i,t) of the 100 events are generated:

{θj(p⃗i,t)∣t=0,t0,2t0,…,nt0}.(55)\{\theta_j(\vec{p}_i,t)|t=0,t_0,2t_0,\ldots,nt_0\}. \tag*{(55)}

Here tt is the time, nn is the number of time steps of each event, jj denotes the jjth event, and t0t_0 is the time step size (much smaller than tcorrt_{\rm corr}). The corresponding sequences of eiθj(p⃗i,t)e^{i\theta_j(\vec{p}_i,t)} read,

{eiθj(p⃗i,t)∣t=0,t0,2t0,…,nt0}.(56)\{e^{i\theta_j(\vec{p}_i,t)}|t=0,t_0,2t_0,\ldots,nt_0\}. \tag*{(56)}

In each event, we resample the random phases after a period of tcorrt_{\rm corr}. In addition, we randomize the time when the phases are updated in each event, such that in different events, the random phases are updated at different times. Using Eq. (39), we can obtain the time correlation function gj(Δt)g_j(\Delta t) of eiθj(p⃗i,t)e^{i\theta_j(\vec{p}_i,t)} for the jjth event. Then, the ensemble average of the time correlation functions is

g(Δt)=1Nevent∑j=1Neventgj(Δt)(Δt≥0).(57)g(\Delta t)=\frac{1}{N_{\rm event}}\sum_{j=1}^{N_{\rm event}}g_j(\Delta t)\quad(\Delta t\geq0). \tag*{(57)}

Since g∗(Δt)=g(−Δt)g^*(\Delta t)=g(-\Delta t), we can obtain the function g(Δt<0)g(\Delta t<0) from g(Δt>0)g(\Delta t>0). The resulting g(Δt)g(\Delta t) is shown in Figure 8. As we can see from Figure 8, the real part of

Time correlation function of the random phases

Figure 8. The ensemble average of the time correlation function g(Δt)g(\Delta t) of the random phases of the gluon field over 100 events, with tcorr=0.067 GeV−1t_{\rm corr}=0.067\ {\rm GeV}^{-1}. Top: real part of the correlation function ℜ[g(Δt)]\Re[g(\Delta t)]. Bottom: imaginary part of the correlation function ℑ[g(Δt)]\Im[g(\Delta t)].

g(Δt)g(\Delta t) reaches a maximum value of around 1.0 at Δt=0\Delta t=0, and the correlation decreases as ∣Δt∣|\Delta t| increases. When ∣Δt∣≥tcorr|\Delta t|\geq t_{\rm corr}, the correlation is close to 0. The imaginary part of g(Δt)g(\Delta t) fluctuates around ℑ[g(Δt)]=0\Im[g(\Delta t)]=0, and its amplitudes are much smaller than ℜ[g(Δt)]\Re[g(\Delta t)]. Through numerical calculation, we find that Im⁡[g(Δt)]\operatorname{Im}[g(\Delta t)] tends to zero when the number of events increases, which confirms the fact that G(E)G(E) is an even function.

The energy distribution of the gluon G(E)G(E), as shown in Figure 9, is the Fourier transform of the correlation function g(Δt)g(\Delta t), see Eq. (43). In Figure 9, we use the ansatz G(E)=asinc⁡2(bE)G(E)=a\operatorname{sinc}^{2}(bE) to fit the numerical results and determine the values of a,ba,b for different tcorrt_{\rm corr}. Through fitting the numerical results, we obtain

Energy distributions of the gluon for four correlation times

Figure 9. (Color online) The energy distribution of the gluon G(E)G(E) from the SSE, with tcorr=0.027 GeV−1t_{\mathrm{corr}}=0.027\,\mathrm{GeV}^{-1}, 0.067 GeV−10.067\,\mathrm{GeV}^{-1}, 0.13 GeV−10.13\,\mathrm{GeV}^{-1}, 0.2 GeV−10.2\,\mathrm{GeV}^{-1} from top panel to bottom panel. The unit of aa and bb in the legend is GeV−1\mathrm{GeV}^{-1}. Purple line: Fourier transform of the time correlation function of the random phase factor g(Δt)g(\Delta t); green line: fit result with the ansatz G(E)=asinc⁡2(bE)G(E)=a\operatorname{sinc}^{2}(bE), see text for the detail.

a=1.0155tcorr+0.0157 GeV−1,b=0.4899tcorr+0.007 GeV−1.(58)\begin{aligned} a&=1.0155t_{\rm corr}+0.0157\ {\rm GeV}^{-1},\\ b&=0.4899t_{\rm corr}+0.007\ {\rm GeV}^{-1}. \tag*{(58)} \end{aligned}

The parameters aa and bb are found to be proportional to tcorrt_{\rm corr}, as illustrated in Figure 10. As expected, the width of the energy distribution of the gluon G(E)G(E) is inversely proportional to that in g(Δt)g(\Delta t). By substituting the fitted G(E)G(E) into our Boltzmann equation Eq. (52), we can numerically solve the time evolution of the heavy quark phase distribution f(x⃗,P⃗,t)f(\vec{x},\vec{P},t). In the remaining part of this section, for conciseness, we will use the lower-case x⃗\vec{x} (p⃗\vec{p}) to denote the 3-position (momentum) of the heavy quark.

Fit parameters as functions of the correlation time

Figure 10. The relation between the fit parameters a,ba,b and the correlation time tcorrt_{\mathrm{corr}}, see text for the detail. Top panel: aa as a function of the correlation time tcorrt_{\mathrm{corr}}. Bottom panel: bb as a function of the correlation time tcorrt_{\mathrm{corr}}.

The momentum distribution fP(p⃗,t)f_P(\vec{p},t) of the heavy quark follows from its phase space distribution f(x⃗,p⃗,t)f(\vec{x},\vec{p},t) as,

fP(p⃗,t)=∫f(x⃗,p⃗,t)dx⃗∫f(x⃗,p⃗,t)dx⃗dp⃗.(59)f_P(\vec{p},t)=\frac{\int f(\vec{x},\vec{p},t)d\vec{x}}{\int f(\vec{x},\vec{p},t)d\vec{x}d\vec{p}}. \tag*{(59)}

In our numerical calculations, we adopt the following Gaussian distribution as the momentum distribution at the initial time t=0t=0,

fP(p⃗,t=0)=1(2π)0.5σexp⁡[−(p⃗−p⃗0)22σ2].(60)f_P(\vec{p},t=0)=\frac{1}{(2\pi)^{0.5}\sigma}\exp\left[-\frac{(\vec{p}-\vec{p}_0)^2}{2\sigma^2}\right]. \tag*{(60)}

where p⃗0\vec{p}_0 represents the momentum at which fP(p⃗,t=0)f_P(\vec{p},t=0) reaches its peak value, and σ\sigma is the width of the initial momentum distribution. For BE, we additionally assume a uniform distribution in the coordinate space at t=0t=0.

Next, we compare the average squared momentum ⟨p2(t)⟩\langle p^2(t)\rangle and momentum distributions fP(p⃗,t)f_P(\vec{p},t) obtained from the SSE and BE. For given momentum distribution, ⟨p2(t)⟩\langle p^2(t)\rangle is defined as,

⟨p2(t)⟩=∫∣p⃗∣2fP(p⃗,t)dp⃗.(61)\langle p^2(t)\rangle=\int|\vec{p}|^2 f_P(\vec{p},t)d\vec{p}. \tag*{(61)}

In order to minimize the impact from the truncation artifacts in momentum space at p⃗=±π GeV\vec{p}=\pm\pi\ {\rm GeV}, we restrict the integration interval in Eq. (61) to [−π3 GeV,π3 GeV][-\frac{\pi}{3}\ {\rm GeV},\frac{\pi}{3}\ {\rm GeV}]. The expected equilibrium momentum distribution of the heavy quark follows the uniform distribution, cf. Eq. (54), so the corresponding ensemble-averaged momentum squared is ⟨p2⟩eq=0.37 GeV2\langle p^2\rangle_{\rm eq}=0.37\ {\rm GeV}^2.

We compare the time evolution of ⟨p2⟩\langle p^{2}\rangle with p⃗0=0\vec{p}_{0}=0 and σ=0.043 GeV\sigma=0.043\,\mathrm{GeV} in Figure 11 and (12). As time evolves, ⟨p2⟩\langle p^{2}\rangle increases from zero toward the equilibrium value ⟨p2⟩eq=0.37 GeV2\langle p^{2}\rangle_{\mathrm{eq}}=0.37\,\mathrm{GeV}^{2}. From the Boltzmann equation in Eq. (52), we can see that in the weak coupling limit, the scattering cross section is proportional to αs\alpha_{s} if the gluon

Time evolution of the heavy-quark momentum squared for three coupling constants

Figure 11. (Color online) The time evolution of ⟨p2(t)⟩\langle p^{2}(t)\rangle from the SSE with different coupling constants between the heavy quark and the gluon field. The evolution time is rescaled with the factor of αs/αs0\alpha_{s}/\alpha_{s0}. The initial momentum distribution at t=0t=0 is centered around p⃗0=0\vec{p}_{0}=0, see text for the detail. The dashed line indicates the equilibrium value ⟨p2⟩eq=0.37 GeV2\langle p^{2}\rangle_{\mathrm{eq}}=0.37\,\mathrm{GeV}^{2}.

mass mgm_{g} does not depend on the coupling constant. In Figure 11, we plot the evolutions of ⟨p2[(αs/αs0)t]⟩\langle p^{2}[(\alpha_{s}/\alpha_{s0})t]\rangle with three different coupling constants αs=0.1\alpha_{s}=0.1, 0.2250.225 and 0.40.4. Here, we rescale the evolution time by the ratio of the coupling constant, αs/αs0\alpha_{s}/\alpha_{s0}, where αs0\alpha_{s0} is the reference coupling constant appearing in the thermal mass of the gluon. We find that the three curves almost overlap with each other; that is, the evolution rate of ⟨p2(t)⟩\langle p^{2}(t)\rangle is approximately proportional to the ratio of αs/αs0\alpha_{s}/\alpha_{s0}. These results indicate that the weak coupling constant condition is approximately satisfied with αs≤0.4\alpha_{s}\leq0.4.

In Figure 12, we compare the time evolution of ⟨p2⟩\langle p^2\rangle from the SSE with that from BE. As αs\alpha_s increases, the evolution rate calculated from the SSE gradually slows down compared to that from BE. This trend suggests that when αs\alpha_s becomes large, nonperturbative effects may start to emerge during the time evolution. In Figure 12, all the results are averaged over 1000 events to further reduce the thermal fluctuations.

Comparison of the time evolution of $\langle p^2(t)\rangle$ from the SSE and the Boltzmann equation

Figure 12. (Color online) Comparison of the time evolution of ⟨p2(t)⟩\langle p^2(t)\rangle from the SSE in Eq. (32) and the Boltzmann equation (BE) in Eq. (52) with αs=0.4,1.6,3.6\alpha_s=0.4, 1.6, 3.6. The dashed line indicates the equilibrium value ⟨p2⟩eq=0.37 GeV2\langle p^2\rangle_{\rm eq}=0.37\ {\rm GeV}^2. The initial momentum distribution at t=0t=0 is centered around p⃗0=0\vec{p}_0=0, see text for the detail.

In Figure 13, we compare the time evolution of ⟨p2⟩\langle p^2\rangle from an initial momentum distribution centered around p⃗0=0.2π GeV\vec{p}_0=0.2\pi\ {\rm GeV} at αs=0.4\alpha_s=0.4. The resulting ⟨p2⟩\langle p^2\rangle decreases with time and approaches the equilibrium value of ⟨p2⟩eq=0.37 GeV2\langle p^2\rangle_{\rm eq}=0.37\ {\rm GeV}^2. The overlapping ⟨p2⟩\langle p^2\rangle evolution trends confirm that the consistency between the BE and SSE in the weak coupling condition is independent of the initial momentum distribution of the heavy quark.

Time evolution of $\langle p^2(t)\rangle$ from the Boltzmann equation and SSE

Figure 13. (Color line) Time evolution of ⟨p2(t)⟩\langle p^2(t)\rangle from the Boltzmann equation and SSE. The initial momentum distribution is peaked at p⃗0=0.2π GeV\vec{p}_0=0.2\pi\ {\rm GeV}. Purple line: result from the Boltzmann equation. Green line: result from the SSE. The dashed line indicates the equilibrium value ⟨p2⟩eq=0.37 GeV2\langle p^2\rangle_{\rm eq}=0.37\ {\rm GeV}^2.

The time evolution of the heavy quark momentum distribution fP(p⃗,t)f_P(\vec{p},t) from the SSE and BE is compared in Figure 14 (with p⃗0=0\vec{p}_0=0) and Figure 15 (with p⃗0=0.2π GeV\vec{p}_0=0.2\pi\ {\rm GeV}). The strong coupling constant used in this calculation is αs=0.4\alpha_s=0.4. In both cases, the width of momentum distributions from the BE and SSE increases with the time tt and the momentum distributions approach the uniform distribution as tt increases, which is consistent with the Boltzmann’s HH-theorem, cf. Eq. (54). One can see that for both initial conditions, the momentum distributions from the BE and SSE are close to each other at each intermediate time, which further confirms the consistency between the BE and SSE approach in the weak coupling condition..

Time evolution of the momentum distribution $f_P(\vec{p},t)$ of the heavy quark from the Boltzmann equation and SSE

Figure 14. (Color online) Time evolution of the momentum distribution fP(p⃗,t)f_P(\vec{p},t) of the heavy quark from the Boltzmann equation and SSE. The peak of the initial distribution is located at p⃗0=0\vec{p}_0=0, see text for the detail. Top panel: results from the Boltzmann equation. Bottom panel: results from the SSE.

Time evolution of the heavy quark momentum distribution

Figure 15. (Color online) Time evolution of the heavy quark momentum distribution fP(p⃗,t)f_P(\vec{p},t) from the Boltzmann equation and SSE. The peak of the initial momentum distribution is located at p⃗0=0.2π GeV\vec{p}_0=0.2\pi\ \mathrm{GeV}, see text for the detail. Top panel: results from the Boltzmann equation. Bottom panel: results from the SSE.

Conclusion

In this study, we investigate the relationship between the Boltzmann equation (BE) approach and the recently constructed stochastic Schrödinger equation (SSE) approach in the Keldysh Green’s function framework. We find that the SSE approach describes the scattering processes between the heavy quark and the off-shell gluons in the thermal medium. The time scale for the phase rotation of the background gluon field is inversely proportional to the energy dispersion of the gluons in the thermal medium. When the coupling constant between the heavy quark and the gluon field is small, this process can be equivalently described by the BE approach with the scattering term obtained from the leading-order perturbative expansion of the heavy quark self-energy in the thermal medium. Furthermore, we perform the numerical calculation of the time evolution of the heavy quark momentum distribution in both the SSE and BE approaches. The resulting momentum distributions are consistent with each other in the weak coupling limit, which confirms the connection between the SSE and BE found in the Keldysh Green’s function framework. As the coupling increases, the results from the BE and SSE start to deviate, which signals the possible emergence of nonperturbative effects.

Compared to BE, the SSE approach has several advantages: First, in the SSE the heavy quark system evolves on the amplitude level and therefore this approach can potentially capture a more complete set of quantum effects, such as the (de)excitation of bound states through the interaction with the medium. Secondly, the SSE provides a straightforward nonperturbative framework to study the time evolution of strong coupling systems in the thermal medium.

The application of the SSE can be extended in several directions: First, we plan to apply the SSE to bound state systems consisting of heavy quarks, such as the heavy quarkonium systems, and study their formation and dissociation in the thermal medium. Secondly, we can implement more realistic background fields to simulate the thermal medium. For example, the transverse components of the gluon field as well as the time dependence and the flow effects of the thermal medium can be considered. Finally, besides the gluon field, we can include time-dependent electromagnetic fields in the SSE and study their effects on the evolution of the heavy quark systems in the thermal medium.

Acknowledgment

We thank Baoyi Chen and Min He for useful discussions. X. Z. is supported by new faculty startup funding by the Institute of Modern Physics, Chinese Academy of Sciences, by Key Research Program of Frontier Sciences, Chinese Academy of Sciences, Grant No. ZDBS-LY-7020, by the Foundation for Key Talents of Gansu Province, by the Central Funds Guiding the Local Science and Technology Development of Gansu Province, Grant No. 22ZY1QA006, by Gansu International Collaboration and Talents Recruitment Base of Particle Physics (2023-2027), by International Partnership Program of the Chinese Academy of Sciences, Grant No. 016GJHZ2022103FN, by National Natural Science Foundation of China, Grant No. 12375143, by National Key R&D Program of China, Grant No. 2023YFA1606903 and by the Strategic Priority Research Program of the Chinese Academy of Sciences, Grant No. XDB34000000.

References

  1. [1]R. Rapp and H. van Hees, arXiv preprint arXiv:0803.0901 (2008).arxiv.org/abs/0803.0901
  2. [2]K. Yagi, T. Hatsuda, and Y. Miake, Quark-gluon plasma: From big bang to little bang, vol. 23 (Cambridge University Press, 2005).
  3. [3]X. Zhao and R. Rapp, Nuclear Physics A 859, 114 (2011).
  4. [4]X. Du, M. He, and R. Rapp, Physical Review C 96, 054901 (2017).
  5. [5]B. Wu, X. Du, M. Sibila, and R. Rapp, The European Physical Journal A 57, 1 (2021).
  6. [6]B. Wu, Z. Tang, M. He, and R. Rapp, arXiv preprint arXiv:2302.11511 (2023).DOI
  7. [7]M. He, H. van Hees, P. B. Gossiaux, R. J. Fries, and R. Rapp, Physical Review E 88, 032138 (2013).
  8. [8]X. Yao, W. Ke, Y. Xu, S. A. Bass, and B. Müller, Journal of High Energy Physics 2021, 1 (2021).
  9. [9]M. Kostin, The journal of chemical physics 57, 3589 (1972).
  10. [10]R. Katz and P.-B. Gossiaux, Annals of Physics 368, 267 (2016).
  11. [11]Y. Akamatsu, M. Asakawa, S. Kajimoto, and A. Rothkopf, Journal of High Energy Physics 2018, 1 (2018).
  12. [12]Y. Akamatsu, Physical Review D 91, 056002 (2015).
  13. [13]D. De Boni, Journal of High Energy Physics 2017, 1 (2017).
  14. [14]J. Schwinger, Journal of Mathematical Physics 2, 407 (1961).
  15. [15]L. V. Keldysh et al., Sov. Phys. JETP 20, 1018 (1965).
  16. [16]B. Wu, In preparation (2024).
  17. [17]M. E. Peskin, An introduction to quantum field theory (CRC press, 2018).
  18. [18]J. I. Kapusta and P. Landshoff, Journal of Physics G: Nuclear and Particle Physics 15, 267 (1989).
  19. [19]L. D. Landau, E. Lifshitz, and L. Pitaevskij, Course of theoretical physics. vol. 10: Physical kinetics (Oxford, 1981).DOI
  20. [20]V. Špička, B. Velický, and A. Kalvová, International Journal of Modern Physics B 28, 1430013 (2014).
  21. [21]H. Kleinert, Path integrals in quantum mechanics, statistics, polymer physics, and financial markets (World scientific, 2009).
  22. [22]H. van Hees (2014).
  23. [23]K. Geiger, Physical Review D 54, 949 (1996).
  24. [24]J. Gamble, arXiv preprint arXiv:0805.3178 (2008).
  25. [25]C. Cercignani, G. M. Kremer, C. Cercignani, and G. M. Kremer, Relativistic boltzmann equation (Springer, 2002).DOI
  26. [26]R. L. Liboff, Kinetic theory: classical, quantum, and relativistic descriptions (Springer Science & Business Media, 2003).
  27. [27]H. van Hees (2015).
  28. [28]S. Weinberg, The quantum theory of fields, vol. 2 (Cambridge university press, 1995).
  29. [29]T. Litaka, Phys. Rev. E 49, 4684 (1994), URL https://link.aps.org/doi/10.1103/PhysRevE.49.4684.
  30. [30]J. A. Gubner, Probability and random processes for electrical and computer engineers (Cambridge University Press, 2006).DOI
  31. [31]K. Olive, K. Agashe, C. Amsler, M. Antonelli, J. Arguin, D. Asner, H. Baer, H. Band, R. Barnett, T. Basaglia, et al., Chinese physics C 38, 1 (2014).

Paper details

vlnc:W_s4fphvyb
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents