INTRODUCTION
The transport of energetic charged particles in turbulent magnetic fields of plasmas has been widely recognized as important in space plasma physics and experimental plasmas [1 ] (Jokipii 1966)
Email: qingang@hit.edu.cn
Webb et al. [26 ] ; Shalchi [11 ] ; Zank [27 ] ; Shalchi [14 ] . Of particular interest are transport coefficients for the spatial, momentum, and pitch-angle scattering, which control the penetration and modulation of cosmic rays in the heliosphere, and the efficiency of diffusive shock acceleration (Jokipii 1966; Schlickeiser [9 ] ; Matthaeus et al. [1 ] ; Qin [5 ] ; Shalchi [12 , 14 ] ). So far, the transport coefficients have two different expressions, i.e., Taylor-Green-Kubo formulation (hereafter Kubo formula) and formulas in terms of the moments (hereafter FIM) (Taylor [18 ] ; Green 1951; Kubo 1962). Based on the Kubo formula, various perpendicular and parallel models have been developed, the BAM model (Bieber & Matthaeus 1997), the nonlinear guiding center theory (NLGC) (Matthaeus et al. [1 ] ), the nonlinear parallel diffusion theory (Qin [5 ] ), the unified nonlinear theory (Shalchi [12 ] ), the modification of NLGC (Qin & Zhang [6 ] ), and others. However, in studies of spatial transport coefficients with adiabatic focusing (i.e., a spatially nonuniform background magnetic field), scientists have found that the Kubo formula and FIM yield different results (Roelof [7 ] ; Earl 1976; Kunstmann 1979; Beeck & Wibberenz 1986; Bieber & Burger 1990; Kóta 2000; Schlickeiser & Shalchi [10 ] ; Shalchi [13 ] ; Litvinenko 2012a,b; Danos et al. 2013; Shalchi & Danos [15 ] ; He & Schlickeiser 2014; Wang & Qin [20 ] ; Wang et al. [25 ] ). Wang & Qin [21 , 22 , 23 ] and demonstrated that FIM has a broader range of applicability. Thereafter, the relationship between FIM and the moments has been thoroughly investigated(Wang & Qin [23 ] ).
For the linear spatial transport equation with constant coefficients, Wang & Qin [24 ] derived formulas for the coefficients of for first, second, and higher-order derivative terms. They found that these coefficients are all equal to the time derivative of statistical quantities that are functions of the moments. For example, the coefficient of the convection equation is the time derivative of the expectation; the coefficient of the diffusion equation is the time derivative of the variance; and the coefficients of higher-order transport equations are all the time derivatives of different statistical quantities (Wang & Qin [24 ] ). When we define a statistical quantity as Q Q Q and a constant transport coefficient as κ \kappa κ , the results obtained by Wang & Qin [24 ] can be expressed in a unified form at late times either as
κ = d d t Q , (1) \kappa= \frac{\mathrm{d}}{\mathrm{d}t}Q, \tag*{(1)} κ = d t d Q , (1)
or as
Q = κ t . (2) Q = \kappa t. \tag*{(2)} Q = κ t . (2)
The latter equation indicates that a statistical quantity is a linear function of time at late times. These results were derived from the spatial transport equations. However, the relationship between momentum transport coefficients and momentum moments has not been explored. In this paper, we investigate the relationship between momentum transport coefficients and momentum moments. In addition, in general, the coefficients of transport equations are not constant and depend on time and position. Therefore, we need to explore the characteristics of variable transport coefficients.
In this article, we investigate the variable transport coefficients. This article is organised as follows. In § 2 , the momentum transport equations is derived from the Fokker-Planck equation with variable coefficients. The logarithmic relationship between the coefficients and statistical quantities is observed in our calculations. In addition, some new statistical quantities are derived to express the momentum transport coefficients. In § 3 , the pitch-angle scattering coefficient taking a logarithmic form is derived. In § 4 , we derive the coefficients of the variable coefficient Fokker-Planck equations and again obtain the logarithmic form. In addition, some well-known Fokker-Planck-type equations with variable coefficients in statistical mechanics are investigated, and the corresponding coefficient formulas are deduced. In § 5 , the coefficients of fractional equations are explored. The results are presented in § 6 .
Moment-based formulas for momentum transport coefficients
In this section, we explore the momentum transport equations and derive the formulas for the coefficients in terms of statistical quantities derived from the variable moments.
Transport equation for energetic particles in momentum space
The transport of energetic cosmic-ray particles in turbulent magnetized plasmas, such as the interplanetary and interstellar media, can be described by the following Fokker-Planck equation
∂ f ∂ t = − ∇ ⋅ ( κ r f ) − ∇ p ⋅ ( κ p f ) + ∇ p 2 ( κ p p f ) . (3) \frac{\partial f}{\partial t} = -\nabla\cdot(\kappa_r f) - \nabla_p \cdot(\kappa_p f) + \nabla_p^2 (\kappa_{pp} f). \tag*{(3)} ∂ t ∂ f = − ∇ ⋅ ( κ r f ) − ∇ p ⋅ ( κ p f ) + ∇ p 2 ( κ pp f ) . (3)
Here, f = f ( r , p , t ) f=f(\mathbf{r},\mathbf{p},t) f = f ( r , p , t ) is the particle distribution function with position r \mathbf{r} r and momentum p \mathbf{p} p , κ r \kappa_r κ r is the spatial convection coefficient, κ p \kappa_p κ p is the momentum convection coefficient, and κ p p \kappa_{pp} κ pp is the momentum diffusion coefficient. After averaging over perpendicular spatial coordinates and the gyrophase in spherical momentum coordinates, we obtain from Equation (3)
∂ f ∂ t = − v μ ∂ f ∂ z − 1 p 2 ∂ ∂ p ( p 2 D p f ) + ∂ ∂ μ ( D μ μ ∂ f ∂ μ + D μ p ∂ f ∂ p ) + 1 p 2 ∂ ∂ p ( p 2 D p p ∂ f ∂ p + p 2 D p μ ∂ f ∂ μ ) . (4) \frac{\partial f}{\partial t}=-v\mu\frac{\partial f}{\partial z}-\frac{1}{p^2}\frac{\partial}{\partial p}(p^2D_p f)+\frac{\partial}{\partial\mu}\left(D_{\mu\mu}\frac{\partial f}{\partial\mu}+D_{\mu p}\frac{\partial f}{\partial p}\right)+\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^2D_{pp}\frac{\partial f}{\partial p}+p^2D_{p\mu}\frac{\partial f}{\partial\mu}\right). \tag*{(4)} ∂ t ∂ f = − v μ ∂ z ∂ f − p 2 1 ∂ p ∂ ( p 2 D p f ) + ∂ μ ∂ ( D μμ ∂ μ ∂ f + D μ p ∂ p ∂ f ) + p 2 1 ∂ p ∂ ( p 2 D pp ∂ p ∂ f + p 2 D p μ ∂ μ ∂ f ) . (4)
This equation contains spatial convection, momentum convection, momentum diffusion, pitch-angle diffusion, and mixed-derivative terms with coefficients D μ p D_{\mu p} D μ p or D p μ D_{p\mu} D p μ . Here, the pitch-angle convection is neglected.
To explore whether any new results could be obtained and to simplify the derivation, we use the following simpler model as our starting point
∂ f ∂ t = − v μ ∂ f ∂ z − 1 p 2 ∂ ∂ p ( p 2 D p f ) + ∂ ∂ μ ( D μ μ ∂ f ∂ μ ) + 1 p 2 ∂ ∂ p ( p 2 D p p ∂ f ∂ p ) , (5) \frac{\partial f}{\partial t}=-v\mu\frac{\partial f}{\partial z}-\frac{1}{p^2}\frac{\partial}{\partial p}(p^2D_p f)+\frac{\partial}{\partial\mu}\left(D_{\mu\mu}\frac{\partial f}{\partial\mu}\right)+\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^2D_{pp}\frac{\partial f}{\partial p}\right), \tag*{(5)} ∂ t ∂ f = − v μ ∂ z ∂ f − p 2 1 ∂ p ∂ ( p 2 D p f ) + ∂ μ ∂ ( D μμ ∂ μ ∂ f ) + p 2 1 ∂ p ∂ ( p 2 D pp ∂ p ∂ f ) , (5)
with the momentum convection and diffusion coefficients, and pitch-angle scattering coefficient as
D μ μ = D μ μ ( μ ) , (6) D_{\mu\mu}=D_{\mu\mu}(\mu), \tag*{(6)} D μμ = D μμ ( μ ) , (6)
D p = p A ( μ ) , (7) D_p=pA(\mu), \tag*{(7)} D p = p A ( μ ) , (7)
D p p = N p 2 D μ μ ( μ ) . (8) D_{pp}=Np^2D_{\mu\mu}(\mu). \tag*{(8)} D pp = N p 2 D μμ ( μ ) . (8)
Here, f = f ( z , p , μ , t ) f=f(z,p,\mu,t) f = f ( z , p , μ , t ) , A ( μ ) A(\mu) A ( μ ) is a function of μ \mu μ , and N N N is constant [9 ] . In Equation (5) , the mixed-derivative terms with D μ p D_{\mu p} D μ p or D p μ D_{p\mu} D p μ are neglected. A more detailed study of Equation (4) will be left for future work. In this paper, D μ μ D_{\mu\mu} D μμ is used frequently to derive the momentum transport equation from the Fokker-Planck equation, as shown in Appendix A. In order to simplify the derivation, a simpler form of D μ μ D_{\mu\mu} D μμ is used; that is, we assume D μ μ = D ( 1 − μ 2 ) D_{\mu\mu}=D(1-\mu^2) D μμ = D ( 1 − μ 2 ) with constant D D D . Simplified forms of D p D_p D p and D p p D_{pp} D pp are also employed in this paper. In future work, the coefficient formulas will be explored for cases in which D μ μ D_{\mu\mu} D μμ , D p D_p D p and D p p D_{pp} D pp depend on other variables or take other forms.
The distribution function f ( z , p , μ , t ) f(z,p,\mu,t) f ( z , p , μ , t ) in Equation (5) can be decomposed into the isotropic part F ( z , p , t ) F(z,p,t) F ( z , p , t ) and the anisotropic part g ( z , p , μ , t ) g(z,p,\mu,t) g ( z , p , μ , t )
f ( z , p , μ , t ) = F ( z , p , t ) + g ( z , p , μ , t ) . (9) f(z,p,\mu,t)=F(z,p,t)+g(z,p,\mu,t). \tag*{(9)} f ( z , p , μ , t ) = F ( z , p , t ) + g ( z , p , μ , t ) . (9)
Inserting the latter formula into Equation (5) and integrating over z z z gives (the detailed derivation is shown in Appendix A)
∂ F ∂ t = − K p 1 p 2 ∂ ∂ p ( p 3 F ) + K 1 1 p 2 ∂ ∂ p ( p 4 ∂ F ∂ p ) + K 2 1 p 2 ∂ ∂ p [ p ∂ ∂ p ( p 3 F ) ] . (10) \frac{\partial F}{\partial t}=-\mathcal{K}_p\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^3F\right)+\mathcal{K}_1\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4\frac{\partial F}{\partial p}\right)+\mathcal{K}_2\frac{1}{p^2}\frac{\partial}{\partial p}\left[p\frac{\partial}{\partial p}\left(p^3F\right)\right]. \tag*{(10)} ∂ t ∂ F = − K p p 2 1 ∂ p ∂ ( p 3 F ) + K 1 p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ F ) + K 2 p 2 1 ∂ p ∂ [ p ∂ p ∂ ( p 3 F ) ] . (10)
Here, the first term on the right-hand side represents the momentum convection, with K p \mathcal{K}_p K p as its coefficient; the second term represents the first momentum diffusion, with K 1 \mathcal{K}_1 K 1 as its coefficient; the third term represents the second momentum diffusion, with K 2 \mathcal{K}_2 K 2 as its coefficient.
In this work, the following formula for the i i i th moment of momentum is frequently used
⟨ p i ⟩ = ∫ − ∞ + ∞ p i F ( p , t ) d p x d p y d p z = 4 π ∫ 0 + ∞ p i F ( p , t ) p 2 d p , (11) \left\langle p^i\right\rangle=\int_{-\infty}^{+\infty}p^iF(p,t)\,dp_xdp_ydp_z=4\pi\int_0^{+\infty}p^iF(p,t)p^2\,dp, \tag*{(11)} ⟨ p i ⟩ = ∫ − ∞ + ∞ p i F ( p , t ) d p x d p y d p z = 4 π ∫ 0 + ∞ p i F ( p , t ) p 2 d p , (11)
where, i i i is a natural numbers. Below, we derive the moment-based transport coefficients for the special case of Equation (10) .
Moment-based formula for momentum convection coefficients
Here, we derive the transport coefficient for the following momentum convection equation
∂ F ∂ t = − K p 1 p 2 ∂ ∂ p ( p 3 F ) , (12) \frac{\partial F}{\partial t}=-\mathcal{K}_p\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^3F\right), \tag*{(12)} ∂ t ∂ F = − K p p 2 1 ∂ p ∂ ( p 3 F ) , (12)
which contains only the first term on the right-hand side of Equation (10) . Integrating Equation (12) by parts, we obtain the first-order moment equation for the momentum distribution
d d t ⟨ p ⟩ = K p ⟨ p ⟩ . (13) \frac{\mathrm{d}}{\mathrm{d}t}\langle p\rangle=\mathcal{K}_p\langle p\rangle. \tag*{(13)} d t d ⟨ p ⟩ = K p ⟨ p ⟩ . (13)
The solution of Equation (13) is
⟨ p ⟩ = ⟨ p ⟩ 0 e K p t , (14) \langle p\rangle=\langle p\rangle_0e^{\mathcal{K}_pt}, \tag*{(14)} ⟨ p ⟩ = ⟨ p ⟩ 0 e K p t , (14)
where ⟨ p ⟩ 0 \langle p\rangle_0 ⟨ p ⟩ 0 is the initial value of ⟨ p ⟩ \langle p\rangle ⟨ p ⟩ . From Equation (14) , we obtain
K p = d d t [ ln ⟨ p ⟩ − ⟨ p ⟩ 0 ] = d d t ln ⟨ p ⟩ ⟨ p ⟩ 0 = d d t ln ⟨ p ⟩ , (15) \mathcal{K}_p=\frac{\mathrm{d}}{\mathrm{d}t}\left[\ln\langle p\rangle-\langle p\rangle_0\right]=\frac{\mathrm{d}}{\mathrm{d}t}\ln\frac{\langle p\rangle}{\langle p\rangle_0}=\frac{\mathrm{d}}{\mathrm{d}t}\ln\langle p\rangle, \tag*{(15)} K p = d t d [ ln ⟨ p ⟩ − ⟨ p ⟩ 0 ] = d t d ln ⟨ p ⟩ 0 ⟨ p ⟩ = d t d ln ⟨ p ⟩ , (15)
which implies that the momentum coefficient can be expressed in terms of the logarithm of the first-order moment. In fact, from Equation (13) , we can also obtain Equation (15) . If we define the first moment as
Q 1 = ⟨ p ⟩ , (16) Q_1=\langle p\rangle, \tag*{(16)} Q 1 = ⟨ p ⟩ , (16)
the momentum convection coefficient can be rewritten as
K p = d d t ln Q 1 . (17) \mathcal{K}_p = \frac{d}{dt}\ln Q_1. \tag*{(17)} K p = d t d ln Q 1 . (17)
The latter equation shows that the momentum convection coefficient is a function of Q 1 Q_1 Q 1 . In a previous study, Wang & Qin [24 ] found that, for constant-coefficient spatial transport equations, the coefficients depend on statistical quantities. Here, we find the similar result. However, for momentum transport, the convection coefficient takes a logarithmic form.
Moment-based formula for momentum diffusion coefficients
If we neglect the convection term in Equation (10) , the momentum transport equation, which contains the first and second diffusion terms, can be written as
∂ F ∂ t = K 1 1 p 2 ∂ ∂ p ( p 4 ∂ F ∂ p ) + K 2 1 p 2 ∂ ∂ p [ p ∂ ∂ p ( p 3 F ) ] . (18) \frac{\partial F}{\partial t}=\mathcal{K}_1\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4\frac{\partial F}{\partial p}\right)+\mathcal{K}_2\frac{1}{p^2}\frac{\partial}{\partial p}\left[p\frac{\partial}{\partial p}\left(p^3F\right)\right]. \tag*{(18)} ∂ t ∂ F = K 1 p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ F ) + K 2 p 2 1 ∂ p ∂ [ p ∂ p ∂ ( p 3 F ) ] . (18)
In what follows, we investigate the relationship between momentum diffusion coefficients and statistical quantities.
The case 1.
Here, we consider the momentum diffusion equation with only the first diffusion term,
∂ F ∂ t = K 1 1 p 2 ∂ ∂ p ( p 4 ∂ F ∂ p ) , (19) \frac{\partial F}{\partial t}=\mathcal{K}_1\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4\frac{\partial F}{\partial p}\right), \tag*{(19)} ∂ t ∂ F = K 1 p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ F ) , (19)
and derive the formula for the momentum diffusion coefficient K 1 \mathcal{K}_1 K 1 . With formula (11), the first order moment can be derived from the latter equation
d d t ⟨ p ⟩ = 4 K 1 ⟨ p ⟩ , (20) \frac{d}{dt}\langle p\rangle=4\mathcal{K}_1\langle p\rangle, \tag*{(20)} d t d ⟨ p ⟩ = 4 K 1 ⟨ p ⟩ , (20)
and the corresponding momentum diffusion coefficient can be found
K 1 = d d t ln ( ⟨ p ⟩ 1 / 4 ) . (21) \mathcal{K}_1=\frac{d}{dt}\ln\left(\langle p\rangle^{1/4}\right). \tag*{(21)} K 1 = d t d ln ( ⟨ p ⟩ 1/4 ) . (21)
By setting
Q 2 = ⟨ p ⟩ 1 / 4 , (22) Q_2=\langle p\rangle^{1/4}, \tag*{(22)} Q 2 = ⟨ p ⟩ 1/4 , (22)
we can rewrite Equation (21) as
K 1 = d d t ln Q 2 . (23) \mathcal{K}_1 = \frac{\mathrm{d}}{\mathrm{d}t}\ln Q_2. \tag*{(23)} K 1 = d t d ln Q 2 . (23)
This formula indicates that the first momentum diffusion coefficient is related to the fourth root of the first moment.
The Case 2.
Retaining only the second term in Equation (18) , we obtain
∂ F ∂ t = K 2 1 p 2 ∂ ∂ p [ p ∂ ∂ p ( p 3 F ) ] . (24) \frac{\partial F}{\partial t}=\mathcal{K}_2\frac{1}{p^2}\frac{\partial}{\partial p}\left[p\frac{\partial}{\partial p}\left(p^3F\right)\right]. \tag*{(24)} ∂ t ∂ F = K 2 p 2 1 ∂ p ∂ [ p ∂ p ∂ ( p 3 F ) ] . (24)
From the latter equation, we can obtain the first order moment equation of momentum as
d d t ⟨ p ⟩ = K 2 ⟨ p ⟩ . (25) \frac{\mathrm{d}}{\mathrm{d}t}\langle p\rangle=\mathcal{K}_2\langle p\rangle. \tag*{(25)} d t d ⟨ p ⟩ = K 2 ⟨ p ⟩ . (25)
Employing this equation, we can derive the formula for the momentum diffusion coefficient
K 2 = d d t ln ⟨ p ⟩ . (26) \mathcal{K}_2=\frac{\mathrm{d}}{\mathrm{d}t}\ln\langle p\rangle. \tag*{(26)} K 2 = d t d ln ⟨ p ⟩ . (26)
With Equation (16) , the latter equation can be rewritten as
K 2 = d d t ln Q 1 . (27) \mathcal{K}_2=\frac{\mathrm{d}}{\mathrm{d}t}\ln Q_1. \tag*{(27)} K 2 = d t d ln Q 1 . (27)
This formula shows that the second momentum diffusion coefficient K 2 \mathcal{K}_2 K 2 takes the same form as the momentum convection coefficient K p \mathcal{K}_p K p . They are all determined by the first-order moment Q 1 Q_1 Q 1 .
The Case 3.
Here, we investigate the momentum diffusion coefficients for the following equation
∂ F ∂ t = K 1 1 p 2 ∂ ∂ p ( p 4 ∂ F ∂ p ) + K 2 1 p 2 ∂ ∂ p [ p ∂ ∂ p ( p 3 F ) ] , (28) \frac{\partial F}{\partial t}=\mathcal{K}_1\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4\frac{\partial F}{\partial p}\right)+\mathcal{K}_2\frac{1}{p^2}\frac{\partial}{\partial p}\left[p\frac{\partial}{\partial p}\left(p^3F\right)\right], \tag*{(28)} ∂ t ∂ F = K 1 p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ F ) + K 2 p 2 1 ∂ p ∂ [ p ∂ p ∂ ( p 3 F ) ] , (28)
which contains two different momentum diffusion terms. The first-order moment equation for the momentum distribution can be readily derived
d d t ⟨ p ⟩ = 4 K 1 ⟨ p ⟩ + K 2 ⟨ p ⟩ , (29) \frac{\mathrm{d}}{\mathrm{d}t}\langle p\rangle=4\mathcal{K}_1\langle p\rangle+\mathcal{K}_2\langle p\rangle, \tag*{(29)} d t d ⟨ p ⟩ = 4 K 1 ⟨ p ⟩ + K 2 ⟨ p ⟩ , (29)
which can be rewritten as
d d t ln ⟨ p ⟩ = 4 K 1 + K 2 . (30) \frac{\mathrm{d}}{\mathrm{d}t}\ln\langle p\rangle=4\mathcal{K}_1+\mathcal{K}_2. \tag*{(30)} d t d ln ⟨ p ⟩ = 4 K 1 + K 2 . (30)
The latter equation is linear in K 1 \mathcal{K}_1 K 1 and K 2 \mathcal{K}_2 K 2 . In order to derive the formulas for K 1 \mathcal{K}_1 K 1 and K 2 \mathcal{K}_2 K 2 , we need to obtain another equation for the momentum distribution, e.g., the second-order moment equation,
d d t ln ⟨ p 2 ⟩ = 10 K 1 + 4 K 2 . (31) \frac{\mathrm{d}}{\mathrm{d}t}\ln\langle p^2\rangle=10\mathcal{K}_1+4\mathcal{K}_2. \tag*{(31)} d t d ln ⟨ p 2 ⟩ = 10 K 1 + 4 K 2 . (31)
Combining Equations (30) and (31) yields
K 1 = d d t ln ( ⟨ p 4 ⟩ ⟨ p 2 ⟩ ) 1 / 6 , (32) \mathcal{K}_1=\frac{\mathrm{d}}{\mathrm{d}t}\ln\left(\frac{\langle p^4\rangle}{\langle p^2\rangle}\right)^{1/6}, \tag*{(32)} K 1 = d t d ln ( ⟨ p 2 ⟩ ⟨ p 4 ⟩ ) 1/6 , (32)
K 2 = d d t ln ( ⟨ p 2 ⟩ 2 ⟨ p ⟩ 5 ) 1 / 3 . (33) \mathcal{K}_2=\frac{\mathrm{d}}{\mathrm{d}t}\ln\left(\frac{\langle p^2\rangle^2}{\langle p\rangle^5}\right)^{1/3}. \tag*{(33)} K 2 = d t d ln ( ⟨ p ⟩ 5 ⟨ p 2 ⟩ 2 ) 1/3 . (33)
Defining two statistical quantities as
Q 3 = ( ⟨ p ⟩ 4 ⟨ p 2 ⟩ ) 1 / 6 , (34) Q_3=\left(\frac{\langle p\rangle^4}{\langle p^2\rangle}\right)^{1/6}, \tag*{(34)} Q 3 = ( ⟨ p 2 ⟩ ⟨ p ⟩ 4 ) 1/6 , (34)
Q 4 = ( ⟨ p 2 ⟩ 2 ⟨ p ⟩ 5 ) 1 / 3 , (35) Q_4=\left(\frac{\langle p^2\rangle^2}{\langle p\rangle^5}\right)^{1/3}, \tag*{(35)} Q 4 = ( ⟨ p ⟩ 5 ⟨ p 2 ⟩ 2 ) 1/3 , (35)
we rewrite Equations (32) and (33) as
K 1 = d d t ln Q 3 , (36) \mathcal{K}_1=\frac{\mathrm{d}}{\mathrm{d}t}\ln Q_3, \tag*{(36)} K 1 = d t d ln Q 3 , (36)
K 2 = d d t ln Q 4 . (37) \mathcal{K}_2=\frac{\mathrm{d}}{\mathrm{d}t}\ln Q_4. \tag*{(37)} K 2 = d t d ln Q 4 . (37)
Here, Q 3 Q_3 Q 3 and Q 4 Q_4 Q 4 are two new statistical quantities.
Moment-based formulas for the momentum transport coefficients
In this subsection, we explore the momentum transport equation with the convection term and one of the diffusion terms.
The case 1.
Here, we investigate the transport coefficients in the momentum equation with the convection term and the first diffusion term
∂ F ∂ t = − K p 1 p 2 ∂ ∂ p ( p 3 F ) + K 1 1 p 2 ∂ ∂ p ( p 4 ∂ F ∂ p ) . (38) \frac{\partial F}{\partial t}=-\mathcal{K}_{p}\frac{1}{p^{2}}\frac{\partial}{\partial p}\left(p^{3}F\right)+\mathcal{K}_{1}\frac{1}{p^{2}}\frac{\partial}{\partial p}\left(p^{4}\frac{\partial F}{\partial p}\right). \tag*{(38)} ∂ t ∂ F = − K p p 2 1 ∂ p ∂ ( p 3 F ) + K 1 p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ F ) . (38)
From the convection-diffusion equation for the momentum distribution, we derive the first-order moment equation
d d t ⟨ p ⟩ = K p ⟨ p ⟩ + 4 K 1 ⟨ p ⟩ , (39) \frac{\mathrm{d}}{\mathrm{d}t}\langle p\rangle=\mathcal{K}_{p}\langle p\rangle+4\mathcal{K}_{1}\langle p\rangle, \tag*{(39)} d t d ⟨ p ⟩ = K p ⟨ p ⟩ + 4 K 1 ⟨ p ⟩ , (39)
which can be rewritten as
d d t ln ⟨ p ⟩ = K p + 4 K 1 . (40) \frac{\mathrm{d}}{\mathrm{d}t}\ln\langle p\rangle=\mathcal{K}_{p}+4\mathcal{K}_{1}. \tag*{(40)} d t d ln ⟨ p ⟩ = K p + 4 K 1 . (40)
Similarly, we derive the equation for the second-order moment of the momentum distribution
d d t ln ⟨ p 2 ⟩ = 2 K p + 10 K 1 . (41) \frac{\mathrm{d}}{\mathrm{d}t}\ln\langle p^{2}\rangle=2\mathcal{K}_{p}+10\mathcal{K}_{1}. \tag*{(41)} d t d ln ⟨ p 2 ⟩ = 2 K p + 10 K 1 . (41)
Considering Equations (40) and (41) gives
K p = d d t ln ⟨ p ⟩ 5 ⟨ p 2 ⟩ 2 , (42) \mathcal{K}_{p}=\frac{\mathrm{d}}{\mathrm{d}t}\ln\frac{\langle p\rangle^{5}}{\langle p^{2}\rangle^{2}}, \tag*{(42)} K p = d t d ln ⟨ p 2 ⟩ 2 ⟨ p ⟩ 5 , (42)
K 1 = d d t ln ( ⟨ p 2 ⟩ ⟨ p ⟩ 2 ) 1 / 2 . (43) \mathcal{K}_{1}=\frac{\mathrm{d}}{\mathrm{d}t}\ln\left(\frac{\langle p^{2}\rangle}{\langle p\rangle^{2}}\right)^{1/2}. \tag*{(43)} K 1 = d t d ln ( ⟨ p ⟩ 2 ⟨ p 2 ⟩ ) 1/2 . (43)
Defining the statistical quantities
Q 5 = ⟨ p ⟩ 5 ⟨ p 2 ⟩ 2 , (44) Q_{5}=\frac{\langle p\rangle^{5}}{\langle p^{2}\rangle^{2}}, \tag*{(44)} Q 5 = ⟨ p 2 ⟩ 2 ⟨ p ⟩ 5 , (44)
Q 6 = ( ⟨ p 2 ⟩ ⟨ p ⟩ 2 ) 1 / 2 , (45) Q_{6}=\left(\frac{\langle p^{2}\rangle}{\langle p\rangle^{2}}\right)^{1/2}, \tag*{(45)} Q 6 = ( ⟨ p ⟩ 2 ⟨ p 2 ⟩ ) 1/2 , (45)
we rewrite Equations (51) and (52) as
K p = d d t ln Q 5 , (46) \mathcal{K}_{p}=\frac{\mathrm{d}}{\mathrm{d}t}\ln Q_{5}, \tag*{(46)} K p = d t d ln Q 5 , (46)
K 1 = d d t ln Q 6 . (47) \mathcal{K}_{1}=\frac{\mathrm{d}}{\mathrm{d}t}\ln Q_{6}. \tag*{(47)} K 1 = d t d ln Q 6 . (47)
The statistical quantities Q 5 Q_{5} Q 5 and Q 6 Q_{6} Q 6 are new.
The case 2.
Retaining the convection term and the second diffusion term in Equation (10) , yields the following momentum convection-diffusion equation
∂ F ∂ t = − K p 1 p 2 ∂ ∂ p ( p 3 F ) + K 2 1 p 2 ∂ ∂ p [ p ∂ ∂ p ( p 3 F ) ] . (48) \frac{\partial F}{\partial t}=-\mathcal{K}_p\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^3F\right)+\mathcal{K}_2\frac{1}{p^2}\frac{\partial}{\partial p}\left[p\frac{\partial}{\partial p}\left(p^3F\right)\right]. \tag*{(48)} ∂ t ∂ F = − K p p 2 1 ∂ p ∂ ( p 3 F ) + K 2 p 2 1 ∂ p ∂ [ p ∂ p ∂ ( p 3 F ) ] . (48)
From the latter equation, the first- and second-order moment equations for the momentum distribution are derived as
d d t ln ⟨ p ⟩ = K p + K 2 . (49) \frac{\mathrm{d}}{\mathrm{d}t}\ln\langle p\rangle=\mathcal{K}_p+\mathcal{K}_2. \tag*{(49)} d t d ln ⟨ p ⟩ = K p + K 2 . (49)
d d t ln ⟨ p 2 ⟩ = 2 K p + 4 K 2 . (50) \frac{\mathrm{d}}{\mathrm{d}t}\ln\left\langle p^2\right\rangle=2\mathcal{K}_p+4\mathcal{K}_2. \tag*{(50)} d t d ln ⟨ p 2 ⟩ = 2 K p + 4 K 2 . (50)
From the latter two equations, we derive
K p = d d t ln ( ⟨ p ⟩ 4 ⟨ p 2 ⟩ ) 1 / 2 , (51) \mathcal{K}_p=\frac{\mathrm{d}}{\mathrm{d}t}\ln\left(\frac{\langle p\rangle^4}{\left\langle p^2\right\rangle}\right)^{1/2}, \tag*{(51)} K p = d t d ln ( ⟨ p 2 ⟩ ⟨ p ⟩ 4 ) 1/2 , (51)
K 2 = d d t ln ( ⟨ p 2 ⟩ ⟨ p ⟩ 2 ) 1 / 2 . (52) \mathcal{K}_2=\frac{\mathrm{d}}{\mathrm{d}t}\ln\left(\frac{\left\langle p^2\right\rangle}{\langle p\rangle^2}\right)^{1/2}. \tag*{(52)} K 2 = d t d ln ( ⟨ p ⟩ 2 ⟨ p 2 ⟩ ) 1/2 . (52)
Defining the following statistical quantity
Q 7 = ( ⟨ p ⟩ 4 ⟨ p 2 ⟩ ) 1 / 2 (53) Q_7=\left(\frac{\langle p\rangle^4}{\left\langle p^2\right\rangle}\right)^{1/2} \tag*{(53)} Q 7 = ( ⟨ p 2 ⟩ ⟨ p ⟩ 4 ) 1/2 (53)
and using the formula for statistical quantity Q 6 Q_6 Q 6 , we rewrite Equations (51) and (52) as
K p = d d t ln Q 7 , (54) \mathcal{K}_p=\frac{\mathrm{d}}{\mathrm{d}t}\ln Q_7, \tag*{(54)} K p = d t d ln Q 7 , (54)
K 2 = d d t ln Q 6 . (55) \mathcal{K}_2=\frac{\mathrm{d}}{\mathrm{d}t}\ln Q_6. \tag*{(55)} K 2 = d t d ln Q 6 . (55)
Here, Q 7 Q_7 Q 7 is a new statistical quantity.
Moment-based formula for pitch-angle diffusion coefficient
The pitch-angle scattering of energetic particles leads to parallel spatial diffusion via isotropization. This process is described by the pitch-angle diffusion coefficient D μ μ D_{\mu\mu} D μμ , whose formula we derive in this section.
The pitch-angle scattering equation is given by
∂ F ∂ t = ∂ ∂ μ ( D μ μ ∂ F ∂ μ ) , (56) \frac{\partial F}{\partial t}=\frac{\partial}{\partial\mu}\left(D_{\mu\mu}\frac{\partial F}{\partial\mu}\right), \tag*{(56)} ∂ t ∂ F = ∂ μ ∂ ( D μμ ∂ μ ∂ F ) , (56)
where F = F ( μ , μ 0 , t ) F = F(\mu,\mu_0,t) F = F ( μ , μ 0 , t ) . For strong magnetic turbulence, the formula takes the form [16 ]
D μ μ = D ( 1 − μ 2 ) (57) D_{\mu\mu}=D(1-\mu^2) \tag*{(57)} D μμ = D ( 1 − μ 2 ) (57)
with constant D D D . Thus, the pitch-angle scattering equation can be rewritten as
∂ F ∂ t = D ∂ ∂ μ ( ( 1 − μ 2 ) ∂ F ∂ μ ) . (58) \frac{\partial F}{\partial t}=D\frac{\partial}{\partial\mu}\left((1-\mu^2)\frac{\partial F}{\partial\mu}\right). \tag*{(58)} ∂ t ∂ F = D ∂ μ ∂ ( ( 1 − μ 2 ) ∂ μ ∂ F ) . (58)
The pitch-angle correlation is given by
⟨ μ μ 0 ⟩ = ∫ − 1 + 1 d μ 0 ∫ − 1 + 1 d μ μ μ 0 F ( μ , μ 0 , t ) , (59) \langle\mu\mu_0\rangle=\int_{-1}^{+1}d\mu_0\int_{-1}^{+1}d\mu\mu\mu_0F(\mu,\mu_0,t), \tag*{(59)} ⟨ μ μ 0 ⟩ = ∫ − 1 + 1 d μ 0 ∫ − 1 + 1 d μμ μ 0 F ( μ , μ 0 , t ) , (59)
where μ \mu μ is pitch-angle, μ 0 \mu_0 μ 0 is initial pitch-angle, and t t t is the time. From Equation (58) , we find
D = − 1 2 d d t ln ⟨ μ μ 0 ⟩ . (60) D=-\frac{1}{2}\frac{d}{dt}\ln\langle\mu\mu_0\rangle. \tag*{(60)} D = − 2 1 d t d ln ⟨ μ μ 0 ⟩ . (60)
With formula (57) , the latter equation can be rewritten as
D μ μ = − 1 2 ( 1 − μ 2 ) d d t ln ⟨ μ μ 0 ⟩ , (61) D_{\mu\mu}=-\frac{1}{2}(1-\mu^2)\frac{d}{dt}\ln\langle\mu\mu_0\rangle, \tag*{(61)} D μμ = − 2 1 ( 1 − μ 2 ) d t d ln ⟨ μ μ 0 ⟩ , (61)
which is identical to the result obtained by Tautz [17 ] . Similarly, we can derive the second-order moment of the pitch-angle distribution as
D = − 1 6 d d t ln ( 1 − 3 ⟨ μ 2 ⟩ ) . (62) D=-\frac{1}{6}\frac{d}{dt}\ln\left(1-3\langle\mu^2\rangle\right). \tag*{(62)} D = − 6 1 d t d ln ( 1 − 3 ⟨ μ 2 ⟩ ) . (62)
Using Equation (57) , we obtain the following formula
D μ μ = − 1 6 ( 1 − μ 2 ) d d t ln ( 1 − 3 ⟨ μ 2 ⟩ ) . (63) D_{\mu\mu}=-\frac{1}{6}(1-\mu^2)\frac{d}{dt}\ln\left(1-3\langle\mu^2\rangle\right). \tag*{(63)} D μμ = − 6 1 ( 1 − μ 2 ) d t d ln ( 1 − 3 ⟨ μ 2 ⟩ ) . (63)
Moment-based formulas for Fokker-Planck coefficients
In this section, we investigate the coefficients of the Fokker-Planck equation, which takes the form
∂ F ∂ t = − ∂ ∂ x [ A ( x ) F ( x , t ) ] + ∂ 2 ∂ x 2 [ B ( x ) F ( x , t ) ] , (64) \frac{\partial F}{\partial t}=-\frac{\partial}{\partial x}\left[A(x)F(x,t)\right]+\frac{\partial^2}{\partial x^2}\left[B(x)F(x,t)\right], \tag*{(64)} ∂ t ∂ F = − ∂ x ∂ [ A ( x ) F ( x , t ) ] + ∂ x 2 ∂ 2 [ B ( x ) F ( x , t ) ] , (64)
where A ( x ) A(x) A ( x ) and B ( x ) B(x) B ( x ) are the first- and second-order coefficients, respectively.
Moment-based formulas for the first-order Fokker-Planck coefficient
Here, we explore the coefficient of the first-order Fokker-Planck equation
∂ F ∂ t = − ∂ ∂ x [ A ( x ) F ( x , t ) ] , (65) \frac{\partial F}{\partial t}=-\frac{\partial}{\partial x}\left[A(x)F(x,t)\right], \tag*{(65)} ∂ t ∂ F = − ∂ x ∂ [ A ( x ) F ( x , t ) ] , (65)
Using the formula for the first-order moment
⟨ x ⟩ = ∫ x F ( x , t ) d x , (66) \langle x\rangle=\int xF(x,t)\,\mathrm{d}x, \tag*{(66)} ⟨ x ⟩ = ∫ x F ( x , t ) d x , (66)
we derive from Equation (65)
d d t ⟨ x ⟩ = ⟨ A ( x ) ⟩ . (67) \frac{\mathrm{d}}{\mathrm{d}t}\langle x\rangle=\langle A(x)\rangle. \tag*{(67)} d t d ⟨ x ⟩ = ⟨ A ( x )⟩ . (67)
The latter formula indicates that the ensemble average of the variable coefficient is a function of the expectation of x x x . When A ( x ) = κ x A(x)=\kappa_x A ( x ) = κ x is constant, Equation (67) becomes
κ x = d d t ⟨ x ⟩ . (68) \kappa_x=\frac{\mathrm{d}}{\mathrm{d}t}\langle x\rangle. \tag*{(68)} κ x = d t d ⟨ x ⟩ . (68)
If the first-order coefficient is a linear function of x x x , i.e., A ( x ) = κ x x A(x)=\kappa_x x A ( x ) = κ x x with constant κ x \kappa_x κ x , we obtain
κ x = d d t ln ⟨ x ⟩ . (69) \kappa_x=\frac{\mathrm{d}}{\mathrm{d}t}\ln\langle x\rangle. \tag*{(69)} κ x = d t d ln ⟨ x ⟩ . (69)
We find that the coefficient κ x \kappa_x κ x takes a logarithmic form in this case. In fact, the first-order Fokker-Planck coefficient can take a more general form, e.g., A ( x ) = κ x x n A(x)=\kappa_x x^n A ( x ) = κ x x n where κ x \kappa_x κ x is constant. However, for this general form, we find
κ x = 1 n ⟨ x n − 1 ⟩ d d t ⟨ x n ⟩ , (70) \kappa_x=\frac{1}{n\langle x^{n-1}\rangle}\frac{\mathrm{d}}{\mathrm{d}t}\langle x^n\rangle, \tag*{(70)} κ x = n ⟨ x n − 1 ⟩ 1 d t d ⟨ x n ⟩ , (70)
which cannot be written as the time derivative of a single statistical quantity. Since the right-hand side of the following formula depends on time t t t
⟨ x n − 1 ⟩ = ∫ x n − 1 F ( x , t ) d x , (71) \langle x^{n-1}\rangle=\int x^{n-1}F(x,t)\,\mathrm{d}x, \tag*{(71)} ⟨ x n − 1 ⟩ = ∫ x n − 1 F ( x , t ) d x , (71)
the moment ⟨ x n − 1 ⟩ \langle x^{n-1}\rangle ⟨ x n − 1 ⟩ is a function of time. Substituting formula (71) into (70) , we find that the coefficient κ x \kappa_x κ x is a function of time, i.e., κ x = κ x ( t ) \kappa_x=\kappa_x(t) κ x = κ x ( t ) . However, κ x = κ x ( t ) \kappa_x=\kappa_x(t) κ x = κ x ( t ) contradicts the assumption that κ x \kappa_x κ x is constant. If we set A ( x , t ) = κ x ( t ) x A(x,t)=\kappa_x(t)x A ( x , t ) = κ x ( t ) x , we can rewrite Equation (65) as
∂ F ∂ t = − ∂ ∂ x [ κ x ( t ) x F ( x , t ) ] . (72) \frac{\partial F}{\partial t}=-\frac{\partial}{\partial x}\left[\kappa_x(t)xF(x,t)\right]. \tag*{(72)} ∂ t ∂ F = − ∂ x ∂ [ κ x ( t ) x F ( x , t ) ] . (72)
From this equation, the formula for the coefficient can be written as
κ x ( t ) = 1 n 1 ⟨ x n − 1 ⟩ d d t ⟨ x n ⟩ . (73) \kappa_x(t) = \frac{1}{n}\frac{1}{\langle x^{n-1}\rangle}\frac{\mathrm{d}}{\mathrm{d}t}\langle x^n\rangle. \tag*{(73)} κ x ( t ) = n 1 ⟨ x n − 1 ⟩ 1 d t d ⟨ x n ⟩ . (73)
We will explore the time-dependent coefficient in future work.
Moment-based formulas for the second-order Fokker-Planck coefficient
In this subsection, we investigate the coefficient of the second-order Fokker-Planck equation
∂ F ∂ t = ∂ 2 ∂ x 2 [ B ( x ) F ( x , t ) ] . (74) \frac{\partial F}{\partial t} = \frac{\partial^2}{\partial x^2}\left[B(x)F(x,t)\right]. \tag*{(74)} ∂ t ∂ F = ∂ x 2 ∂ 2 [ B ( x ) F ( x , t ) ] . (74)
Here, B ( x ) B(x) B ( x ) is the second-order Fokker-Planck coefficient. Using Equation (66) , we can obtain the equations for the first- and second-order moments as
d d t ⟨ x ⟩ = 0 , (75) \frac{\mathrm{d}}{\mathrm{d}t}\langle x\rangle= 0, \tag*{(75)} d t d ⟨ x ⟩ = 0 , (75)
d d t ⟨ x 2 ⟩ = 2 ⟨ B ( x ) ⟩ . (76) \frac{\mathrm{d}}{\mathrm{d}t}\langle x^2\rangle= 2\langle B(x)\rangle. \tag*{(76)} d t d ⟨ x 2 ⟩ = 2 ⟨ B ( x )⟩ . (76)
If the second-order Fokker-Planck coefficient is constant, i.e., with B ( x ) = κ x x B(x)=\kappa_{xx} B ( x ) = κ xx constant, we find
κ x x = 1 2 d d t ⟨ x 2 ⟩ . (77) \kappa_{xx} = \frac{1}{2}\frac{\mathrm{d}}{\mathrm{d}t}\langle x^2\rangle. \tag*{(77)} κ xx = 2 1 d t d ⟨ x 2 ⟩ . (77)
If the second-order Fokker-Planck coefficient is quadratic in x x x , i.e., B ( x ) = κ x x x 2 B(x)=\kappa_{xx}x^2 B ( x ) = κ xx x 2 with κ x x \kappa_{xx} κ xx constant, we derive
κ x x = d d t ln ( ⟨ x 2 ⟩ 1 / 2 ) . (78) \kappa_{xx} = \frac{\mathrm{d}}{\mathrm{d}t}\ln\left(\langle x^2\rangle^{1/2}\right). \tag*{(78)} κ xx = d t d ln ( ⟨ x 2 ⟩ 1/2 ) . (78)
However, when B ( x ) B(x) B ( x ) takes other forms, e.g., B ( x ) = κ x n B(x)=\kappa x^n B ( x ) = κ x n with κ \kappa κ constant and n = 3 , 4 , 5 , … n=3,4,5,\ldots n = 3 , 4 , 5 , … , we cannot obtain the above relationship.
Moment-based formulas for the Fokker-Planck coefficients
For the Fokker-Planck equation with the first- and second-order terms
∂ F ∂ t = − ∂ ∂ x [ A ( x ) F ( x , t ) ] + ∂ 2 ∂ x 2 [ B ( x ) F ( x , t ) ] , (79) \frac{\partial F}{\partial t}=-\frac{\partial}{\partial x}\left[A(x)F(x,t)\right]+\frac{\partial^2}{\partial x^2}\left[B(x)F(x,t)\right], \tag*{(79)} ∂ t ∂ F = − ∂ x ∂ [ A ( x ) F ( x , t ) ] + ∂ x 2 ∂ 2 [ B ( x ) F ( x , t ) ] , (79)
we can readily obtain the equations for the first- and second-order moments as
d d t ⟨ t ⟩ = ⟨ A ( x ) ⟩ , (80) \frac{\mathrm{d}}{\mathrm{d}t}\langle t\rangle= \langle A(x)\rangle, \tag*{(80)} d t d ⟨ t ⟩ = ⟨ A ( x )⟩ , (80)
d d t ⟨ x 2 ⟩ = 2 ⟨ x A ( x ) ⟩ + 2 ⟨ B ( x ) ⟩ . (81) \frac{\mathrm{d}}{\mathrm{d}t}\langle x^2\rangle= 2\langle xA(x)\rangle+ 2\langle B(x)\rangle. \tag*{(81)} d t d ⟨ x 2 ⟩ = 2 ⟨ x A ( x )⟩ + 2 ⟨ B ( x )⟩ . (81)
If A ( x ) = κ x = constant A(x)=\kappa_x=\text{constant} A ( x ) = κ x = constant and B ( x ) = κ x x = constant B(x)=\kappa_{xx}=\text{constant} B ( x ) = κ xx = constant , the latter formulas become
κ x = d d t ⟨ x ⟩ , (82) \kappa_x=\frac{\mathrm{d}}{\mathrm{d}t}\langle x\rangle, \tag*{(82)} κ x = d t d ⟨ x ⟩ , (82)
κ x x = 1 2 d d t [ ⟨ x 2 ⟩ − ⟨ x ⟩ 2 ] . (83) \kappa_{xx}=\frac{1}{2}\frac{\mathrm{d}}{\mathrm{d}t}\left[\langle x^2\rangle-\langle x\rangle^2\right]. \tag*{(83)} κ xx = 2 1 d t d [ ⟨ x 2 ⟩ − ⟨ x ⟩ 2 ] . (83)
However, when A ( x ) A(x) A ( x ) and B ( x ) B(x) B ( x ) take other forms, we cannot find the simple relationship between transport coefficients and statistical quantities. For example, if A ( x ) = κ x A(x)=\kappa_x A ( x ) = κ x and B ( x ) = κ x x x B(x)=\kappa_{xx}x B ( x ) = κ xx x , the coefficients take the form
κ x = d d t ⟨ x ⟩ , (84) \kappa_x=\frac{\mathrm{d}}{\mathrm{d}t}\langle x\rangle, \tag*{(84)} κ x = d t d ⟨ x ⟩ , (84)
κ x x = 1 2 ⟨ x ⟩ d d t [ ⟨ x 2 ⟩ − ⟨ x ⟩ 2 ] . (85) \kappa_{xx}=\frac{1}{2\langle x\rangle}\frac{\mathrm{d}}{\mathrm{d}t}\left[\langle x^2\rangle-\langle x\rangle^2\right]. \tag*{(85)} κ xx = 2 ⟨ x ⟩ 1 d t d [ ⟨ x 2 ⟩ − ⟨ x ⟩ 2 ] . (85)
Moment-based coefficient formulas for some Fokker-Planck-type equations
The Ornstein-Uhlenbeck processes are important in physics, finance, data science, and other fields. Its governing equation, which takes the following form, is often used in related research
∂ F ∂ t = γ ∂ ∂ x ( x F ( x , t ) ) + Q ∂ 2 F ∂ x 2 (86) \frac{\partial F}{\partial t}=\gamma\frac{\partial}{\partial x}\left(xF(x,t)\right)+Q\frac{\partial^2F}{\partial x^2} \tag*{(86)} ∂ t ∂ F = γ ∂ x ∂ ( x F ( x , t ) ) + Q ∂ x 2 ∂ 2 F (86)
Here, γ \gamma γ and Q Q Q are independent of the coordinate x x x . Equation (79) reduces to Equation (86) for A ( x ) = − γ x A(x)=-\gamma x A ( x ) = − γ x and B ( x ) = Q B(x)=Q B ( x ) = Q , with Q Q Q constant. Employing the same method, we find the coefficient formula for the Ornstein-Uhlenbeck equation
γ = − d d t ln ⟨ x ⟩ , (87) \gamma=-\frac{\mathrm{d}}{\mathrm{d}t}\ln\langle x\rangle, \tag*{(87)} γ = − d t d ln ⟨ x ⟩ , (87)
Q = 1 2 d d t ⟨ x 2 ⟩ − ⟨ x 2 ⟩ d d t ln ⟨ x ⟩ = 1 2 d d t ⟨ x 2 ⟩ + γ ⟨ x 2 ⟩ , (88) Q=\frac{1}{2}\frac{\mathrm{d}}{\mathrm{d}t}\langle x^2\rangle-\langle x^2\rangle\frac{\mathrm{d}}{\mathrm{d}t}\ln\langle x\rangle=\frac{1}{2}\frac{\mathrm{d}}{\mathrm{d}t}\langle x^2\rangle+\gamma\langle x^2\rangle, \tag*{(88)} Q = 2 1 d t d ⟨ x 2 ⟩ − ⟨ x 2 ⟩ d t d ln ⟨ x ⟩ = 2 1 d t d ⟨ x 2 ⟩ + γ ⟨ x 2 ⟩ , (88)
Equation (87) indicates that the first-order coefficient γ \gamma γ is a function of the expectation, and Equation (88) shows that the second-order coefficient Q Q Q is a function of the second-order moment and γ \gamma γ . In addition, the Shimizu-Yamada model is a well-known formulation describing many-body systems with attractive coupling forces and time delays
∂ F ∂ t = ∂ ∂ x [ γ x + κ ( x − ⟨ x ⟩ ) ] F + Q ∂ 2 F ∂ x 2 , (89) \frac{\partial F}{\partial t} = \frac{\partial}{\partial x}\left[\gamma x+\kappa(x-\langle x\rangle)\right]F + Q\frac{\partial^2 F}{\partial x^2}, \tag*{(89)} ∂ t ∂ F = ∂ x ∂ [ γ x + κ ( x − ⟨ x ⟩) ] F + Q ∂ x 2 ∂ 2 F , (89)
where γ \gamma γ , κ \kappa κ and Q Q Q are independent of x x x . Using the same method as in the previous subsection and starting from the latter model, we derive
γ = − d d t ln ⟨ x ⟩ , (90) \gamma= -\frac{\mathrm{d}}{\mathrm{d}t}\ln\langle x\rangle, \tag*{(90)} γ = − d t d ln ⟨ x ⟩ , (90)
Q = 1 2 d d t ⟨ x 2 ⟩ + 2 ⟨ x 2 ⟩ γ − [ ⟨ x 2 ⟩ − ⟨ x ⟩ 2 ] = 1 2 d d t ⟨ x 2 ⟩ + 2 ⟨ x 2 ⟩ γ − σ 2 . (91) Q = \frac{1}{2}\frac{\mathrm{d}}{\mathrm{d}t}\langle x^2\rangle+ 2\langle x^2\rangle\gamma-\left[\langle x^2\rangle-\langle x\rangle^2\right] = \frac{1}{2}\frac{\mathrm{d}}{\mathrm{d}t}\langle x^2\rangle+ 2\langle x^2\rangle\gamma-\sigma^2. \tag*{(91)} Q = 2 1 d t d ⟨ x 2 ⟩ + 2 ⟨ x 2 ⟩ γ − [ ⟨ x 2 ⟩ − ⟨ x ⟩ 2 ] = 2 1 d t d ⟨ x 2 ⟩ + 2 ⟨ x 2 ⟩ γ − σ 2 . (91)
Here, σ 2 = ⟨ x 2 ⟩ − ⟨ x ⟩ 2 \sigma^2=\langle x^2\rangle-\langle x\rangle^2 σ 2 = ⟨ x 2 ⟩ − ⟨ x ⟩ 2 . The first formula indicates that the coefficient γ \gamma γ takes the logarithmic form of the first-order moment; the second formula shows that the second coefficient Q Q Q is a function of the second-order moment and the variance.
The Desai-Zwanzig model is a well-known description for the order-disorder phase transition and takes the form
∂ F ∂ t = − ∂ ∂ x [ a x − b x 3 − κ ( x − ⟨ x ⟩ ) ] F + Q ∂ 2 F ∂ x 2 . (92) \frac{\partial F}{\partial t}=-\frac{\partial}{\partial x}\left[ax-bx^3-\kappa(x-\langle x\rangle)\right]F+Q\frac{\partial^2 F}{\partial x^2}. \tag*{(92)} ∂ t ∂ F = − ∂ x ∂ [ a x − b x 3 − κ ( x − ⟨ x ⟩) ] F + Q ∂ x 2 ∂ 2 F . (92)
Here, a a a , b b b , κ \kappa κ , and Q Q Q are coefficients. From the latter equation, we can obtain
d d t ⟨ x ⟩ = a ⟨ x ⟩ − b ⟨ x 3 ⟩ , (93) \frac{\mathrm{d}}{\mathrm{d}t}\langle x\rangle=a\langle x\rangle-b\langle x^3\rangle, \tag*{(93)} d t d ⟨ x ⟩ = a ⟨ x ⟩ − b ⟨ x 3 ⟩ , (93)
d d t ⟨ x 2 ⟩ = 2 [ a ⟨ x 2 ⟩ − b ⟨ x 4 ⟩ − κ ⟨ x 2 ⟩ + κ ⟨ x ⟩ 2 ] + 2 Q , (94) \frac{\mathrm{d}}{\mathrm{d}t}\langle x^2\rangle=2\left[a\langle x^2\rangle-b\langle x^4\rangle-\kappa\langle x^2\rangle+\kappa\langle x\rangle^2\right]+2Q, \tag*{(94)} d t d ⟨ x 2 ⟩ = 2 [ a ⟨ x 2 ⟩ − b ⟨ x 4 ⟩ − κ ⟨ x 2 ⟩ + κ ⟨ x ⟩ 2 ] + 2 Q , (94)
d d t ⟨ x 3 ⟩ = 3 [ a ⟨ x 3 ⟩ − b ⟨ x 5 ⟩ − κ ⟨ x 3 ⟩ + κ ⟨ x ⟩ ⟨ x 2 ⟩ ] + 6 Q ⟨ x ⟩ , (95) \frac{\mathrm{d}}{\mathrm{d}t}\langle x^3\rangle=3\left[a\langle x^3\rangle-b\langle x^5\rangle-\kappa\langle x^3\rangle+\kappa\langle x\rangle\langle x^2\rangle\right]+6Q\langle x\rangle, \tag*{(95)} d t d ⟨ x 3 ⟩ = 3 [ a ⟨ x 3 ⟩ − b ⟨ x 5 ⟩ − κ ⟨ x 3 ⟩ + κ ⟨ x ⟩ ⟨ x 2 ⟩ ] + 6 Q ⟨ x ⟩ , (95)
d d t ⟨ x 4 ⟩ = 4 [ a ⟨ x 4 ⟩ − b ⟨ x 6 ⟩ − κ ⟨ x 4 ⟩ + κ ⟨ x ⟩ ⟨ x 3 ⟩ ] + 12 Q ⟨ x 2 ⟩ . (96) \frac{\mathrm{d}}{\mathrm{d}t}\langle x^4\rangle=4\left[a\langle x^4\rangle-b\langle x^6\rangle-\kappa\langle x^4\rangle+\kappa\langle x\rangle\langle x^3\rangle\right]+12Q\langle x^2\rangle. \tag*{(96)} d t d ⟨ x 4 ⟩ = 4 [ a ⟨ x 4 ⟩ − b ⟨ x 6 ⟩ − κ ⟨ x 4 ⟩ + κ ⟨ x ⟩ ⟨ x 3 ⟩ ] + 12 Q ⟨ x 2 ⟩ . (96)
With Cramer’s rule, the formulas for a a a , b b b , κ \kappa κ , and Q Q Q can be derived from the linear equations. These coefficients are all functions of the moments.
The Klein-Kramers equation is another well-known model in statistical mechanics and it takes the form
∂ f ∂ t = − v ∂ f ∂ x − V ′ ( x ) m ∂ f ∂ v + γ m ∂ ∂ v ( v f ) + γ k B T m 2 ∂ 2 f ∂ v 2 . (97) \frac{\partial f}{\partial t}=-v\frac{\partial f}{\partial x}-\frac{V'(x)}{m}\frac{\partial f}{\partial v}+\frac{\gamma}{m}\frac{\partial}{\partial v}(vf)+\frac{\gamma k_{\mathrm{B}}T}{m^2}\frac{\partial^2 f}{\partial v^2}. \tag*{(97)} ∂ t ∂ f = − v ∂ x ∂ f − m V ′ ( x ) ∂ v ∂ f + m γ ∂ v ∂ ( v f ) + m 2 γ k B T ∂ v 2 ∂ 2 f . (97)
Here, x x x and v v v are variables, and m m m , γ \gamma γ , k B k_{\mathrm{B}} k B , and T T T are constants. Using the definition of the moment
⟨ x i v j ⟩ = ∬ x i v j f ( x , v , t ) d x d v , (98) \langle x^i v^j\rangle= \iint x^i v^j f(x,v,t)\,\mathrm{d}x\mathrm{d}v, \tag*{(98)} ⟨ x i v j ⟩ = ∬ x i v j f ( x , v , t ) d x d v , (98)
we obtain the following formulas from the Klein-Kramers equation,
d d t ⟨ x ⟩ = ⟨ v ⟩ , (99) \frac{\mathrm{d}}{\mathrm{d}t}\langle x\rangle= \langle v\rangle, \tag*{(99)} d t d ⟨ x ⟩ = ⟨ v ⟩ , (99)
d d t ⟨ v ⟩ = 1 m ⟨ V ′ ( x ) ⟩ − γ m ⟨ v ⟩ . (100) \frac{\mathrm{d}}{\mathrm{d}t}\langle v\rangle= \frac{1}{m}\langle V'(x)\rangle- \frac{\gamma}{m}\langle v\rangle. \tag*{(100)} d t d ⟨ v ⟩ = m 1 ⟨ V ′ ( x )⟩ − m γ ⟨ v ⟩ . (100)
From the latter equations, we obtain
⟨ V ′ ( x ) ⟩ = m d d t ⟨ v ⟩ + γ d d t ⟨ x ⟩ = d d t [ m ⟨ v ⟩ + γ ⟨ x ⟩ ] . (101) \langle V'(x)\rangle= m\frac{\mathrm{d}}{\mathrm{d}t}\langle v\rangle+ \gamma\frac{\mathrm{d}}{\mathrm{d}t}\langle x\rangle= \frac{\mathrm{d}}{\mathrm{d}t}\left[m\langle v\rangle+\gamma\langle x\rangle\right]. \tag*{(101)} ⟨ V ′ ( x )⟩ = m d t d ⟨ v ⟩ + γ d t d ⟨ x ⟩ = d t d [ m ⟨ v ⟩ + γ ⟨ x ⟩ ] . (101)
If we set
Q 8 = m ⟨ v ⟩ + γ ⟨ x ⟩ , (102) Q_8 = m\langle v\rangle+\gamma\langle x\rangle, \tag*{(102)} Q 8 = m ⟨ v ⟩ + γ ⟨ x ⟩ , (102)
the coefficient can be rewritten as
⟨ V ′ ( x ) ⟩ = d d t Q 8 . (103) \langle V'(x)\rangle= \frac{\mathrm{d}}{\mathrm{d}t}Q_8. \tag*{(103)} ⟨ V ′ ( x )⟩ = d t d Q 8 . (103)
In addition, if V ′ ( x ) V'(x) V ′ ( x ) is known, we find
γ = ⟨ V ′ ( x ) ⟩ − m d d t ⟨ v ⟩ d d t ⟨ x ⟩ . (104) \gamma= \frac{\langle V'(x)\rangle-m\frac{\mathrm{d}}{\mathrm{d}t}\langle v\rangle}{\frac{\mathrm{d}}{\mathrm{d}t}\langle x\rangle}. \tag*{(104)} γ = d t d ⟨ x ⟩ ⟨ V ′ ( x )⟩ − m d t d ⟨ v ⟩ . (104)
Moment-based coefficient formula for fractional transport equation
The fractional transport equations, which describe the anomalous diffusion of energetic charged particles in turbulent magnetic fields, have been extensively explored over the past decades (Mazur et al. 2000; Ruffolo et al. 2003; del-Castillo-Negrete et al 2004 ; Trenchi et al. 2013; de Oliveira & Machado 2014; Perri et al. 2015; Zimbardo et al. 2021). Starting from the Continuous Time Random Walk method (Montroll & Weis 1965), del-Castillo-Negrete et al (2004 ) derived the well-known fractional equation
∂ α F ∂ t α = χ ∂ β F ∂ z β . (105) \frac{\partial^\alpha F}{\partial t^\alpha}=\chi\frac{\partial^\beta F}{\partial z^\beta}. \tag*{(105)} ∂ t α ∂ α F = χ ∂ z β ∂ β F . (105)
Here, 0 < α < 1 0 < \alpha< 1 0 < α < 1 and 0 < β < 2 0 < \beta< 2 0 < β < 2 , F = F ( z , t ) F = F(z,t) F = F ( z , t ) is the distribution function of energetic particles, and χ \chi χ is the transport coefficient. In addition, the time derivative used in the latter equation is the Caputo derivative
∂ α F ∂ t α = 1 Γ ( m − α ) ∫ 0 t d τ ( t − τ ) m − α − 1 d m d τ m F , (106) \frac{\partial^\alpha F}{\partial t^\alpha} = \frac{1}{\Gamma(m-\alpha)} \int_0^t d\tau\,(t-\tau)^{m-\alpha-1}\frac{d^m}{d\tau^m}F, \tag*{(106)} ∂ t α ∂ α F = Γ ( m − α ) 1 ∫ 0 t d τ ( t − τ ) m − α − 1 d τ m d m F , (106)
where m m m is the ceiling of α \alpha α . The space derivative is defined according to the Riesz definition
∂ β F ∂ z β = − 1 2 cos π β 2 1 Γ ( n − β ) d n d z n ( ∫ − ∞ z ( z − ξ ) n − β − 1 F ( ξ ) d ξ + ( − 1 ) n ∫ z + ∞ ( ξ − z ) n − β − 1 F ( ξ ) d ξ ) (107) \frac{\partial^\beta F}{\partial z^\beta} = -\frac{1}{2\cos\frac{\pi\beta}{2}}\frac{1}{\Gamma(n-\beta)}\frac{d^n}{dz^n}\left(\int_{-\infty}^{z}(z-\xi)^{n-\beta-1}F(\xi)d\xi+(-1)^n\int_z^{+\infty}(\xi-z)^{n-\beta-1}F(\xi)d\xi\right) \tag*{(107)} ∂ z β ∂ β F = − 2 cos 2 π β 1 Γ ( n − β ) 1 d z n d n ( ∫ − ∞ z ( z − ξ ) n − β − 1 F ( ξ ) d ξ + ( − 1 ) n ∫ z + ∞ ( ξ − z ) n − β − 1 F ( ξ ) d ξ ) (107)
Here, the left-sided and right-sided Weyl derivative are used. In this section, we investigate the transport coefficient χ \chi χ and derive its formula expressed in terms of the moments of the distribution function F ( z , t ) F(z,t) F ( z , t ) .
Applying the Caputo derivative operator to the α \alpha α -order moment of the distribution function F ( z , t ) F(z,t) F ( z , t ) yields
∂ α ∂ t α ⟨ ∣ z ∣ α ⟩ = χ ∫ − ∞ + ∞ d z ∣ z ∣ α 1 Γ ( m − α ) ∫ 0 t d τ ( t − τ ) m − α − 1 d m d τ m F ( z , τ ) . (108) \frac{\partial^\alpha}{\partial t^\alpha}\langle|z|^\alpha\rangle= \chi\int_{-\infty}^{+\infty} dz|z|^\alpha\frac{1}{\Gamma(m-\alpha)}\int_0^t d\tau\,(t-\tau)^{m-\alpha-1}\frac{d^m}{d\tau^m}F(z,\tau). \tag*{(108)} ∂ t α ∂ α ⟨ ∣ z ∣ α ⟩ = χ ∫ − ∞ + ∞ d z ∣ z ∣ α Γ ( m − α ) 1 ∫ 0 t d τ ( t − τ ) m − α − 1 d τ m d m F ( z , τ ) . (108)
where m m m is the ceiling of α \alpha α . Similarly, we have
∂ β ∂ z β ⟨ ∣ z ∣ α ⟩ = − ∫ − ∞ + ∞ d z ∣ z ∣ α 1 2 cos π β 2 1 Γ ( n − β ) d n d z n ( ∫ − ∞ z ( z − ξ ) n − β − 1 F ( ξ ) d ξ + ( − 1 ) n ∫ z + ∞ ( ξ − z ) n − β − 1 F ( ξ ) d ξ ) . (109) \frac{\partial^\beta}{\partial z^\beta}\langle|z|^\alpha\rangle= -\int_{-\infty}^{+\infty} dz|z|^\alpha\frac{1}{2\cos\frac{\pi\beta}{2}}\frac{1}{\Gamma(n-\beta)}\frac{d^n}{dz^n}\left(\int_{-\infty}^{z}(z-\xi)^{n-\beta-1}F(\xi)d\xi+(-1)^n\int_z^{+\infty}(\xi-z)^{n-\beta-1}F(\xi)d\xi\right). \tag*{(109)} ∂ z β ∂ β ⟨ ∣ z ∣ α ⟩ = − ∫ − ∞ + ∞ d z ∣ z ∣ α 2 cos 2 π β 1 Γ ( n − β ) 1 d z n d n ( ∫ − ∞ z ( z − ξ ) n − β − 1 F ( ξ ) d ξ + ( − 1 ) n ∫ z + ∞ ( ξ − z ) n − β − 1 F ( ξ ) d ξ ) . (109)
In the following subsection, we derive the formula for the coefficient χ \chi χ .
The fractional transport equation
∂ F ∂ t = χ ∂ β F ∂ z β (110) \frac{\partial F}{\partial t}=\chi\frac{\partial^\beta F}{\partial z^\beta} \tag*{(110)} ∂ t ∂ F = χ ∂ z β ∂ β F (110)
describes the evolution of the distribution function of energetic particles propagating in magnetic turbulence. The distribution function of the fractional transport equation follows the Lévy distribution. In this subsection, we derive the formula for the coefficient χ \chi χ in Equation (110) .
Combining Equations (109) and (110) , we find
∫ − ∞ + ∞ d z ∣ z ∣ α ∂ F ∂ t = d d t ⟨ ∣ z ∣ α ⟩ = χ ∫ − ∞ + ∞ d z ∣ z ∣ α ∂ β F ∂ z β = χ ∫ − ∞ + ∞ d z ∣ z ∣ α [ − 1 2 cos π β 2 1 Γ ( n − β ) d n d z n ( ∫ − ∞ z ( z − ξ ) n − β − 1 F ( ξ ) d ξ + ( − 1 ) n ∫ z + ∞ ( ξ − z ) n − β − 1 F ( ξ ) d ξ ) ] = ( − 1 ) n + 1 χ 2 cos π β 2 Γ ( α + 1 ) Γ ( n − β ) Γ ( α − n + 1 ) [ I 1 + ( − 1 ) n I 2 ] (111) \begin{aligned}
\int_{-\infty}^{+\infty} dz|z|^\alpha\frac{\partial F}{\partial t}=\frac{d}{dt}\langle|z|^\alpha\rangle=\chi\int_{-\infty}^{+\infty} dz|z|^\alpha\frac{\partial^\beta F}{\partial z^\beta} \\
&=\chi\int_{-\infty}^{+\infty} dz|z|^\alpha\left[-\frac{1}{2\cos\frac{\pi\beta}{2}}\frac{1}{\Gamma(n-\beta)}\frac{d^n}{dz^n}\left(\int_{-\infty}^{z}(z-\xi)^{n-\beta-1}F(\xi)d\xi+(-1)^n\int_z^{+\infty}(\xi-z)^{n-\beta-1}F(\xi)d\xi\right)\right] \\
&=(-1)^{n+1}\frac{\chi}{2\cos\frac{\pi\beta}{2}}\frac{\Gamma(\alpha+1)}{\Gamma(n-\beta)\Gamma(\alpha-n+1)}[I_1+(-1)^nI_2] \tag*{(111)}
\end{aligned} ∫ − ∞ + ∞ d z ∣ z ∣ α ∂ t ∂ F = d t d ⟨ ∣ z ∣ α ⟩ = χ ∫ − ∞ + ∞ d z ∣ z ∣ α ∂ z β ∂ β F = χ ∫ − ∞ + ∞ d z ∣ z ∣ α [ − 2 cos 2 π β 1 Γ ( n − β ) 1 d z n d n ( ∫ − ∞ z ( z − ξ ) n − β − 1 F ( ξ ) d ξ + ( − 1 ) n ∫ z + ∞ ( ξ − z ) n − β − 1 F ( ξ ) d ξ ) ] = ( − 1 ) n + 1 2 cos 2 π β χ Γ ( n − β ) Γ ( α − n + 1 ) Γ ( α + 1 ) [ I 1 + ( − 1 ) n I 2 ] (111)
with
I 1 = ∫ − ∞ + ∞ d z ∣ z ∣ α − n ∫ − ∞ z d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (112) I_1=\int_{-\infty}^{+\infty} dz|z|^{\alpha-n}\int_{-\infty}^{z}d\xi\,(z-\xi)^{n-\beta-1}F(\xi), \tag*{(112)} I 1 = ∫ − ∞ + ∞ d z ∣ z ∣ α − n ∫ − ∞ z d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (112)
I 2 = ∫ − ∞ + ∞ d z ∣ z ∣ α − n ∫ z + ∞ d ξ ( ξ − z ) n − β − 1 F ( ξ ) . (113) I_2=\int_{-\infty}^{+\infty} dz|z|^{\alpha-n}\int_{z}^{+\infty}d\xi\,(\xi-z)^{n-\beta-1}F(\xi). \tag*{(113)} I 2 = ∫ − ∞ + ∞ d z ∣ z ∣ α − n ∫ z + ∞ d ξ ( ξ − z ) n − β − 1 F ( ξ ) . (113)
The derivation of the formulas I 1 I_1 I 1 and I 2 I_2 I 2 can be found in Appendix § 9 , and with the results obtained there, the transport coefficient χ \chi χ can be rewritten as
χ = 2 cos π β 2 Γ ( α − n + 1 ) Γ ( α + 1 ) 1 Γ ( β − α ) Γ ( n − α ) cos 2 α ( π + 2 k π ) 2 − Γ ( α − n + 1 ) Γ ( α − β + 1 ) cos 2 β ( π + 2 k π ) 2 d d t ⟨ ∣ z ∣ α ⟩ 1 ⟨ ∣ z ∣ α − β ⟩ . (114) \chi=2\cos\frac{\pi\beta}{2}\frac{\Gamma(\alpha-n+1)}{\Gamma(\alpha+1)}
\frac{1}{\frac{\Gamma(\beta-\alpha)}{\Gamma(n-\alpha)}\cos^2\frac{\alpha(\pi+2k\pi)}{2}-\frac{\Gamma(\alpha-n+1)}{\Gamma(\alpha-\beta+1)}\cos^2\frac{\beta(\pi+2k\pi)}{2}}
\frac{\mathrm{d}}{\mathrm{d}t}\langle|z|^\alpha\rangle\frac{1}{\langle|z|^{\alpha-\beta}\rangle}. \tag*{(114)} χ = 2 cos 2 π β Γ ( α + 1 ) Γ ( α − n + 1 ) Γ ( n − α ) Γ ( β − α ) cos 2 2 α ( π + 2 k π ) − Γ ( α − β + 1 ) Γ ( α − n + 1 ) cos 2 2 β ( π + 2 k π ) 1 d t d ⟨ ∣ z ∣ α ⟩ ⟨ ∣ z ∣ α − β ⟩ 1 . (114)
The latter formula shows that the coefficient χ \chi χ is a function of ⟨ ∣ z ∣ α ⟩ \langle|z|^\alpha\rangle ⟨ ∣ z ∣ α ⟩ and ⟨ ∣ z ∣ α − β ⟩ \langle|z|^{\alpha-\beta}\rangle ⟨ ∣ z ∣ α − β ⟩ .
Moment-based coefficient formula for fraction equation ∂ α F / ∂ t α = χ ∂ 2 F / ∂ z 2 \partial^\alpha F/\partial t^\alpha=\chi\partial^2F/\partial z^2 ∂ α F / ∂ t α = χ ∂ 2 F / ∂ z 2
For β = 2 \beta=2 β = 2 , Equation (105) becomes
∂ α F ∂ t α = χ ∂ 2 F ∂ z 2 . (115) \frac{\partial^\alpha F}{\partial t^\alpha}=\chi\frac{\partial^2F}{\partial z^2}. \tag*{(115)} ∂ t α ∂ α F = χ ∂ z 2 ∂ 2 F . (115)
In this subsection, we derive the transport coefficient χ \chi χ in the latter equation. Using the same procedure as in the preceding subsections yields the following formula
d α d t α ⟨ z 2 ⟩ = 2 κ α ∫ − ∞ + ∞ d z F ( z ) . (116) \frac{\mathrm{d}^\alpha}{\mathrm{d}t^\alpha}\langle z^2\rangle=2\kappa_\alpha\int_{-\infty}^{+\infty}dzF(z). \tag*{(116)} d t α d α ⟨ z 2 ⟩ = 2 κ α ∫ − ∞ + ∞ d z F ( z ) . (116)
Given the normalization condition and assuming the distribution function is even, we have
∫ − ∞ + ∞ d z F ( z , t ) = 2 ∫ 0 + ∞ d z F ( z , t ) = 1 , (117) \int_{-\infty}^{+\infty}dzF(z,t)=2\int_0^{+\infty}dzF(z,t)=1, \tag*{(117)} ∫ − ∞ + ∞ d z F ( z , t ) = 2 ∫ 0 + ∞ d z F ( z , t ) = 1 , (117)
From Equations (116) and (117) , we find
χ = 1 2 d α d t α ⟨ z 2 ⟩ , (118) \chi=\frac{1}{2}\frac{\mathrm{d}^\alpha}{\mathrm{d}t^\alpha}\langle z^2\rangle, \tag*{(118)} χ = 2 1 d t α d α ⟨ z 2 ⟩ , (118)
which indicates that the transport coefficient χ \chi χ is a function of the second-order moment. As done in Section (5.1), the power-law dependence of the second-order moment of z z z can be obtained
⟨ z 2 ⟩ ∼ t α . (119) \langle z^2\rangle\sim t^\alpha. \tag*{(119)} ⟨ z 2 ⟩ ∼ t α . (119)
If α = 1 \alpha=1 α = 1 , the latter formula describes a normal Markovian process; for 1 < α < 2 1<\alpha<2 1 < α < 2 and 0 < α < 1 0<\alpha<1 0 < α < 1 , it corresponds to the subdiffusion and superdiffusion, respectively.
In this subsection, we investigate the transport coefficients of the following fractional equation
∂ F ∂ t = χ β ∂ β F ∂ z β + χ α ∂ α F ∂ z α , (120) \frac{\partial F}{\partial t} = \chi_{\beta}\frac{\partial^{\beta}F}{\partial z^{\beta}} + \chi_{\alpha}\frac{\partial^{\alpha}F}{\partial z^{\alpha}}, \tag*{(120)} ∂ t ∂ F = χ β ∂ z β ∂ β F + χ α ∂ z α ∂ α F , (120)
where 0 < α , β < 2 0 < \alpha,\beta< 2 0 < α , β < 2 . Applying the Fourier transform
F ^ ( k ) = ∫ − ∞ + ∞ F ( z ) e − i k z d z , (121) \hat{F}(k) = \int_{-\infty}^{+\infty} F(z)e^{-ikz}\,\mathrm{d}z, \tag*{(121)} F ^ ( k ) = ∫ − ∞ + ∞ F ( z ) e − ik z d z , (121)
we find that Equation (120) becomes
∂ F ^ ∂ t = − κ β ∣ k ∣ β F ^ − κ α ∣ k ∣ α F ^ . (122) \frac{\partial\hat{F}}{\partial t} = -\kappa_{\beta}|k|^{\beta}\hat{F} - \kappa_{\alpha}|k|^{\alpha}\hat{F}. \tag*{(122)} ∂ t ∂ F ^ = − κ β ∣ k ∣ β F ^ − κ α ∣ k ∣ α F ^ . (122)
By defining the following formulas for the moments in Fourier space
⟨ E ⟩ = ∫ − ∞ + ∞ d k F ^ ( k ) , (123) \langle E\rangle= \int_{-\infty}^{+\infty} \mathrm{d}k\hat{F}(k), \tag*{(123)} ⟨ E ⟩ = ∫ − ∞ + ∞ d k F ^ ( k ) , (123)
⟨ ∣ k ∣ α ⟩ = ∫ − ∞ + ∞ d k F ^ ( k ) ∣ k ∣ α , (124) \langle|k|^{\alpha}\rangle= \int_{-\infty}^{+\infty} \mathrm{d}k\hat{F}(k)|k|^{\alpha}, \tag*{(124)} ⟨ ∣ k ∣ α ⟩ = ∫ − ∞ + ∞ d k F ^ ( k ) ∣ k ∣ α , (124)
we obtain the zeroth- and β \beta β -th order moment equation as
d d t ⟨ E ⟩ = − κ β ⟨ ∣ k ∣ β ⟩ − κ α ⟨ ∣ k ∣ α ⟩ , (125) \frac{\mathrm{d}}{\mathrm{d}t}\langle E\rangle= -\kappa_{\beta}\langle|k|^{\beta}\rangle- \kappa_{\alpha}\langle|k|^{\alpha}\rangle, \tag*{(125)} d t d ⟨ E ⟩ = − κ β ⟨ ∣ k ∣ β ⟩ − κ α ⟨ ∣ k ∣ α ⟩ , (125)
d d t ⟨ ∣ k ∣ β ⟩ = − κ β ⟨ ∣ k ∣ 2 β ⟩ − κ α ⟨ ∣ k ∣ β + α ⟩ . (126) \frac{\mathrm{d}}{\mathrm{d}t}\langle|k|^{\beta}\rangle= -\kappa_{\beta}\langle|k|^{2\beta}\rangle- \kappa_{\alpha}\langle|k|^{\beta+\alpha}\rangle. \tag*{(126)} d t d ⟨ ∣ k ∣ β ⟩ = − κ β ⟨ ∣ k ∣ 2 β ⟩ − κ α ⟨ ∣ k ∣ β + α ⟩ . (126)
Solving the two linear equations gives
κ β = ⟨ ∣ k ∣ α ⟩ d d t ⟨ ∣ k ∣ β ⟩ − ⟨ ∣ k ∣ β + α ⟩ d d t ⟨ E ⟩ ⟨ ∣ k ∣ β ⟩ ⟨ ∣ k ∣ β + α ⟩ − ⟨ ∣ k ∣ α ⟩ ⟨ ∣ k ∣ 2 β ⟩ , (127) \kappa_{\beta} = \frac{\langle|k|^{\alpha}\rangle\frac{\mathrm{d}}{\mathrm{d}t}\langle|k|^{\beta}\rangle- \langle|k|^{\beta+\alpha}\rangle\frac{\mathrm{d}}{\mathrm{d}t}\langle E\rangle}{\langle|k|^{\beta}\rangle\langle|k|^{\beta+\alpha}\rangle- \langle|k|^{\alpha}\rangle\langle|k|^{2\beta}\rangle}, \tag*{(127)} κ β = ⟨ ∣ k ∣ β ⟩ ⟨ ∣ k ∣ β + α ⟩ − ⟨ ∣ k ∣ α ⟩ ⟨ ∣ k ∣ 2 β ⟩ ⟨ ∣ k ∣ α ⟩ d t d ⟨ ∣ k ∣ β ⟩ − ⟨ ∣ k ∣ β + α ⟩ d t d ⟨ E ⟩ , (127)
κ α = ⟨ ∣ k ∣ 2 β ⟩ d d t ⟨ E ⟩ − ⟨ ∣ k ∣ β ⟩ d d t ⟨ ∣ k ∣ β ⟩ ⟨ ∣ k ∣ β ⟩ ⟨ ∣ k ∣ β + α ⟩ − ⟨ ∣ k ∣ α ⟩ ⟨ ∣ k ∣ 2 β ⟩ . (128) \kappa_{\alpha} = \frac{\langle|k|^{2\beta}\rangle\frac{\mathrm{d}}{\mathrm{d}t}\langle E\rangle- \langle|k|^{\beta}\rangle\frac{\mathrm{d}}{\mathrm{d}t}\langle|k|^{\beta}\rangle}{\langle|k|^{\beta}\rangle\langle|k|^{\beta+\alpha}\rangle- \langle|k|^{\alpha}\rangle\langle|k|^{2\beta}\rangle}. \tag*{(128)} κ α = ⟨ ∣ k ∣ β ⟩ ⟨ ∣ k ∣ β + α ⟩ − ⟨ ∣ k ∣ α ⟩ ⟨ ∣ k ∣ 2 β ⟩ ⟨ ∣ k ∣ 2 β ⟩ d t d ⟨ E ⟩ − ⟨ ∣ k ∣ β ⟩ d t d ⟨ ∣ k ∣ β ⟩ . (128)
Since the moments ⟨ ∣ k ∣ α ⟩ \langle|k|^{\alpha}\rangle ⟨ ∣ k ∣ α ⟩ , ⟨ ∣ k ∣ β ⟩ \langle|k|^{\beta}\rangle ⟨ ∣ k ∣ β ⟩ , and other moments are all functions of time, the coefficients are also functions of time. Similarly, after applying the Fourier transform to the fractional equation,
∂ F ∂ t = χ ∂ α F ∂ z α , (129) \frac{\partial F}{\partial t} = \chi\frac{\partial^{\alpha}F}{\partial z^{\alpha}}, \tag*{(129)} ∂ t ∂ F = χ ∂ z α ∂ α F , (129)
we derive
∂ F ^ ∂ t = − χ ∣ k ∣ α F ^ . (130) \frac{\partial\widehat{F}}{\partial t}=-\chi|k|^\alpha\widehat{F}. \tag*{(130)} ∂ t ∂ F = − χ ∣ k ∣ α F . (130)
The corresponding zeroth moment equation can be found
d d t ⟨ E ⟩ = − χ ⟨ ∣ k ∣ α ⟩ . (131) \frac{\mathrm{d}}{\mathrm{d}t}\langle E\rangle=-\chi\langle|k|^\alpha\rangle. \tag*{(131)} d t d ⟨ E ⟩ = − χ ⟨ ∣ k ∣ α ⟩ . (131)
Thus, the formula for the transport coefficient can be obtained
κ = − 1 ⟨ ∣ k ∣ α ⟩ d d t ⟨ E ⟩ . (132) \kappa=-\frac{1}{\langle|k|^\alpha\rangle}\frac{\mathrm{d}}{\mathrm{d}t}\langle E\rangle. \tag*{(132)} κ = − ⟨ ∣ k ∣ α ⟩ 1 d t d ⟨ E ⟩ . (132)
Because ⟨ ∣ k ∣ α ⟩ \langle|k|^\alpha\rangle ⟨ ∣ k ∣ α ⟩ is a function of time, the coefficient κ \kappa κ is dependent on time.
SUMMARY AND CONCLUSION
Transport coefficients are important physical quantities describing stochastic processes. In previous studies, the spatial transport coefficients (including parallel and perpendicular) have been assumed constant and can be expressed in terms of various statistical quantities, such as the mean and variance[24 ] . However, the momentum transport coefficients have not been explored. In this paper, we investigate whether the momentum transport formulas can be expresses in terms of moments. In addition, pitch-angle scattering coefficient depend on the variables. Whether a relationship exists between variable transport coefficients and statistical quantities remain unclear. In this paper, we explore this issue.
Employing the method from our previous studies, we derive the coefficients for momentum, pitch-angle, and spatial transport in terms of the moments. Starting with the momentum transport equations with variable coefficients, i.e., Equation (5) , we show that the coefficients are the logarithmic functions of statistical quantities. For the momentum equation containing only one term, e.g., the convection term or the diffusion term, the transport coefficients K p \mathcal{K}_p K p , K 1 \mathcal{K}_1 K 1 and K 2 \mathcal{K}_2 K 2 are all logarithmic functions of the mathematical expectation. However, if the momentum transport equations contain two terms, either the first and second diffusion terms, or the convection term and a diffusion term, the transport coefficients are all logarithmic functions of some new statistical quantities. The results show that the momentum transport coefficients take logarithmic forms, which are different from those of the constant spatial transport coefficients. Similarly, we find that the pitch-angle diffusion coefficient also takes the logarithmic form, which is identical to the result obtained by Tautz [17 ] . In addition, for the first-order Fokker-Planck equation with the coefficient A ( x ) = κ x x A(x)=\kappa_x x A ( x ) = κ x x and the second-order Fokker-Planck equation with the coefficient B ( x ) = κ x x x 2 B(x)=\kappa_{xx}x^2 B ( x ) = κ xx x 2 , the coefficients κ x \kappa_x κ x and κ x x \kappa_{xx} κ xx also take a logarithmic form.
There are many Fokker-Planck-type transport models in space plasma physics, statistical physics, finance, and other fields. The Ornstein-Uhlenbeck equation, where the first coefficient is not constant and the second coefficient is independent of x x x , is well-known in finance and data science. Using the same procedure, we find that the first coefficient γ \gamma γ is a logarithmic function of the expectation, and the second coefficient Q Q Q is a function of the second-order moment and the variance. The Shimizu-Yamada equation, the Desai-Zwanzig equation and the Klein-Kramers equation are all well-known in statistical mechanics, and their coefficients are all functions of the moments.
Anomalous diffusion processes, including superdiffusion and subdiffusion, have been extensively investigated in recent decades. The fractional transport equation with the Caputo time derivative and the Riesz space derivative, derived by del-Castillo-Negrete et al (2004), is one of the best-known models. In this paper, we find that the transport coefficients of the fractional equation can be expressed in terms of the moments.
The coefficient formulas derived in this paper are the simple theoretical results. Because we assume an isotropic form D μ μ = D ( 1 − μ 2 ) D_{\mu\mu}=D(1-\mu^2) D μμ = D ( 1 − μ 2 ) with a constant D D D , the results obtained in this paper are only applicable to isotropic pitch-angle scattering and to pitch-angle diffusion that is almost independent of particle speed. In addition, because the mixed-derivative terms with D μ p D_{\mu p} D μ p or D p μ D_{p\mu} D p μ are neglected, the formulas obtained in this paper are only applicable to cases in which the coupling effects between pitch-angle and particle momentum are very weak relative to the other physical effects. Additionally, for simplicity, we assume that D p p = N p 2 D μ μ D_{pp}=Np^2D_{\mu\mu} D pp = N p 2 D μμ with N N N constant. Actually, the parameter N N N is a function of particle speed and other variables. Thus, the results obtained in this paper are only applicable to the cases in which D p p D_{pp} D pp is approximately quadratic in particle momentum. This paper is only the starting point for our research on this topic. As the next step, we will explore cases that include the terms with D μ p D_{\mu p} D μ p or D p μ D_{p\mu} D p μ , and cases in which D μ μ D_{\mu\mu} D μμ depends on particle momentum. In addition, we will explore the formulas for the momentum transport coefficients when D p p D_{pp} D pp , D p D_p D p and D μ μ D_{\mu\mu} D μμ depend on other variables or take other forms.
Acknowledgments
This work was supported by the grants NNSFC 42374189 and NNSFC 42374190. This work was supported by the Shenzhen Science and Technology Program under Grant No. JCYJ20250604145503005 and No. JCYJ20210324132812029. The authors would like to acknowledge the support of National Key Research and Development Program of China No.2021YFA0718600 and No.2022YFA1604600. This work was supported by Shenzhen Key Laboratory Launching Project (No. ZDSYS20210702140800001). The authors thank the anonymous referees for their valuable comments.
Appendix The derivation of the momentum transport equation
The Fokker–Planck equation with momentum transport terms is is given by Schlickeiser [9 ]
∂ f ∂ t = − v μ ∂ f ∂ z + ∂ ∂ μ ( D μ μ ∂ f ∂ μ ) − 1 p 2 ∂ ∂ p ( p 2 D p f ) + 1 p 2 ∂ ∂ p ( p 2 D p p ∂ f ∂ p ) (133) \frac{\partial f}{\partial t}=-v\mu\frac{\partial f}{\partial z}+\frac{\partial}{\partial\mu}\left(D_{\mu\mu}\frac{\partial f}{\partial\mu}\right)-\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^2D_p f\right)+\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^2D_{pp}\frac{\partial f}{\partial p}\right) \tag*{(133)} ∂ t ∂ f = − v μ ∂ z ∂ f + ∂ μ ∂ ( D μμ ∂ μ ∂ f ) − p 2 1 ∂ p ∂ ( p 2 D p f ) + p 2 1 ∂ p ∂ ( p 2 D pp ∂ p ∂ f ) (133)
with
D p = p A ( μ ) , (134) D_p=pA(\mu), \tag*{(134)} D p = p A ( μ ) , (134)
D p p = N p 2 D μ μ . (135) D_{pp}=Np^2D_{\mu\mu}. \tag*{(135)} D pp = N p 2 D μμ . (135)
Here, N N N is constant. In fact, the coefficients can take more general forms, e.g.,
D p = p n A ( μ ) , (136) D_p=p^nA(\mu), \tag*{(136)} D p = p n A ( μ ) , (136)
D p p = N p m D μ μ , (137) D_{pp}=Np^mD_{\mu\mu}, \tag*{(137)} D pp = N p m D μμ , (137)
and these will be explored in future work.
Decomposing the distribution function f = f ( z , p , μ , t ) f=f(z,p,\mu,t) f = f ( z , p , μ , t ) into the isotropic part F = F ( z , p , t ) F=F(z,p,t) F = F ( z , p , t ) and the anisotropic part g = g ( z , p , μ , t ) g=g(z,p,\mu,t) g = g ( z , p , μ , t ) yields
f = F + g . (138) f=F+g. \tag*{(138)} f = F + g . (138)
Here, the following formulas hold
∫ − 1 + 1 g d μ = 0 , (139) \int_{-1}^{+1} g\,\mathrm{d}\mu= 0, \tag*{(139)} ∫ − 1 + 1 g d μ = 0 , (139)
∫ − 1 + 1 f d μ = F . (140) \int_{-1}^{+1} f\,\mathrm{d}\mu= F. \tag*{(140)} ∫ − 1 + 1 f d μ = F . (140)
Inserting Equation (A-6) into Equation (A-1) gives
∂ F ∂ t + ∂ g ∂ t = − v μ ∂ F ∂ z − v μ ∂ g ∂ z + ∂ ∂ μ ( D μ μ ∂ g ∂ μ ) − 1 p 2 ∂ ∂ p ( p 3 A ( μ ) F ) − 1 p 2 ∂ ∂ p ( p 3 A ( μ ) g ) + N 1 p 2 ∂ ∂ p ( p 4 D μ μ ∂ F ∂ p ) + N 1 p 2 ∂ ∂ p ( p 4 D μ μ ∂ g ∂ p ) (141) \frac{\partial F}{\partial t}+\frac{\partial g}{\partial t}
=-v\mu\frac{\partial F}{\partial z}-v\mu\frac{\partial g}{\partial z}
+\frac{\partial}{\partial\mu}\left(D_{\mu\mu}\frac{\partial g}{\partial\mu}\right)
-\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^3A(\mu)F\right)
-\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^3A(\mu)g\right)
+N\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4D_{\mu\mu}\frac{\partial F}{\partial p}\right)
+N\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4D_{\mu\mu}\frac{\partial g}{\partial p}\right) \tag*{(141)} ∂ t ∂ F + ∂ t ∂ g = − v μ ∂ z ∂ F − v μ ∂ z ∂ g + ∂ μ ∂ ( D μμ ∂ μ ∂ g ) − p 2 1 ∂ p ∂ ( p 3 A ( μ ) F ) − p 2 1 ∂ p ∂ ( p 3 A ( μ ) g ) + N p 2 1 ∂ p ∂ ( p 4 D μμ ∂ p ∂ F ) + N p 2 1 ∂ p ∂ ( p 4 D μμ ∂ p ∂ g ) (141)
By integrating the latter equation over μ \mu μ from − 1 -1 − 1 to + 1 +1 + 1 , we obtain the following momentum equation
∂ F ∂ t = − κ p 1 p 2 ∂ ∂ p ( p 3 F ) + κ p p ( 0 ) 1 p 2 ∂ ∂ p ( p 4 ∂ F ∂ p ) + T 1 + T 2 + T 3 , (142) \frac{\partial F}{\partial t}
=-\kappa_p\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^3F\right)
+\kappa_{pp}^{(0)}\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4\frac{\partial F}{\partial p}\right)
+T_1+T_2+T_3, \tag*{(142)} ∂ t ∂ F = − κ p p 2 1 ∂ p ∂ ( p 3 F ) + κ pp ( 0 ) p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ F ) + T 1 + T 2 + T 3 , (142)
where the constant coefficients are defined as
κ p = 1 2 ∫ − 1 + 1 d μ A ( μ ) , (143) \kappa_p=\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu A(\mu), \tag*{(143)} κ p = 2 1 ∫ − 1 + 1 d μ A ( μ ) , (143)
κ p p ( 0 ) = N 2 ∫ − 1 + 1 d μ D μ μ . (144) \kappa_{pp}^{(0)}=\frac{N}{2}\int_{-1}^{+1}\mathrm{d}\mu D_{\mu\mu}. \tag*{(144)} κ pp ( 0 ) = 2 N ∫ − 1 + 1 d μ D μμ . (144)
The three terms T 1 T_1 T 1 , T 2 T_2 T 2 , and T 3 T_3 T 3 are given by
T 1 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ g , (145) T_1=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu g, \tag*{(145)} T 1 = − 2 v ∂ z ∂ ∫ − 1 + 1 d μμg , (145)
T 2 = − 1 2 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) g , (146) T_2=-\frac{1}{2p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)g, \tag*{(146)} T 2 = − 2 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) g , (146)
T 3 = 1 p 2 ∂ ∂ p p 4 ∂ ∂ p N 2 ∫ − 1 + 1 d μ D μ μ g . (147) T_3=\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial}{\partial p}\frac{N}{2}\int_{-1}^{+1}\mathrm{d}\mu D_{\mu\mu}g. \tag*{(147)} T 3 = p 2 1 ∂ p ∂ p 4 ∂ p ∂ 2 N ∫ − 1 + 1 d μ D μμ g . (147)
Integrating Equation (A-9) over μ \mu μ from − 1 -1 − 1 to μ \mu μ gives
∂ g ∂ μ = 1 D μ μ Φ ( μ ) (148) \frac{\partial g}{\partial\mu}=\frac{1}{D_{\mu\mu}}\Phi(\mu) \tag*{(148)} ∂ μ ∂ g = D μμ 1 Φ ( μ ) (148)
with the function
Φ ( μ ) = [ ∂ F ∂ t ( μ + 1 ) + ∂ ∂ t ∫ − 1 μ d μ g + v μ 2 − 1 2 ∂ F ∂ z + v ∂ ∂ z ∫ − 1 μ d μ μ g + 1 p 2 ∂ ∂ p p 3 F ∫ − 1 μ d μ A ( μ ) + 1 p 2 ∂ ∂ p p 3 ∫ − 1 μ d μ A ( μ ) g − 1 p 2 ∂ ∂ p ( p 4 ∂ F ∂ p N ∫ − 1 μ d μ D μ μ ) − N 1 p 2 ∂ ∂ p ( p 4 ∂ ∂ p ∫ − 1 μ d μ D μ μ g ) ] . (149) \begin{aligned}
\Phi(\mu)=\Bigg[&
\frac{\partial F}{\partial t}(\mu+1)
+\frac{\partial}{\partial t}\int_{-1}^{\mu}\mathrm{d}\mu g
+v\frac{\mu^2-1}{2}\frac{\partial F}{\partial z}
+v\frac{\partial}{\partial z}\int_{-1}^{\mu}\mathrm{d}\mu\mu g \\
&+\frac{1}{p^2}\frac{\partial}{\partial p}p^3F\int_{-1}^{\mu}\mathrm{d}\mu A(\mu)
+\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{\mu}\mathrm{d}\mu A(\mu)g \\
&-\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4\frac{\partial F}{\partial p}N\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}\right)
-N\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4\frac{\partial}{\partial p}\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}g\right)
\Bigg]. \tag*{(149)}
\end{aligned} Φ ( μ ) = [ ∂ t ∂ F ( μ + 1 ) + ∂ t ∂ ∫ − 1 μ d μg + v 2 μ 2 − 1 ∂ z ∂ F + v ∂ z ∂ ∫ − 1 μ d μμg + p 2 1 ∂ p ∂ p 3 F ∫ − 1 μ d μ A ( μ ) + p 2 1 ∂ p ∂ p 3 ∫ − 1 μ d μ A ( μ ) g − p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ F N ∫ − 1 μ d μ D μμ ) − N p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ ∫ − 1 μ d μ D μμ g ) ] . (149)
From formula (A-16) we can find
g ( μ ) = ( ∫ − 1 μ d μ D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μ μ ) Φ ( μ ) . (150) g(\mu)=\left(\int_{-1}^{\mu} \frac{\mathrm{d}\mu}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\frac{\mathrm{d}\mu}{D_{\mu\mu}}\right)\Phi(\mu). \tag*{(150)} g ( μ ) = ( ∫ − 1 μ D μμ d μ − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ D μμ d μ ) Φ ( μ ) . (150)
Deriving the formula for T1
The term T 1 T_1 T 1 can be rewritten as
T 1 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ g = ∑ n = 1 8 Z n (151) T_1=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu g=\sum_{n=1}^{8}Z_n \tag*{(151)} T 1 = − 2 v ∂ z ∂ ∫ − 1 + 1 d μμg = n = 1 ∑ 8 Z n (151)
with
Z 1 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ ∂ F ∂ t ( μ + 1 ) Z 2 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ ∂ ∂ t ∫ − 1 μ d μ g Z 3 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ v μ 2 − 1 2 ∂ F ∂ z Z 4 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ ∂ ∂ z ∫ − 1 μ d μ g Z 5 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ 1 p 2 ∂ ∂ p p 3 F ∫ − 1 μ d μ A ( μ ) Z 6 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ 1 p 2 ∂ ∂ p p 3 ∫ − 1 μ d μ A ( μ ) g Z 7 = v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ 1 p 2 ∂ ∂ p p 4 ∂ F ∂ p N ∫ − 1 μ d μ D μ μ Z 8 = v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ N 1 p 2 ∂ ∂ p p 4 ∂ ∂ p ∫ − 1 μ d μ D μ μ g (152) \begin{aligned}
Z_1&=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{\partial F}{\partial t}(\mu+1)\\
Z_2&=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{\partial}{\partial t}\int_{-1}^{\mu}\mathrm{d}\mu g\\
Z_3&=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}v\frac{\mu^2-1}{2}\frac{\partial F}{\partial z}\\
Z_4&=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{\partial}{\partial z}\int_{-1}^{\mu}\mathrm{d}\mu g\\
Z_5&=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{1}{p^2}\frac{\partial}{\partial p}p^3F\int_{-1}^{\mu}\mathrm{d}\mu A(\mu)\\
Z_6&=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{\mu}\mathrm{d}\mu A(\mu)g\\
Z_7&=\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial F}{\partial p}N\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}\\
Z_8&=\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}N\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial}{\partial p}\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}g \tag*{(152)}
\end{aligned} Z 1 Z 2 Z 3 Z 4 Z 5 Z 6 Z 7 Z 8 = − 2 v ∂ z ∂ ∫ − 1 + 1 d μμ ∫ − 1 μ d μ D μμ 1 ∂ t ∂ F ( μ + 1 ) = − 2 v ∂ z ∂ ∫ − 1 + 1 d μμ ∫ − 1 μ d μ D μμ 1 ∂ t ∂ ∫ − 1 μ d μg = − 2 v ∂ z ∂ ∫ − 1 + 1 d μμ ∫ − 1 μ d μ D μμ 1 v 2 μ 2 − 1 ∂ z ∂ F = − 2 v ∂ z ∂ ∫ − 1 + 1 d μμ ∫ − 1 μ d μ D μμ 1 ∂ z ∂ ∫ − 1 μ d μg = − 2 v ∂ z ∂ ∫ − 1 + 1 d μμ ∫ − 1 μ d μ D μμ 1 p 2 1 ∂ p ∂ p 3 F ∫ − 1 μ d μ A ( μ ) = − 2 v ∂ z ∂ ∫ − 1 + 1 d μμ ∫ − 1 μ d μ D μμ 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 μ d μ A ( μ ) g = 2 v ∂ z ∂ ∫ − 1 + 1 d μμ ∫ − 1 μ d μ D μμ 1 p 2 1 ∂ p ∂ p 4 ∂ p ∂ F N ∫ − 1 μ d μ D μμ = 2 v ∂ z ∂ ∫ − 1 + 1 d μμ ∫ − 1 μ d μ D μμ 1 N p 2 1 ∂ p ∂ p 4 ∂ p ∂ ∫ − 1 μ d μ D μμ g (152)
In this section, we retain only the terms involving momentum derivative operator ∂ / ∂ p \partial/\partial p ∂ / ∂ p , and neglect the vanishing terms. In addition, we set the third-order and higher derivative terms to zero, retaining only the lower-order terms. Based on these requirements, we obtain
Z 1 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ ∂ F ∂ t ( μ + 1 ) = 0 , (153) Z_1=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\,\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{\partial F}{\partial t}(\mu+1)=0, \tag*{(153)} Z 1 = − 2 v ∂ z ∂ ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ D μμ 1 ∂ t ∂ F ( μ + 1 ) = 0 , (153)
Z 2 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ ∂ ∂ t ∫ − 1 μ d μ g = 0 , (154) Z_2=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\,\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{\partial}{\partial t}\int_{-1}^{\mu}\mathrm{d}\mu g=0, \tag*{(154)} Z 2 = − 2 v ∂ z ∂ ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ D μμ 1 ∂ t ∂ ∫ − 1 μ d μg = 0 , (154)
Z 3 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ v μ 2 − 1 2 ∂ F ∂ z = 0 , (155) Z_3=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\,\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}v\frac{\mu^2-1}{2}\frac{\partial F}{\partial z}=0, \tag*{(155)} Z 3 = − 2 v ∂ z ∂ ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ D μμ 1 v 2 μ 2 − 1 ∂ z ∂ F = 0 , (155)
Z 4 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ v ∂ ∂ z ∫ − 1 μ d μ μ g = 0 , (156) Z_4=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\,\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}v\frac{\partial}{\partial z}\int_{-1}^{\mu}\mathrm{d}\mu\mu g=0, \tag*{(156)} Z 4 = − 2 v ∂ z ∂ ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ D μμ 1 v ∂ z ∂ ∫ − 1 μ d μμg = 0 , (156)
Z 5 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ 1 p 2 ∂ ∂ p p 3 ∫ − 1 μ d μ A ( μ ) = − κ z p ( 1 ) 1 p ∂ ∂ p p 3 ∂ F ∂ z , (157) Z_5=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\,\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{\mu}\mathrm{d}\mu A(\mu)=-\kappa_{zp}^{(1)}\frac{1}{p}\frac{\partial}{\partial p}p^3\frac{\partial F}{\partial z}, \tag*{(157)} Z 5 = − 2 v ∂ z ∂ ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ D μμ 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 μ d μ A ( μ ) = − κ z p ( 1 ) p 1 ∂ p ∂ p 3 ∂ z ∂ F , (157)
Z 6 = − v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ 1 p 2 ∂ ∂ p p 3 ∫ − 1 μ d μ A ( μ ) g = 0 , (158) Z_6=-\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\,\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{\mu}\mathrm{d}\mu A(\mu)g=0, \tag*{(158)} Z 6 = − 2 v ∂ z ∂ ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ D μμ 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 μ d μ A ( μ ) g = 0 , (158)
Z 7 = v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ 1 p 2 ∂ ∂ p p 4 ∂ F ∂ p N ∫ − 1 μ d μ D μ μ = 0 , (159) Z_7=\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\,\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial F}{\partial p}N\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}=0, \tag*{(159)} Z 7 = 2 v ∂ z ∂ ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ D μμ 1 p 2 1 ∂ p ∂ p 4 ∂ p ∂ F N ∫ − 1 μ d μ D μμ = 0 , (159)
Z 8 = v 2 ∂ ∂ z ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ N 1 p 2 ∂ ∂ p p 4 ∂ ∂ p ∫ − 1 μ d μ D μ μ g = 0 (160) Z_8=\frac{v}{2}\frac{\partial}{\partial z}\int_{-1}^{+1}\mathrm{d}\mu\,\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}N\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial}{\partial p}\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}g=0 \tag*{(160)} Z 8 = 2 v ∂ z ∂ ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ D μμ 1 N p 2 1 ∂ p ∂ p 4 ∂ p ∂ ∫ − 1 μ d μ D μμ g = 0 (160)
with
κ z p ( 1 ) = 1 2 ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ ∫ − 1 μ d μ A ( μ ) . (161) \kappa_{zp}^{(1)}=\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\,\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\int_{-1}^{\mu}\mathrm{d}\mu A(\mu). \tag*{(161)} κ z p ( 1 ) = 2 1 ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ D μμ 1 ∫ − 1 μ d μ A ( μ ) . (161)
Combining Equations (A-21)-(A-29), we derive
T 1 = − κ z p ( 1 ) 1 p ∂ ∂ p p 3 ∂ F ∂ z . (162) T_1=-\kappa_{zp}^{(1)}\frac{1}{p}\frac{\partial}{\partial p}p^3\frac{\partial F}{\partial z}. \tag*{(162)} T 1 = − κ z p ( 1 ) p 1 ∂ p ∂ p 3 ∂ z ∂ F . (162)
T2
The term T 2 T_2 T 2 can be rewritten as
T 2 = − 1 2 1 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) g = ∑ n = 1 8 X n (163) T_2=-\frac{1}{2}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)g=\sum_{n=1}^{8}X_n \tag*{(163)} T 2 = − 2 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) g = n = 1 ∑ 8 X n (163)
with
X 1 = − 1 2 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ∂ F ∂ t ( μ + 1 ) , (164) X_1=-\frac{1}{2p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{\partial F}{\partial t}(\mu+1), \tag*{(164)} X 1 = − 2 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ∂ t ∂ F ( μ + 1 ) , (164)
X 2 = − 1 2 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ∂ ∂ t ∫ − 1 μ d μ g , (165) X_2=-\frac{1}{2p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{\partial}{\partial t}\int_{-1}^{\mu}\mathrm{d}\mu g, \tag*{(165)} X 2 = − 2 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ∂ t ∂ ∫ − 1 μ d μg , (165)
X 3 = − 1 2 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ν μ 2 − 1 2 ∂ F ∂ z , (166) X_3=-\frac{1}{2p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\nu\frac{\mu^2-1}{2}\frac{\partial F}{\partial z}, \tag*{(166)} X 3 = − 2 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ν 2 μ 2 − 1 ∂ z ∂ F , (166)
X 4 = − 1 2 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ν ∂ ∂ z ∫ − 1 μ d μ g , (167) X_4=-\frac{1}{2p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\nu\frac{\partial}{\partial z}\int_{-1}^{\mu}\mathrm{d}\mu g, \tag*{(167)} X 4 = − 2 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ν ∂ z ∂ ∫ − 1 μ d μg , (167)
X 5 = 1 2 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) 1 p 2 ∂ ∂ p p 3 F ∫ − 1 μ d μ A ( μ ) , (168) X_5=\frac{1}{2p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{1}{p^2}\frac{\partial}{\partial p}p^3F\int_{-1}^{\mu}\mathrm{d}\mu A(\mu), \tag*{(168)} X 5 = 2 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) p 2 1 ∂ p ∂ p 3 F ∫ − 1 μ d μ A ( μ ) , (168)
X 6 = 1 2 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) 1 p 2 ∂ ∂ p p 3 ∫ − 1 μ d μ A ( μ ) g , (169) X_6=\frac{1}{2p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{\mu}\mathrm{d}\mu A(\mu)g, \tag*{(169)} X 6 = 2 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) p 2 1 ∂ p ∂ p 3 ∫ − 1 μ d μ A ( μ ) g , (169)
X 7 = 1 2 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) 1 p 2 ∂ ∂ p p 4 ∂ ∂ p N ∫ − 1 μ d μ D μ μ , (170) X_7=\frac{1}{2p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial}{\partial p}N\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}, \tag*{(170)} X 7 = 2 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) p 2 1 ∂ p ∂ p 4 ∂ p ∂ N ∫ − 1 μ d μ D μμ , (170)
X 8 = 1 2 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) 1 p 2 ∂ ∂ p p 4 ∂ ∂ p ∫ − 1 μ d μ D μ μ g (171) X_8=\frac{1}{2p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial}{\partial p}\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}g \tag*{(171)} X 8 = 2 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) p 2 1 ∂ p ∂ p 4 ∂ p ∂ ∫ − 1 μ d μ D μμ g (171)
Using the same methods and requirements as in the latter subsection, we obtain
X 1 = − 1 2 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ∂ F ∂ t ( μ + 1 ) = − κ p t ( 2 ) 1 p 2 ∂ ∂ p p 3 ∂ F ∂ t , (172) X_1=-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{\partial F}{\partial t}(\mu+1)=-\kappa_{pt}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\frac{\partial F}{\partial t}, \tag*{(172)} X 1 = − 2 1 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ∂ t ∂ F ( μ + 1 ) = − κ pt ( 2 ) p 2 1 ∂ p ∂ p 3 ∂ t ∂ F , (172)
X 2 = − 1 2 1 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ∂ ∂ t ∫ − 1 μ d μ g = 0 , (173) X_2=-\frac{1}{2}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{\partial}{\partial t}\int_{-1}^{\mu}\mathrm{d}\mu g=0, \tag*{(173)} X 2 = − 2 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ∂ t ∂ ∫ − 1 μ d μg = 0 , (173)
X 3 = − 1 2 1 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) μ 2 − 1 2 ∂ F ∂ z = − κ p z ( 2 ) 1 p 2 ∂ ∂ p p 4 ∂ F ∂ z , (174) \begin{aligned}
X_3={}&-\frac{1}{2}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{\mu^2-1}{2}\frac{\partial F}{\partial z}\\
={}&-\kappa_{pz}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial F}{\partial z}, \tag*{(174)}
\end{aligned} X 3 = = − 2 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) 2 μ 2 − 1 ∂ z ∂ F − κ p z ( 2 ) p 2 1 ∂ p ∂ p 4 ∂ z ∂ F , (174)
X 4 = − 1 2 1 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ∂ ∂ z ∫ − 1 μ d μ g = 0 , (175) X_4=-\frac{1}{2}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{\partial}{\partial z}\int_{-1}^{\mu}\mathrm{d}\mu g=0, \tag*{(175)} X 4 = − 2 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ∂ z ∂ ∫ − 1 μ d μg = 0 , (175)
X 5 = 1 2 1 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) 1 p 2 ∂ ∂ p p 3 F ∫ − 1 μ d μ A ( μ ) = κ p p ( 2 ) 1 p 2 ∂ ∂ p p ∂ ∂ p p 3 F , (176) \begin{aligned}
X_5={}&\frac{1}{2}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{1}{p^2}\frac{\partial}{\partial p}p^3F\int_{-1}^{\mu}\mathrm{d}\mu A(\mu)\\
={}&\kappa_{pp}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p\frac{\partial}{\partial p}p^3F, \tag*{(176)}
\end{aligned} X 5 = = 2 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) p 2 1 ∂ p ∂ p 3 F ∫ − 1 μ d μ A ( μ ) κ pp ( 2 ) p 2 1 ∂ p ∂ p ∂ p ∂ p 3 F , (176)
X 6 = 1 2 1 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) 1 p 2 ∂ ∂ p p 3 ∫ − 1 μ d μ A ( μ ) g = 0 , (177) X_6=\frac{1}{2}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{\mu}\mathrm{d}\mu A(\mu)g=0, \tag*{(177)} X 6 = 2 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) p 2 1 ∂ p ∂ p 3 ∫ − 1 μ d μ A ( μ ) g = 0 , (177)
X 7 = 1 2 1 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) 1 p 2 ∂ ∂ p p 4 ∂ F ∂ p N ∫ − 1 μ d μ D μ μ = 0 , (178) \begin{aligned}
X_7={}&\frac{1}{2}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial F}{\partial p}N\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}\\
={}&0, \tag*{(178)}
\end{aligned} X 7 = = 2 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) p 2 1 ∂ p ∂ p 4 ∂ p ∂ F N ∫ − 1 μ d μ D μμ 0 , (178)
X 8 = 1 2 1 p 2 ∂ ∂ p p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) N 1 p 2 ∂ ∂ p p 4 ∂ ∂ p ∫ − 1 μ d μ D μ μ g = 0 (179) X_8=\frac{1}{2}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)N\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial}{\partial p}\int_{-1}^{\mu}\mathrm{d}\mu D_{\mu\mu}g=0 \tag*{(179)} X 8 = 2 1 p 2 1 ∂ p ∂ p 3 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) N p 2 1 ∂ p ∂ p 4 ∂ p ∂ ∫ − 1 μ d μ D μμ g = 0 (179)
With all of the latter formulas, we find
T 2 = − κ p t ( 2 ) 1 p 2 ∂ ∂ p p 3 ∂ F ∂ t − κ p z ( 2 ) 1 p 2 ∂ ∂ p p 4 ∂ F ∂ z + κ p p ( 2 ) 1 p 2 ∂ ∂ p p ∂ ∂ p p 3 F (180) T_2=-\kappa_{pt}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\frac{\partial F}{\partial t}-\kappa_{pz}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial F}{\partial z}+\kappa_{pp}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p\frac{\partial}{\partial p}p^3F \tag*{(180)} T 2 = − κ pt ( 2 ) p 2 1 ∂ p ∂ p 3 ∂ t ∂ F − κ p z ( 2 ) p 2 1 ∂ p ∂ p 4 ∂ z ∂ F + κ pp ( 2 ) p 2 1 ∂ p ∂ p ∂ p ∂ p 3 F (180)
with
κ p t ( 2 ) = 1 2 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ( μ + 1 ) , (181) \kappa_{pt}^{(2)}=\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)(\mu+1), \tag*{(181)} κ pt ( 2 ) = 2 1 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ( μ + 1 ) , (181)
κ p z ( 2 ) = 1 2 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) μ 2 − 1 2 , (182) \kappa_{pz}^{(2)}=\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\frac{\mu^2-1}{2}, \tag*{(182)} κ p z ( 2 ) = 2 1 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) 2 μ 2 − 1 , (182)
κ p p ( 2 ) = 1 2 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ∫ − 1 μ d μ A ( μ ) . (183) \kappa_{pp}^{(2)}=\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu A(\mu)\left(\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}\mathrm{d}\mu\int_{-1}^{\mu}\mathrm{d}\mu\frac{1}{D_{\mu\mu}}\right)\int_{-1}^{\mu}\mathrm{d}\mu A(\mu). \tag*{(183)} κ pp ( 2 ) = 2 1 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ∫ − 1 μ d μ A ( μ ) . (183)
T3
Similarly, we find
T 3 = 1 p 2 ∂ ∂ p p 4 ∂ ∂ p N 2 ∫ − 1 + 1 d μ D μ μ g = 0 (184) T_3=\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial}{\partial p}\frac{N}{2}\int_{-1}^{+1}\mathrm{d}\mu D_{\mu\mu}g=0 \tag*{(184)} T 3 = p 2 1 ∂ p ∂ p 4 ∂ p ∂ 2 N ∫ − 1 + 1 d μ D μμ g = 0 (184)
The momentum transport equation
Inserting the formulas for T 1 T_1 T 1 , T 2 T_2 T 2 , and T 3 T_3 T 3 into Equation (A-10) yields
∂ F ∂ t = − κ p 1 p 2 ∂ ∂ p p 3 F + κ p p ( 0 ) 1 p 2 ∂ ∂ p p 4 ∂ F ∂ p − κ z p ( 1 ) 1 p ∂ ∂ p p 3 ∂ F ∂ z − κ p t ( 2 ) 1 p 2 ∂ ∂ p p 3 ∂ F ∂ t − κ p z ( 2 ) 1 p 2 ∂ ∂ p p 4 ∂ F ∂ z + κ p p ( 2 ) 1 p 2 ∂ ∂ p p ∂ ∂ p p 3 F (185) \frac{\partial F}{\partial t}=-\kappa_p\frac{1}{p^2}\frac{\partial}{\partial p}p^3F+\kappa_{pp}^{(0)}\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial F}{\partial p}-\kappa_{zp}^{(1)}\frac{1}{p}\frac{\partial}{\partial p}p^3\frac{\partial F}{\partial z}
-\kappa_{pt}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\frac{\partial F}{\partial t}-\kappa_{pz}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial F}{\partial z}+\kappa_{pp}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p\frac{\partial}{\partial p}p^3F \tag*{(185)} ∂ t ∂ F = − κ p p 2 1 ∂ p ∂ p 3 F + κ pp ( 0 ) p 2 1 ∂ p ∂ p 4 ∂ p ∂ F − κ z p ( 1 ) p 1 ∂ p ∂ p 3 ∂ z ∂ F − κ pt ( 2 ) p 2 1 ∂ p ∂ p 3 ∂ t ∂ F − κ p z ( 2 ) p 2 1 ∂ p ∂ p 4 ∂ z ∂ F + κ pp ( 2 ) p 2 1 ∂ p ∂ p ∂ p ∂ p 3 F (185)
with
κ p = 1 2 ∫ − 1 + 1 d μ A ( μ ) , (186) \kappa_p=\frac{1}{2}\int_{-1}^{+1}d\mu A(\mu), \tag*{(186)} κ p = 2 1 ∫ − 1 + 1 d μ A ( μ ) , (186)
κ p p ( 0 ) = N 2 ∫ − 1 + 1 d μ D μ μ , (187) \kappa_{pp}^{(0)}=\frac{N}{2}\int_{-1}^{+1}d\mu D_{\mu\mu}, \tag*{(187)} κ pp ( 0 ) = 2 N ∫ − 1 + 1 d μ D μμ , (187)
κ p t ( 2 ) = 1 2 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ( μ + 1 ) , (188) \kappa_{pt}^{(2)}=\frac{1}{2}\int_{-1}^{+1}d\mu A(\mu)\left(\int_{-1}^{\mu}d\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}d\mu\int_{-1}^{\mu}d\mu\frac{1}{D_{\mu\mu}}\right)(\mu+1), \tag*{(188)} κ pt ( 2 ) = 2 1 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ( μ + 1 ) , (188)
κ p z ( 2 ) = 1 2 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) μ 2 − 1 2 , (189) \kappa_{pz}^{(2)}=\frac{1}{2}\int_{-1}^{+1}d\mu A(\mu)\left(\int_{-1}^{\mu}d\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}d\mu\int_{-1}^{\mu}d\mu\frac{1}{D_{\mu\mu}}\right)\frac{\mu^2-1}{2}, \tag*{(189)} κ p z ( 2 ) = 2 1 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) 2 μ 2 − 1 , (189)
κ p p ( 2 ) = 1 2 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ 1 D μ μ − 1 2 ∫ − 1 + 1 d μ ∫ − 1 μ d μ 1 D μ μ ) ∫ − 1 μ d μ A ( μ ) , (190) \kappa_{pp}^{(2)}=\frac{1}{2}\int_{-1}^{+1}d\mu A(\mu)\left(\int_{-1}^{\mu}d\mu\frac{1}{D_{\mu\mu}}-\frac{1}{2}\int_{-1}^{+1}d\mu\int_{-1}^{\mu}d\mu\frac{1}{D_{\mu\mu}}\right)\int_{-1}^{\mu}d\mu A(\mu), \tag*{(190)} κ pp ( 2 ) = 2 1 ∫ − 1 + 1 d μ A ( μ ) ( ∫ − 1 μ d μ D μμ 1 − 2 1 ∫ − 1 + 1 d μ ∫ − 1 μ d μ D μμ 1 ) ∫ − 1 μ d μ A ( μ ) , (190)
κ z p ( 1 ) = 1 2 ∫ − 1 + 1 d μ μ ∫ − 1 μ d μ 1 D μ μ ∫ − 1 μ d μ A ( μ ) . (191) \kappa_{zp}^{(1)}=\frac{1}{2}\int_{-1}^{+1}d\mu\mu\int_{-1}^{\mu}d\mu\frac{1}{D_{\mu\mu}}\int_{-1}^{\mu}d\mu A(\mu). \tag*{(191)} κ z p ( 1 ) = 2 1 ∫ − 1 + 1 d μμ ∫ − 1 μ d μ D μμ 1 ∫ − 1 μ d μ A ( μ ) . (191)
The momentum transport equation (A-53) contains terms involving the first-order spatial derivative operator ∂ / ∂ p \partial/\partial p ∂ / ∂ p . However, this equation does not contain the spatial convection term. Therefore, we cannot perform the iteration for the first-order spatial convection. For simplicity, we neglect the terms in Equation (A-53) involving only the first-order spatial derivative operator
∂ F ∂ t = − κ p 1 p 2 ∂ ∂ p p 3 F + κ p p ( 0 ) 1 p 2 ∂ ∂ p p 4 ∂ F ∂ p + κ p p ( 2 ) 1 p 2 ∂ ∂ p p ∂ ∂ p p 3 F − κ p t ( 2 ) 1 p 2 ∂ ∂ p p 3 ∂ F ∂ t (192) \frac{\partial F}{\partial t}=-\kappa_p\frac{1}{p^2}\frac{\partial}{\partial p}p^3F+\kappa_{pp}^{(0)}\frac{1}{p^2}\frac{\partial}{\partial p}p^4\frac{\partial F}{\partial p}+\kappa_{pp}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p\frac{\partial}{\partial p}p^3F-\kappa_{pt}^{(2)}\frac{1}{p^2}\frac{\partial}{\partial p}p^3\frac{\partial F}{\partial t} \tag*{(192)} ∂ t ∂ F = − κ p p 2 1 ∂ p ∂ p 3 F + κ pp ( 0 ) p 2 1 ∂ p ∂ p 4 ∂ p ∂ F + κ pp ( 2 ) p 2 1 ∂ p ∂ p ∂ p ∂ p 3 F − κ pt ( 2 ) p 2 1 ∂ p ∂ p 3 ∂ t ∂ F (192)
Inserting Equation (refA42) into itself, we obtain
∂ F ∂ t = − κ p 1 p 2 ∂ ∂ p ( p 3 F ) + κ p p ( 0 ) 1 p 2 ∂ ∂ p ( p 4 ∂ F ∂ p ) + ( κ p t ( 2 ) κ p + κ p p ( 2 ) ) 1 p 2 ∂ ∂ p [ p ∂ ∂ p ( p 3 F ) ] (193) \frac{\partial F}{\partial t}=-\kappa_p\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^3F\right)+\kappa_{pp}^{(0)}\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4\frac{\partial F}{\partial p}\right)+\left(\kappa_{pt}^{(2)}\kappa_p+\kappa_{pp}^{(2)}\right)\frac{1}{p^2}\frac{\partial}{\partial p}\left[p\frac{\partial}{\partial p}\left(p^3F\right)\right] \tag*{(193)} ∂ t ∂ F = − κ p p 2 1 ∂ p ∂ ( p 3 F ) + κ pp ( 0 ) p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ F ) + ( κ pt ( 2 ) κ p + κ pp ( 2 ) ) p 2 1 ∂ p ∂ [ p ∂ p ∂ ( p 3 F ) ] (193)
The latter equation can be rewritten as
∂ F ∂ t = − K p 1 p 2 ∂ ∂ p ( p 3 F ) + K 1 1 p 2 ∂ ∂ p ( p 4 ∂ F ∂ p ) + K 2 1 p 2 ∂ ∂ p [ p ∂ ∂ p ( p 3 F ) ] (194) \frac{\partial F}{\partial t}=-\mathcal{K}_p\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^3F\right)+\mathcal{K}_1\frac{1}{p^2}\frac{\partial}{\partial p}\left(p^4\frac{\partial F}{\partial p}\right)+\mathcal{K}_2\frac{1}{p^2}\frac{\partial}{\partial p}\left[p\frac{\partial}{\partial p}\left(p^3F\right)\right] \tag*{(194)} ∂ t ∂ F = − K p p 2 1 ∂ p ∂ ( p 3 F ) + K 1 p 2 1 ∂ p ∂ ( p 4 ∂ p ∂ F ) + K 2 p 2 1 ∂ p ∂ [ p ∂ p ∂ ( p 3 F ) ] (194)
with
K p = κ p , (195) \mathcal{K}_p=\kappa_p, \tag*{(195)} K p = κ p , (195)
K 1 = κ p p ( 0 ) , (196) \mathcal{K}_1=\kappa_{pp}^{(0)}, \tag*{(196)} K 1 = κ pp ( 0 ) , (196)
K 2 = κ p t ( 2 ) κ p + κ p p ( 2 ) . (197) \mathcal{K}_2=\kappa_{pt}^{(2)}\kappa_p+\kappa_{pp}^{(2)}. \tag*{(197)} K 2 = κ pt ( 2 ) κ p + κ pp ( 2 ) . (197)
The derivation of the formula for the moment of the fractional transport equation
After integrating by parts, we can rewrite the integrals I 1 I_1 I 1 and I 2 I_2 I 2 as
I 1 = J 1 + J 2 + J 3 , (198) I_1=J_1+J_2+J_3, \tag*{(198)} I 1 = J 1 + J 2 + J 3 , (198)
I 2 = J 4 + J 5 + J 6 (199) I_2=J_4+J_5+J_6 \tag*{(199)} I 2 = J 4 + J 5 + J 6 (199)
with
J 1 = ∫ − ∞ 0 d z ∣ z ∣ α − n ∫ − ∞ z d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (200) J_1=\int_{-\infty}^{0} dz|z|^{\alpha-n}\int_{-\infty}^{z}d\xi\,(z-\xi)^{n-\beta-1}F(\xi), \tag*{(200)} J 1 = ∫ − ∞ 0 d z ∣ z ∣ α − n ∫ − ∞ z d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (200)
J 2 = ∫ 0 + ∞ d z ∣ z ∣ α − n ∫ − ∞ 0 d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (201) J_2=\int_{0}^{+\infty} dz|z|^{\alpha-n}\int_{-\infty}^{0}d\xi\,(z-\xi)^{n-\beta-1}F(\xi), \tag*{(201)} J 2 = ∫ 0 + ∞ d z ∣ z ∣ α − n ∫ − ∞ 0 d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (201)
J 3 = ∫ 0 + ∞ d z ∣ z ∣ α − n ∫ 0 z d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (202) J_3=\int_{0}^{+\infty} dz|z|^{\alpha-n}\int_{0}^{z}d\xi\,(z-\xi)^{n-\beta-1}F(\xi), \tag*{(202)} J 3 = ∫ 0 + ∞ d z ∣ z ∣ α − n ∫ 0 z d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (202)
J 4 = ∫ − ∞ 0 d z ∣ z ∣ α − n ∫ z 0 d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (203) J_4=\int_{-\infty}^{0} dz|z|^{\alpha-n}\int_{z}^{0}d\xi\,(z-\xi)^{n-\beta-1}F(\xi), \tag*{(203)} J 4 = ∫ − ∞ 0 d z ∣ z ∣ α − n ∫ z 0 d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (203)
J 5 = ∫ − ∞ 0 d z ∣ z ∣ α − n ∫ 0 + ∞ d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (204) J_5=\int_{-\infty}^{0} dz|z|^{\alpha-n}\int_{0}^{+\infty}d\xi\,(z-\xi)^{n-\beta-1}F(\xi), \tag*{(204)} J 5 = ∫ − ∞ 0 d z ∣ z ∣ α − n ∫ 0 + ∞ d ξ ( z − ξ ) n − β − 1 F ( ξ ) , (204)
J 6 = ∫ 0 + ∞ d z ∣ z ∣ α − n ∫ z + ∞ d ξ ( z − ξ ) n − β − 1 F ( ξ ) . (205) J_6=\int_{0}^{+\infty} dz|z|^{\alpha-n}\int_{z}^{+\infty}d\xi\,(z-\xi)^{n-\beta-1}F(\xi). \tag*{(205)} J 6 = ∫ 0 + ∞ d z ∣ z ∣ α − n ∫ z + ∞ d ξ ( z − ξ ) n − β − 1 F ( ξ ) . (205)
Defining d z = ξ d η dz=\xi d\eta d z = ξ d η , we can rewrite J 1 J_1 J 1 as
J 1 = 1 ( − 1 ) β − n ∫ − ∞ 0 d ξ F ( ξ ) ∣ ξ ∣ α − n ∫ 0 1 d η η α − n ( 1 − η ) n − β − 1 ( z − ξ ) n − β − 1 = 1 ( − 1 ) β − n Γ ( α − n + 1 ) Γ ( n − β ) Γ ( α − β + 1 ) ∫ − ∞ 0 d ξ F ( ξ ) ∣ ξ ∣ α − β . (206) \begin{aligned}
J_1=\frac{1}{(-1)^{\beta-n}}\int_{-\infty}^{0}d\xi F(\xi)|\xi|^{\alpha-n}\int_{0}^{1}d\eta\eta^{\alpha-n}(1-\eta)^{n-\beta-1}(z-\xi)^{n-\beta-1} \\
&=\frac{1}{(-1)^{\beta-n}}\frac{\Gamma(\alpha-n+1)\Gamma(n-\beta)}{\Gamma(\alpha-\beta+1)}\int_{-\infty}^{0}d\xi F(\xi)|\xi|^{\alpha-\beta}. \tag*{(206)}
\end{aligned} J 1 = ( − 1 ) β − n 1 ∫ − ∞ 0 d ξ F ( ξ ) ∣ ξ ∣ α − n ∫ 0 1 d η η α − n ( 1 − η ) n − β − 1 ( z − ξ ) n − β − 1 = ( − 1 ) β − n 1 Γ ( α − β + 1 ) Γ ( α − n + 1 ) Γ ( n − β ) ∫ − ∞ 0 d ξ F ( ξ ) ∣ ξ ∣ α − β . (206)
Similarly, with ξ = − ξ ′ \xi=-\xi' ξ = − ξ ′ and z = − z ′ z=-z' z = − z ′ , we obtain
J 2 = ∫ 0 + ∞ d ξ ′ F ( − ξ ′ ) ∫ − ∞ 0 d z ′ ( ξ ′ − z ′ ) n − β − 1 ∣ z ′ ∣ α − n . (207) J_2=\int_{0}^{+\infty}d\xi'F(-\xi')\int_{-\infty}^{0}dz'(\xi'-z')^{n-\beta-1}|z'|^{\alpha-n}. \tag*{(207)} J 2 = ∫ 0 + ∞ d ξ ′ F ( − ξ ′ ) ∫ − ∞ 0 d z ′ ( ξ ′ − z ′ ) n − β − 1 ∣ z ′ ∣ α − n . (207)
In addition, using the relation z = ξ / η z=\xi/\eta z = ξ / η , we derive
J 3 = Γ ( β − α ) Γ ( n − β ) Γ ( n − α ) ∫ 0 + ∞ d ξ F ( ξ ) ∣ ξ ∣ α − β , (208) J_{3}=\frac{\Gamma(\beta-\alpha)\Gamma(n-\beta)}{\Gamma(n-\alpha)}\int_{0}^{+\infty}\mathrm{d}\xi F(\xi)|\xi|^{\alpha-\beta}, \tag*{(208)} J 3 = Γ ( n − α ) Γ ( β − α ) Γ ( n − β ) ∫ 0 + ∞ d ξ F ( ξ ) ∣ ξ ∣ α − β , (208)
J 4 = 1 ( − 1 ) α − n Γ ( β − α ) Γ ( n − β ) Γ ( n − α ) ∫ − ∞ 0 d ξ F ( ξ ) ∣ ξ ∣ α − β , (209) J_{4}=\frac{1}{(-1)^{\alpha-n}}\frac{\Gamma(\beta-\alpha)\Gamma(n-\beta)}{\Gamma(n-\alpha)}\int_{-\infty}^{0}\mathrm{d}\xi F(\xi)|\xi|^{\alpha-\beta}, \tag*{(209)} J 4 = ( − 1 ) α − n 1 Γ ( n − α ) Γ ( β − α ) Γ ( n − β ) ∫ − ∞ 0 d ξ F ( ξ ) ∣ ξ ∣ α − β , (209)
J 5 = ∫ 0 + ∞ d ξ F ( ξ ) ∫ − ∞ 0 d z ( ξ − z ) n − β − 1 ∣ z ∣ β − n . (210) J_{5}=\int_{0}^{+\infty}\mathrm{d}\xi F(\xi)\int_{-\infty}^{0}\mathrm{d}z\,(\xi-z)^{n-\beta-1}|z|^{\beta-n}. \tag*{(210)} J 5 = ∫ 0 + ∞ d ξ F ( ξ ) ∫ − ∞ 0 d z ( ξ − z ) n − β − 1 ∣ z ∣ β − n . (210)
With z = ξ η z=\xi\eta z = ξ η , J 6 J_{6} J 6 becomes
J 6 = Γ ( α − n + 1 ) Γ ( n − β ) Γ ( α − β + 1 ) ∫ 0 + ∞ d ξ F ( ξ ) ∣ ξ ∣ α − β . (211) J_{6}=\frac{\Gamma(\alpha-n+1)\Gamma(n-\beta)}{\Gamma(\alpha-\beta+1)}\int_{0}^{+\infty}\mathrm{d}\xi F(\xi)|\xi|^{\alpha-\beta}. \tag*{(211)} J 6 = Γ ( α − β + 1 ) Γ ( α − n + 1 ) Γ ( n − β ) ∫ 0 + ∞ d ξ F ( ξ ) ∣ ξ ∣ α − β . (211)
Thus, Equation (111) becomes
∫ − ∞ + ∞ d z ∣ z ∣ α ∂ F ∂ t = ( − 1 ) n + 1 χ 2 cos π β 2 Γ ( α + 1 ) Γ ( n − β ) Γ ( α − n + 1 ) W . (212) \int_{-\infty}^{+\infty}\mathrm{d}z|z|^{\alpha}\frac{\partial F}{\partial t}=(-1)^{n+1}\frac{\chi}{2\cos\frac{\pi\beta}{2}}\frac{\Gamma(\alpha+1)}{\Gamma(n-\beta)\Gamma(\alpha-n+1)}W. \tag*{(212)} ∫ − ∞ + ∞ d z ∣ z ∣ α ∂ t ∂ F = ( − 1 ) n + 1 2 cos 2 π β χ Γ ( n − β ) Γ ( α − n + 1 ) Γ ( α + 1 ) W . (212)
Here,
W = W 1 + W 2 + W 3 (213) W=W_{1}+W_{2}+W_{3} \tag*{(213)} W = W 1 + W 2 + W 3 (213)
with
W 1 = J 1 + ( − 1 ) n J 6 , (214) W_{1}=J_{1}+(-1)^{n}J_{6}, \tag*{(214)} W 1 = J 1 + ( − 1 ) n J 6 , (214)
W 2 = J 2 + ( − 1 ) n J 5 , (215) W_{2}=J_{2}+(-1)^{n}J_{5}, \tag*{(215)} W 2 = J 2 + ( − 1 ) n J 5 , (215)
W 3 = J 3 + ( − 1 ) n J 4 . (216) W_{3}=J_{3}+(-1)^{n}J_{4}. \tag*{(216)} W 3 = J 3 + ( − 1 ) n J 4 . (216)
If we assume the distribution function is an even function in x x x , i.e., f ( ξ ) = f ( − ξ ) f(\xi)=f(-\xi) f ( ξ ) = f ( − ξ ) , we derive
W 1 = 1 ( − 1 ) β − n Γ ( α − n + 1 ) Γ ( n − β ) Γ ( α − β + 1 ) ∫ − ∞ 0 d ξ F ( ξ ) ∣ ξ ∣ α − β + ( − 1 ) n Γ ( α − n + 1 ) Γ ( n − β ) Γ ( α − β + 1 ) ∫ 0 + ∞ d ξ F ( ξ ) ∣ ξ ∣ α − β = 1 2 ( − 1 ) n Γ ( α − n + 1 ) Γ ( n − β ) Γ ( α − β + 1 ) [ 1 ( − 1 ) β + 1 ] ⟨ ∣ z ∣ α − β ⟩ . (217) \begin{aligned}
W_{1}&=\frac{1}{(-1)^{\beta-n}}\frac{\Gamma(\alpha-n+1)\Gamma(n-\beta)}{\Gamma(\alpha-\beta+1)}\int_{-\infty}^{0}\mathrm{d}\xi F(\xi)|\xi|^{\alpha-\beta} \\
&\quad+(-1)^{n}\frac{\Gamma(\alpha-n+1)\Gamma(n-\beta)}{\Gamma(\alpha-\beta+1)}\int_{0}^{+\infty}\mathrm{d}\xi F(\xi)|\xi|^{\alpha-\beta} \\
&=\frac{1}{2}(-1)^{n}\frac{\Gamma(\alpha-n+1)\Gamma(n-\beta)}{\Gamma(\alpha-\beta+1)}
\left[\frac{1}{(-1)^{\beta}}+1\right]\langle|z|^{\alpha-\beta}\rangle. \tag*{(217)}
\end{aligned} W 1 = ( − 1 ) β − n 1 Γ ( α − β + 1 ) Γ ( α − n + 1 ) Γ ( n − β ) ∫ − ∞ 0 d ξ F ( ξ ) ∣ ξ ∣ α − β + ( − 1 ) n Γ ( α − β + 1 ) Γ ( α − n + 1 ) Γ ( n − β ) ∫ 0 + ∞ d ξ F ( ξ ) ∣ ξ ∣ α − β = 2 1 ( − 1 ) n Γ ( α − β + 1 ) Γ ( α − n + 1 ) Γ ( n − β ) [ ( − 1 ) β 1 + 1 ] ⟨ ∣ z ∣ α − β ⟩ . (217)
and
W 2 = [ 1 + ( − 1 ) n ] 1 2 ∫ − ∞ 0 d ξ F ( ξ ) ∫ 0 + ∞ d z ∣ z ∣ β − n ( z − ξ ) α − n + 1 (218) W_{2}=\left[1+(-1)^{n}\right]\frac{1}{2}\int_{-\infty}^{0}\mathrm{d}\xi F(\xi)\int_{0}^{+\infty}\mathrm{d}z\frac{|z|^{\beta-n}}{(z-\xi)^{\alpha-n+1}} \tag*{(218)} W 2 = [ 1 + ( − 1 ) n ] 2 1 ∫ − ∞ 0 d ξ F ( ξ ) ∫ 0 + ∞ d z ( z − ξ ) α − n + 1 ∣ z ∣ β − n (218)
Similarly, for F ( ξ ) = F ( − ξ ) F(\xi)=F(-\xi) F ( ξ ) = F ( − ξ ) we have
W 3 = 1 2 Γ ( β − α ) Γ ( n − β ) Γ ( n − α ) [ 1 + 1 ( − 1 ) α ] ⟨ ∣ ξ ∣ α − β ⟩ (219) W_3=\frac{1}{2}\frac{\Gamma(\beta-\alpha)\Gamma(n-\beta)}{\Gamma(n-\alpha)}\left[1+\frac{1}{(-1)^\alpha}\right]\left\langle|\xi|^{\alpha-\beta}\right\rangle \tag*{(219)} W 3 = 2 1 Γ ( n − α ) Γ ( β − α ) Γ ( n − β ) [ 1 + ( − 1 ) α 1 ] ⟨ ∣ ξ ∣ α − β ⟩ (219)
and
W = 1 2 ( − 1 ) n Γ ( α − n + 1 ) Γ ( n − β ) Γ ( α − β + 1 ) [ 1 ( − 1 ) β + 1 ] ⟨ ∣ ξ ∣ α − β ⟩ + 1 2 Γ ( β − α ) Γ ( n − β ) Γ ( n − α ) [ 1 + 1 ( − 1 ) α ] ⟨ ∣ ξ ∣ α − β ⟩ (220) W=\frac{1}{2}(-1)^n\frac{\Gamma(\alpha-n+1)\Gamma(n-\beta)}{\Gamma(\alpha-\beta+1)}\left[\frac{1}{(-1)^\beta}+1\right]\left\langle|\xi|^{\alpha-\beta}\right\rangle
+\frac{1}{2}\frac{\Gamma(\beta-\alpha)\Gamma(n-\beta)}{\Gamma(n-\alpha)}\left[1+\frac{1}{(-1)^\alpha}\right]\left\langle|\xi|^{\alpha-\beta}\right\rangle \tag*{(220)} W = 2 1 ( − 1 ) n Γ ( α − β + 1 ) Γ ( α − n + 1 ) Γ ( n − β ) [ ( − 1 ) β 1 + 1 ] ⟨ ∣ ξ ∣ α − β ⟩ + 2 1 Γ ( n − α ) Γ ( β − α ) Γ ( n − β ) [ 1 + ( − 1 ) α 1 ] ⟨ ∣ ξ ∣ α − β ⟩ (220)
Since 0 < β < α < n = 1 0<\beta<\alpha<n=1 0 < β < α < n = 1 and β < 1 / 2 \beta<1/2 β < 1/2 ,
W = 1 2 { Γ ( β − α ) Γ ( n − α ) ( 1 + 1 ( − 1 ) α ) − Γ ( α − n + 1 ) Γ ( α − β + 1 ) ( 1 + 1 ( − 1 ) β ) } Γ ( n − β ) ⟨ ∣ ξ ∣ α − β ⟩ = ( Γ ( β − α ) Γ ( n − α ) cos 2 α π 2 − Γ ( α − n + 1 ) Γ ( α − β + 1 ) cos 2 β π 2 ) Γ ( n − β ) ⟨ ∣ z ∣ α − β ⟩ . (221) \begin{aligned}
W={}&\frac{1}{2}\left\{\frac{\Gamma(\beta-\alpha)}{\Gamma(n-\alpha)}\left(1+\frac{1}{(-1)^\alpha}\right)-\frac{\Gamma(\alpha-n+1)}{\Gamma(\alpha-\beta+1)}\left(1+\frac{1}{(-1)^\beta}\right)\right\}\Gamma(n-\beta)\left\langle|\xi|^{\alpha-\beta}\right\rangle\\
={}&\left(\frac{\Gamma(\beta-\alpha)}{\Gamma(n-\alpha)}\cos^2\frac{\alpha\pi}{2}-\frac{\Gamma(\alpha-n+1)}{\Gamma(\alpha-\beta+1)}\cos^2\frac{\beta\pi}{2}\right)\Gamma(n-\beta)\left\langle|z|^{\alpha-\beta}\right\rangle. \tag*{(221)}
\end{aligned} W = = 2 1 { Γ ( n − α ) Γ ( β − α ) ( 1 + ( − 1 ) α 1 ) − Γ ( α − β + 1 ) Γ ( α − n + 1 ) ( 1 + ( − 1 ) β 1 ) } Γ ( n − β ) ⟨ ∣ ξ ∣ α − β ⟩ ( Γ ( n − α ) Γ ( β − α ) cos 2 2 α π − Γ ( α − β + 1 ) Γ ( α − n + 1 ) cos 2 2 β π ) Γ ( n − β ) ⟨ ∣ z ∣ α − β ⟩ . (221)
Inserting the latter equation into Equation (B-80) gives
d d t ⟨ ∣ z ∣ α ⟩ = χ ( − 1 ) n + 1 2 cos π β 2 Γ ( α + 1 ) Γ ( n − β ) Γ ( α − n + 1 ) ( Γ ( β − α ) Γ ( n − α ) cos 2 α π 2 − Γ ( α − n + 1 ) Γ ( α − β + 1 ) cos 2 β π 2 ) Γ ( n − β ) ⟨ ∣ z ∣ α − β ⟩ , (222) \frac{d}{dt}\left\langle|z|^\alpha\right\rangle=\chi\frac{(-1)^{n+1}}{2\cos\frac{\pi\beta}{2}}\frac{\Gamma(\alpha+1)}{\Gamma(n-\beta)\Gamma(\alpha-n+1)}\left(\frac{\Gamma(\beta-\alpha)}{\Gamma(n-\alpha)}\cos^2\frac{\alpha\pi}{2}-\frac{\Gamma(\alpha-n+1)}{\Gamma(\alpha-\beta+1)}\cos^2\frac{\beta\pi}{2}\right)\Gamma(n-\beta)\left\langle|z|^{\alpha-\beta}\right\rangle, \tag*{(222)} d t d ⟨ ∣ z ∣ α ⟩ = χ 2 cos 2 π β ( − 1 ) n + 1 Γ ( n − β ) Γ ( α − n + 1 ) Γ ( α + 1 ) ( Γ ( n − α ) Γ ( β − α ) cos 2 2 α π − Γ ( α − β + 1 ) Γ ( α − n + 1 ) cos 2 2 β π ) Γ ( n − β ) ⟨ ∣ z ∣ α − β ⟩ , (222)
from which the formula of coefficient χ \chi χ can be obtained.