电子顶点函数
形状因子
现在我们来讨论顶点修正。一阶修正为:添加一个额外的虚光子连接两条腿,构成一个圈。
一般情况,我们用下图表示所有的 amputated 圈图的总和。
令它对应的取值为:− i e Γ μ ( p ′ , p ) -ie\Gamma^{\mu}(p',p) − i e Γ μ ( p ′ , p ) ,跃迁矩阵元可以写为:
i M = i e 2 ( u ˉ ( p ′ ) Γ ( p ′ , p ) u ( p ) ) 1 q 2 ( u ˉ ( k ′ ) γ μ u ( k ) ) (1) i\mathcal{M}= ie^2(\bar{u}(p')\Gamma(p',p)u(p))\frac{1}{q^2}(\bar{u}(k')\gamma_{\mu}u(k))\tag{1}
i M = i e 2 ( u ˉ ( p ′ ) Γ ( p ′ , p ) u ( p ) ) q 2 1 ( u ˉ ( k ′ ) γ μ u ( k ) ) ( 1 )
更一般的讨论,Γ μ \Gamma^{\mu} Γ μ 是对顶点的修正,也出现在其他的过程中。以下哈密顿量描述了电子与电磁场的相互作用:
Δ H i n t = ∫ d 3 x e A μ c l j μ \Delta H_{int} = \int d^3x eA_{\mu}^{cl} j^{\mu}
Δ H i n t = ∫ d 3 x e A μ c l j μ
其中 j μ = ψ ˉ ( x ) γ μ ψ ( x ) j^{\mu} = \bar{\psi}(x)\gamma^{\mu}\psi(x) j μ = ψ ˉ ( x ) γ μ ψ ( x ) ,A μ c l A^{cl}_{\mu} A μ c l 为一个固定的经典势。根据微扰论,散射矩阵元为:
i M ( 2 π ) δ ( p ′ 0 − p 0 ) = − i e u ˉ ( p ′ ) γ μ u ( p ) ⋅ A ~ μ c l ( p ′ − p ) i\mathcal{M}(2\pi)\delta(p'^0-p^0) = -ie\bar{u}(p')\gamma^{\mu}u(p)\cdot\tilde{A}^{cl}_{\mu}(p'-p)
i M ( 2 π ) δ ( p ′ 0 − p 0 ) = − i e u ˉ ( p ′ ) γ μ u ( p ) ⋅ A ~ μ c l ( p ′ − p )
当考虑顶点修正的情况,需要将散射矩阵元的表达式修改为:
i M ( 2 π ) δ ( p 0 − p ) = − i e u ˉ ( p ′ ) Γ μ ( p ′ , p ) u ( p ) ⋅ A ~ μ c l ( p ′ − p ) i\mathcal{M}(2\pi)\delta(p^0-p) = -ie\bar{u}(p')\Gamma^{\mu}(p',p)u(p)\cdot\tilde{A}^{cl}_{\mu}(p'-p)
i M ( 2 π ) δ ( p 0 − p ) = − i e u ˉ ( p ′ ) Γ μ ( p ′ , p ) u ( p ) ⋅ A ~ μ c l ( p ′ − p )
Γ μ ( p ′ , p ) \Gamma^{\mu}(p',p) Γ μ ( p ′ , p ) 应当是一个 p , p ′ , γ μ , m , e , ⋯ p,p',\gamma^{\mu},m,e,\cdots p , p ′ , γ μ , m , e , ⋯ 构成的函数。其领头阶来自于树图的贡献:
Γ μ = γ μ \Gamma^{\mu} = \gamma^{\mu}
Γ μ = γ μ
我们考虑次领头阶的贡献,Γ μ \Gamma^{\mu} Γ μ 矩阵可以写为:
Γ μ = γ μ ⋅ A + ( p ′ μ + p μ ) ⋅ B + ( p ′ μ − p μ ) ⋅ C (2) \Gamma^{\mu} = \gamma^{\mu}\cdot A + (p'^{\mu}+p^{\mu})\cdot B + (p'^{\mu}-p^{\mu})\cdot C \tag{2}
Γ μ = γ μ ⋅ A + ( p ′ μ + p μ ) ⋅ B + ( p ′ μ − p μ ) ⋅ C ( 2 )
现在考虑 A , B , C A,B,C A , B , C 可能具有的形式。若含有 p \cancel{p} p 或 p ′ \cancel{p}' p ′ ,可以化简为:
p u ( p ) = m ⋅ u ( p ) u ˉ ( p ′ ) p ′ = u ˉ ( p ′ ) ⋅ m \begin{aligned}
\cancel{p} u(p) = m\cdot u(p)\\
\bar{u}(p') \cancel{p}'= \bar{u}(p')\cdot m \\
\end{aligned}
p u ( p ) = m ⋅ u ( p ) u ˉ ( p ′ ) p ′ = u ˉ ( p ′ ) ⋅ m
那么实际上含有 p \cancel{p} p 或 p ′ \cancel{p}' p ′ 就等价于一些常数。那么唯一可能出现的非平凡的标量为:
q 2 = − 2 p ′ ⋅ p + 2 m 2 q^2 = -2p'\cdot p + 2m^2
q 2 = − 2 p ′ ⋅ p + 2 m 2
Ward 等式要求 q μ Γ μ = 0 q_{\mu}\Gamma^{\mu} = 0 q μ Γ μ = 0 ,根据 ( 2 ) (2) ( 2 ) 式,必须要求 C = 0 C = 0 C = 0 。
利用 Gordon 等式:
u ˉ ( p ′ ) γ μ u ( p ) = u ˉ ( p ′ ) ( p ′ μ + p μ 2 m + i σ μ ν q ν 2 m ) u ( p ) (3) \bar{u}(p')\gamma^{\mu}u(p) = \bar{u}(p')(\frac{p'^{\mu}+p^{\mu}}{2m} + \frac{i\sigma^{\mu\nu}q_{\nu}}{2m})u(p) \tag{3}
u ˉ ( p ′ ) γ μ u ( p ) = u ˉ ( p ′ ) ( 2 m p ′ μ + p μ + 2 m i σ μ ν q ν ) u ( p ) ( 3 )
可以将 ( 2 ) (2) ( 2 ) 式改写为:
Γ μ = γ μ F 1 ( q 2 ) + i σ μ ν q ν 2 m F 2 ( q 2 ) \begin{aligned}
\Gamma^{\mu} = \gamma^{\mu} F_1(q^2) + \frac{i\sigma^{\mu\nu}q_{\nu}}{2m}F_2(q^2)
\end{aligned}
Γ μ = γ μ F 1 ( q 2 ) + 2 m i σ μ ν q ν F 2 ( q 2 )
其中 F 1 ( q 2 ) , F 2 ( q 2 ) F_1(q^2),F_2(q^2) F 1 ( q 2 ) , F 2 ( q 2 ) 称为 形状因子 form factor 。F 1 F_1 F 1 称为 Dirac 形状因子,F 2 F_2 F 2 称为 Pauli 形状因子。
我们首先对静电场中的非相对论散射进行计算。取 A μ c l ( x ) = ( ϕ ( x ) , 0 ) A_{\mu}^{cl}(x) = (\phi(\bm{x}),\bm{0}) A μ c l ( x ) = ( ϕ ( x ) , 0 ) ,得到:
A ~ μ c l ( q ) = ( ( 2 π ) δ ( q 0 ) ϕ ~ ( q ) , 0 ) \tilde{A}^{cl}_{\mu}(q) = ((2\pi)\delta(q^0)\tilde{\phi}(\bm{q}),\bm{0})
A ~ μ c l ( q ) = ( ( 2 π ) δ ( q 0 ) ϕ ~ ( q ) , 0 )
得到跃迁矩阵元为:
i M = − i e u ˉ ( p ′ ) Γ 0 ( p ′ , p ) u ( p ) ϕ ~ ( q ) i\mathcal{M} = -ie\bar{u}(p')\Gamma^0(p',p)u(p)\tilde{\phi}(\bm{q})
i M = − i e u ˉ ( p ′ ) Γ 0 ( p ′ , p ) u ( p ) ϕ ~ ( q )
我们假设静电场变换的很缓慢,如此 ϕ ( q ) \phi(\bm{q}) ϕ ( q ) 主要集中在 q = 0 \bm{q} = 0 q = 0 处。我们取极限 q → 0 \bm{q}\rightarrow 0 q → 0 。此时 F 1 ( q 2 ) F_1(q^2) F 1 ( q 2 ) 项对散射具有主要贡献,且有:
u ˉ ( p ′ ) γ 0 u ( p ) = u † ( p ′ ) u ( p ) = 2 E p ξ † ξ ≈ 2 m ξ † ξ \bar{u}(p')\gamma^0 u(p) = u^\dagger(p')u(p) = 2E_p\xi^\dagger\xi \approx 2m\xi^\dagger\xi
u ˉ ( p ′ ) γ 0 u ( p ) = u † ( p ′ ) u ( p ) = 2 E p ξ † ξ ≈ 2 m ξ † ξ
将跃迁矩阵元写为:
i M = − i e F 1 ( 0 ) ϕ ~ ( q ) ⋅ 2 m ξ † ξ i\mathcal{M} = -ie F_1(0)\tilde{\phi}(\bm{q})\cdot 2m\xi^\dagger\xi
i M = − i e F 1 ( 0 ) ϕ ~ ( q ) ⋅ 2 m ξ † ξ
根据 Born 近似,这相当于是在一个势为 V ( x ) = e F 1 ( 0 ) ϕ ( x ) V(\bm{x}) = eF_1(0)\phi(\bm{x}) V ( x ) = e F 1 ( 0 ) ϕ ( x ) 的场中的散射。
类似的,我们可以计算电子在磁场中的散射。
此时四维矢势取为 A μ c l ( x ) = ( 0 , A c l ( x ) ) A_{\mu}^{cl}(x) = (0,\bm{A}^{cl}(\bm{x})) A μ c l ( x ) = ( 0 , A c l ( x ) ) ,跃迁矩阵元成为:
i M = i e ( u ˉ ( p ′ ) ( γ i F 1 + i σ i ν q ν 2 m F 2 ) u ( p ) ) A ~ c l i ( q ) i\mathcal{M} = ie(\bar{u}(p')(\gamma^iF_1 + \frac{i\sigma^{i\nu}q_\nu}{2m}F_2)u(p))\tilde{A}^i_{cl}(\bm{q})
i M = i e ( u ˉ ( p ′ ) ( γ i F 1 + 2 m i σ i ν q ν F 2 ) u ( p ) ) A ~ c l i ( q )
做近似:
u ( p ) = ( p ⋅ σ ξ p ⋅ σ ˉ ξ ) ≈ m ( ( 1 − p ⋅ σ / 2 m ) ξ ( 1 + p ⋅ σ / 2 m ) ξ ) u(p) = \begin{pmatrix}
\sqrt{p\cdot \sigma}\xi\\
\sqrt{p\cdot \bar{\sigma}}\xi\\
\end{pmatrix} \approx \sqrt{m} \begin{pmatrix}
(1-\bm{p}\cdot\bm{\sigma}/2m)\xi\\
(1+\bm{p}\cdot\bm{\sigma}/2m)\xi\\
\end{pmatrix}
u ( p ) = ( p ⋅ σ ξ p ⋅ σ ˉ ξ ) ≈ m ( ( 1 − p ⋅ σ / 2 m ) ξ ( 1 + p ⋅ σ / 2 m ) ξ )
得到 F 1 F_1 F 1 项可以写为:
u ˉ ( p ′ ) γ i u ( p ) = 2 m ξ ′ † ( p ′ ⋅ σ 2 m σ i + σ i p ⋅ σ 2 m ) ξ = 2 m ξ ′ † ( − i 2 m ϵ i j k q j σ k ) ξ \begin{aligned}
\bar{u}(p')\gamma^iu(p) &= 2m\xi'^\dagger(\frac{\bm{p}'\cdot\bm{\sigma}}{2m}\sigma^i+\sigma^i\frac{\bm{p}\cdot\bm{\sigma}}{2m})\xi\\
&= 2m\xi'^\dagger(-\frac{i}{2m}\epsilon^{ijk}q^j\sigma^k)\xi\\
\end{aligned}
u ˉ ( p ′ ) γ i u ( p ) = 2 m ξ ′ † ( 2 m p ′ ⋅ σ σ i + σ i 2 m p ⋅ σ ) ξ = 2 m ξ ′ † ( − 2 m i ϵ i j k q j σ k ) ξ
考虑到 F 2 F_2 F 2 项前的因子实际上可以写为:
u ˉ ( p ′ ) i σ i ν q ν 2 m u ( p ) = 2 m ξ ′ † ( − i 2 m ϵ i j k q j σ k ) ξ \bar{u}(p')\frac{i\sigma^{i\nu}q_\nu}{2m}u(p) = 2m\xi'^\dagger(-\frac{i}{2m}\epsilon^{ijk}q^j\sigma^k)\xi
u ˉ ( p ′ ) 2 m i σ i ν q ν u ( p ) = 2 m ξ ′ † ( − 2 m i ϵ i j k q j σ k ) ξ
可得:
i M = i e ( u ˉ ( p ′ ) ( γ i F 1 + i σ i ν q ν 2 m F 2 ) u ( p ) ) A ~ c l i ( q ) ≈ q → 0 i ( 2 m ) e ( ξ ′ † ( − i 2 m ϵ i j k q j σ k ) ( F 1 ( 0 ) + F 2 ( 0 ) ) ξ ) A ~ c l i ( q ) = − i ( 2 m ) e ( ξ ′ † − 1 2 m σ k ( F 1 ( 0 ) + F 2 ( 0 ) ) ξ ) B ~ k ( q ) \begin{aligned}
i\mathcal{M} &= ie(\bar{u}(p')(\gamma^iF_1 + \frac{i\sigma^{i\nu}q_\nu}{2m}F_2)u(p))\tilde{A}^i_{cl}(\bm{q})\\
& \underset{q\rightarrow 0}{\approx} i(2m)e(\xi'^\dagger(-\frac{i}{2m}\epsilon^{ijk}q^j\sigma^k)(F_1(0)+F_2(0))\xi)\tilde{A}^i_{cl}(\bm{q})\\
&=-i(2m)e(\xi'^\dagger\frac{-1}{2m}\sigma^k(F_1(0)+F_2(0))\xi)\tilde{B}^k(\bm{q})\\
\end{aligned}
i M = i e ( u ˉ ( p ′ ) ( γ i F 1 + 2 m i σ i ν q ν F 2 ) u ( p ) ) A ~ c l i ( q ) q → 0 ≈ i ( 2 m ) e ( ξ ′ † ( − 2 m i ϵ i j k q j σ k ) ( F 1 ( 0 ) + F 2 ( 0 ) ) ξ ) A ~ c l i ( q ) = − i ( 2 m ) e ( ξ ′ † 2 m − 1 σ k ( F 1 ( 0 ) + F 2 ( 0 ) ) ξ ) B ~ k ( q )
其中用到了傅里叶变换之后的磁场与矢势的关系:
B ~ k ( q ) = − i ϵ i j k q i A ~ c l j ( q ) \tilde{B}^k(\bm{q}) = -i\epsilon^{ijk}q^i\tilde{A}_{cl}^j(\bm{q})
B ~ k ( q ) = − i ϵ i j k q i A ~ c l j ( q )
利用 Born 近似,可以写出对应的散射势:
V ( x ) = − ⟨ μ ⟩ ⋅ B ( x ) ⟨ μ ⟩ = e m [ F 1 ( 0 ) + F 2 ( 0 ) ] ξ ′ † σ 2 ξ \begin{aligned}
&V(\bm{x}) = -\langle \bm{\mu} \rangle \cdot \bm{B}(\bm{x})\\
&\langle\bm{\mu}\rangle = \frac{e}{m}[F_1(0) + F_2(0)]\xi'^\dagger\frac{\bm{\sigma}}{2}\xi
\end{aligned}
V ( x ) = − ⟨ μ ⟩ ⋅ B ( x ) ⟨ μ ⟩ = m e [ F 1 ( 0 ) + F 2 ( 0 ) ] ξ ′ † 2 σ ξ
利用电子磁矩的表达式:
μ = g ( e 2 m ) S \bm{\mu} = g(\frac{e}{2m})\bm{S}
μ = g ( 2 m e ) S
得到 Lande g-factor 为:
g = 2 [ F 1 ( 0 ) + F 2 ( 0 ) ] = 2 + O ( α ) g = 2[F_1(0) + F_2(0)] = 2 +\mathcal{O}(\alpha)
g = 2 [ F 1 ( 0 ) + F 2 ( 0 ) ] = 2 + O ( α )
一阶圈图的计算
现在考虑一阶圈图的贡献:
u ˉ ( p ′ ) δ Γ μ ( p ′ , p ) u ( p ) = ∫ d 4 k ( 2 π ) 4 − i g ν ρ ( k − p ) 2 + i ϵ u ˉ ( p ′ ) ( − i e γ ν ) i ( γ ⋅ k ′ + m ) k ′ 2 − m 2 + i ϵ γ μ i ( γ ⋅ k + m ) k 2 − m 2 + i ϵ ( − i e γ ρ ) u ( p ) = 2 i e 2 ∫ d 4 k ( 2 π ) 4 u ˉ ( p ′ ) ( γ ⋅ k γ μ γ ⋅ k ′ + m 2 γ μ − 2 m ( k + k ′ ) μ ) u ( p ) ( ( k − p ) 2 + i ϵ ) ( k ′ 2 − m 2 + i ϵ ) ( k 2 − m 2 + i ϵ ) (4) \begin{aligned}
&\bar{u}(p')\delta\Gamma^{\mu}(p',p)u(p)\\
&= \int \frac{d^4k}{(2\pi)^4}\frac{-ig_{\nu\rho}}{(k-p)^2+i\epsilon}\bar{u}(p')(-ie\gamma^{\nu})\frac{i(\gamma\cdot k' +m)}{k'^2-m^2+i\epsilon} \gamma^{\mu} \frac{i(\gamma\cdot k +m)}{k^2-m^2+i\epsilon}(-ie\gamma^{\rho})u(p)\\
&= 2ie^2 \int \frac{d^4k}{(2\pi)^4} \frac{\bar{u}(p')(\gamma\cdot k\gamma^{\mu}\gamma\cdot k' + m^2 \gamma^{\mu} - 2m(k+k')^{\mu})u(p)}{((k-p)^2 + i\epsilon)(k'^2 -m^2 + i\epsilon)(k^2 -m^2 + i\epsilon)}
\end{aligned}\tag{4}
u ˉ ( p ′ ) δ Γ μ ( p ′ , p ) u ( p ) = ∫ ( 2 π ) 4 d 4 k ( k − p ) 2 + i ϵ − i g ν ρ u ˉ ( p ′ ) ( − i e γ ν ) k ′ 2 − m 2 + i ϵ i ( γ ⋅ k ′ + m ) γ μ k 2 − m 2 + i ϵ i ( γ ⋅ k + m ) ( − i e γ ρ ) u ( p ) = 2 i e 2 ∫ ( 2 π ) 4 d 4 k ( ( k − p ) 2 + i ϵ ) ( k ′ 2 − m 2 + i ϵ ) ( k 2 − m 2 + i ϵ ) u ˉ ( p ′ ) ( γ ⋅ k γ μ γ ⋅ k ′ + m 2 γ μ − 2 m ( k + k ′ ) μ ) u ( p ) ( 4 )
我们现在的任务成为了:对上述积分的计算。为了达成这一目的,我们接下来介绍 费曼参数化 Feynman parameterization 的方法。
我们先从如下公式开始:
1 A B = ∫ 0 1 d x 1 [ x A + ( 1 − x ) B ] 2 = ∫ 0 1 d x d y δ ( x + y − 1 ) 1 ( x A + y B ) 2 \begin{aligned}
\frac{1}{AB} &= \int_{0}^{1}dx \frac{1}{[xA+(1-x)B]^2}\\
&= \int_{0}^{1} dxdy \delta(x+y-1)\frac{1}{(xA+yB)^2}
\end{aligned}
A B 1 = ∫ 0 1 d x [ x A + ( 1 − x ) B ] 2 1 = ∫ 0 1 d x d y δ ( x + y − 1 ) ( x A + y B ) 2 1
这个公式是容易证明的:
∫ 0 1 d x 1 [ x A + ( 1 − x ) B ] 2 = 1 B − A 1 x A + ( 1 − x ) B ∣ 0 1 = 1 A B \begin{aligned}
\int_{0}^{1}dx \frac{1}{[xA+(1-x)B]^2} &=\frac{1}{B-A} \frac{1}{xA+(1-x)B}|_{0}^{1}\\
&= \frac{1}{AB}\\
\end{aligned}
∫ 0 1 d x [ x A + ( 1 − x ) B ] 2 1 = B − A 1 x A + ( 1 − x ) B 1 ∣ 0 1 = A B 1
将上式对 B B B 求偏导,得到更一般的式子:
1 A B n = ∫ 0 1 d x d y δ ( x + y − 1 ) n y n − 1 ( x A + y B ) n + 1 \frac{1}{AB^n} = \int_{0}^{1} dxdy \delta(x+y-1)\frac{ny^{n-1}}{(xA+yB)^{n+1}}
A B n 1 = ∫ 0 1 d x d y δ ( x + y − 1 ) ( x A + y B ) n + 1 n y n − 1
推广到多个因子的情形,可以使用数学归纳法推导:
1 A 1 A 2 ⋯ A n = ∫ 0 1 d x 1 d x 2 ⋯ d x n δ ( Σ x i − 1 ) ( n − 1 ) ! ( x 1 A 1 + x 2 A 2 + ⋯ + x n A n ) n \frac{1}{A_1A_2\cdots A_n} = \int_{0}^{1} dx_1dx_2\cdots dx_n \delta(\Sigma x_i-1)\frac{(n-1)!}{(x_1A_1+x_2A_2+\cdots+x_nA_n)^n}
A 1 A 2 ⋯ A n 1 = ∫ 0 1 d x 1 d x 2 ⋯ d x n δ ( Σ x i − 1 ) ( x 1 A 1 + x 2 A 2 + ⋯ + x n A n ) n ( n − 1 ) !
对 A i A_i A i 求 m i m_i m i 次偏导,得到以下一般式子:
1 A 1 m 1 A 2 m 2 ⋯ A n m n = ∫ 0 1 d x 1 ⋯ d x n δ ( Σ x i − 1 ) ∏ x i m i − 1 ( ∑ x i A i ) Σ m i Γ ( m 1 + ⋯ + m 2 ) Γ ( m 1 ) ⋯ Γ ( m n ) \frac{1}{A_1^{m_1}A_2^{m_2}\cdots A_n^{m_n}} = \int_{0}^{1}dx_1\cdots dx_n \delta(\Sigma x_i -1)\frac{\prod x_i^{m_i-1}}{(\sum x_iA_i)^{\Sigma m_i}}\frac{\Gamma(m_1+\cdots+m_2)}{\Gamma(m_1)\cdots \Gamma(m_n)}
A 1 m 1 A 2 m 2 ⋯ A n m n 1 = ∫ 0 1 d x 1 ⋯ d x n δ ( Σ x i − 1 ) ( ∑ x i A i ) Σ m i ∏ x i m i − 1 Γ ( m 1 ) ⋯ Γ ( m n ) Γ ( m 1 + ⋯ + m 2 )
这个公式在 m i m_i m i 不为整数时仍然成立。
因此,可以将圈图积分 ( 14 ) (14) ( 1 4 ) 的分母项写为:
1 ( ( k − p ) 2 + i ϵ ) ( k ′ 2 − m 2 + i ϵ ) ( k 2 − m 2 + i ϵ ) = ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) 2 D 3 \frac{1}{((k-p)^2+i\epsilon)(k'^2-m^2+i\epsilon)(k^2-m^2+i\epsilon)} = \int_{0}^{1}dxdydz \delta(x+y+z-1)\frac{2}{D^3}
( ( k − p ) 2 + i ϵ ) ( k ′ 2 − m 2 + i ϵ ) ( k 2 − m 2 + i ϵ ) 1 = ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) D 3 2
其中 D D D 为:
D = x ( k 2 − m 2 ) + y ( k ′ 2 − m 2 ) + z ( k − p ) 2 + ( x + y + z ) i ϵ = k 2 + 2 k ⋅ ( y q − z p ) + y q 2 + z p 2 − ( x + y ) m 2 + i ϵ \begin{aligned}
D &= x(k^2-m^2) + y(k'^2-m^2) + z(k-p)^2 + (x+y+z)i\epsilon\\
&= k^2 + 2k\cdot(yq-zp) + yq^2 +zp^2 - (x+y)m^2 + i\epsilon\\
\end{aligned}
D = x ( k 2 − m 2 ) + y ( k ′ 2 − m 2 ) + z ( k − p ) 2 + ( x + y + z ) i ϵ = k 2 + 2 k ⋅ ( y q − z p ) + y q 2 + z p 2 − ( x + y ) m 2 + i ϵ
其中已经代入 x + y + z = 1 , k ′ = k + q x+y+z =1,k' = k+q x + y + z = 1 , k ′ = k + q 。
作代换:l ≡ k + y q − z p l\equiv k+yq-zp l ≡ k + y q − z p
得到:
D = l 2 − Δ + i ϵ , Δ = − x y q 2 + ( 1 − z ) 2 m 2 D = l^2 - \Delta + i\epsilon,\quad \Delta = -xyq^2+(1-z)^2m^2
D = l 2 − Δ + i ϵ , Δ = − x y q 2 + ( 1 − z ) 2 m 2
主要到在 Δ \Delta Δ 项中,q 2 < 0 , Δ > 0 q^2<0,\Delta > 0 q 2 < 0 , Δ > 0 ,我们可以将其看作一个有效质量项。D D D 仅仅依赖于 l l l 的大小。根据对称性,注意到有以下式子成立:
∫ d 4 l ( 2 π ) 4 l μ D 3 = 0 ∫ d 4 l ( 2 π ) 4 l μ l ν D 3 = ∫ d 4 l ( 2 π ) 4 1 4 g μ ν l 2 D 3 \begin{aligned}
&\int \frac{d^4\mathcal{l}}{(2\pi)^4}\frac{\mathcal{l}^{\mu}}{D^3} = 0\\
&\int \frac{d^4\mathcal{l}}{(2\pi)^4}\frac{\mathcal{l}^{\mu}\mathcal{l}^{\nu}}{D^3} = \int \frac{d^4\mathcal{l}}{(2\pi)^4}\frac{\frac{1}{4}g^{\mu\nu}l^2}{D^3}\\
\end{aligned}
∫ ( 2 π ) 4 d 4 l D 3 l μ = 0 ∫ ( 2 π ) 4 d 4 l D 3 l μ l ν = ∫ ( 2 π ) 4 d 4 l D 3 4 1 g μ ν l 2
圈图积分的分子为:
u ˉ ( p ′ ) ( k γ μ k ′ + m 2 γ μ − 2 m ( k + k ′ ) μ ) u ( p ) = u ˉ ( p ′ ) [ − 1 2 γ μ l 2 + ( − y q + z p ) γ μ ( ( 1 − y ) q + z p ) ) + m 2 γ μ − 2 m ( ( 1 − 2 y ) q μ + 2 z p μ ) ] u ( p ) = u ˉ ( p ′ ) ( γ μ ( − 1 2 l 2 + ( 1 − x ) ( 1 − y ) q 2 + ( 1 − 2 z − z 2 ) m 2 ) + ( p ′ μ + p μ ) ⋅ m z ( z − 1 ) + q μ ⋅ m ( z − 2 ) ( x − y ) ) u ( p ) \begin{aligned}
&\bar{u}(p')(\cancel{k}\gamma^{\mu}\cancel{k}' + m^2\gamma^{\mu}- 2m(k+k')^{\mu})u(p)\\
&= \bar{u}(p')[-\frac{1}{2}\gamma^{\mu}l^2 + (-y\cancel{q}+z\cancel{p})\gamma^{\mu}((1-y)\cancel{q} + z\cancel{p}))\\
&\qquad + m^2\gamma^{\mu} - 2m((1-2y)q^{\mu}+2zp^{\mu})]u(p)\\
&= \bar{u}(p')(\gamma^{\mu}(-\frac{1}{2}l^2+(1-x)(1-y)q^2 + (1-2z-z^2)m^2)\\
&\qquad +(p'^{\mu} + p^{\mu})\cdot mz(z-1) + q^{\mu}\cdot m(z-2)(x-y) )u(p)\\
\end{aligned}
u ˉ ( p ′ ) ( k γ μ k ′ + m 2 γ μ − 2 m ( k + k ′ ) μ ) u ( p ) = u ˉ ( p ′ ) [ − 2 1 γ μ l 2 + ( − y q + z p ) γ μ ( ( 1 − y ) q + z p ) ) + m 2 γ μ − 2 m ( ( 1 − 2 y ) q μ + 2 z p μ ) ] u ( p ) = u ˉ ( p ′ ) ( γ μ ( − 2 1 l 2 + ( 1 − x ) ( 1 − y ) q 2 + ( 1 − 2 z − z 2 ) m 2 ) + ( p ′ μ + p μ ) ⋅ m z ( z − 1 ) + q μ ⋅ m ( z − 2 ) ( x − y ) ) u ( p )
最终一阶圈图的贡献可以写为:
u ˉ ( p ′ ) δ Γ μ ( p ′ , p ) u ( p ) = 2 i e 2 ∫ d 4 l ( 2 π ) 4 ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) 2 D 3 × u ˉ ( p ′ ) ( γ μ ( − 1 2 l 2 + ( 1 − x ) ( 1 − y ) q 2 + ( 1 − 4 z + z 2 ) m 2 ) i σ μ ν q ν 2 m ( 2 m 2 z ( 1 − z ) ) ) u ( p ) \begin{aligned}
\bar{u}(p')\delta\Gamma^{\mu}(p',p)u(p) &= 2ie^2\int \frac{d^4l}{(2\pi)^4}\int_{0}^{1}dxdydz \delta(x+y+z-1)\frac{2}{D^3}\\
&\qquad \times \bar{u}(p')(\gamma^{\mu}(-\frac{1}{2}l^2+(1-x)(1-y)q^2 + (1-4z+z^2)m^2)\\
&\qquad \frac{i\sigma^{\mu\nu}q_{\nu}}{2m}(2m^2z(1-z)))u(p)\\
\end{aligned}
u ˉ ( p ′ ) δ Γ μ ( p ′ , p ) u ( p ) = 2 i e 2 ∫ ( 2 π ) 4 d 4 l ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) D 3 2 × u ˉ ( p ′ ) ( γ μ ( − 2 1 l 2 + ( 1 − x ) ( 1 − y ) q 2 + ( 1 − 4 z + z 2 ) m 2 ) 2 m i σ μ ν q ν ( 2 m 2 z ( 1 − z ) ) ) u ( p )
现在我们需要计算对 l 0 l^0 l 0 的积分。其极点为:± ( ∣ l ∣ 2 + Δ − i ϵ ) \pm(\sqrt{|\bm{l}|^2 + \Delta} - i\epsilon) ± ( ∣ l ∣ 2 + Δ − i ϵ ) 。然后再在球坐标下对空间求积分。不过这里,我们采用一个更为简单的方法进行计算。我们观察对 l 0 l^0 l 0 的围线积分。考虑被积函数无穷远围线的贡献为零。因此,我们可以将整个积分围线逆时针旋转 90 ° 90\degree 9 0 ° 。如此我们可以定义 Euclidean 四维变量 l E l_E l E :
l 0 ≡ i l E 0 ; l = l E l^0 \equiv il_E^0;\quad \bm{l} = \bm{l}_E
l 0 ≡ i l E 0 ; l = l E
如此:l 0 l^0 l 0 积分上下限 ± i ∞ \pm i\infty ± i ∞ 对应于 l E 0 l_E^0 l E 0 的积分上下限 ± ∞ \pm\infty ± ∞ 。由此,我们可以使用四维球坐标计算上述积分。这种技巧称为 wick rotation 。
我们先计算下述积分:
∫ d 4 l ( 2 π ) 4 1 ∣ l 2 − Δ ∣ m = i ( − 1 ) m 1 ( 2 π ) 4 ∫ d 4 l E 1 ∣ l E 2 + Δ ∣ m = i ( − 1 ) m ( 2 π ) 4 ∫ d Ω 4 ∫ 0 ∞ d l E l E 3 ∣ l E 2 + Δ ∣ m \begin{aligned}
\int \frac{d^4 l}{(2\pi)^4}\frac{1}{|l^2-\Delta|^m} &= \frac{i}{(-1)^m}\frac{1}{(2\pi)^4}\int d^4l_E\frac{1}{|l_E^2+\Delta|^m}\\
&= \frac{i(-1)^m}{(2\pi)^4}\int d\Omega_4 \int_{0}^{\infty}dl_E \frac{l_E^3}{|l_E^2 + \Delta|^m}
\end{aligned}
∫ ( 2 π ) 4 d 4 l ∣ l 2 − Δ ∣ m 1 = ( − 1 ) m i ( 2 π ) 4 1 ∫ d 4 l E ∣ l E 2 + Δ ∣ m 1 = ( 2 π ) 4 i ( − 1 ) m ∫ d Ω 4 ∫ 0 ∞ d l E ∣ l E 2 + Δ ∣ m l E 3
其中 ∫ d Ω 4 \int d\Omega_4 ∫ d Ω 4 表示在一个四维球面上做积分。总的立体角为 2 π 2 2\pi^2 2 π 2 。我们引入四维球坐标:
x = ( r sin ω sin θ cos ϕ , r sin ω sin θ sin ϕ , r sin ω cos θ , r cos ω ) ω , θ ∈ [ 0 , π ] , ϕ ∈ [ 0 , 2 π ) \begin{aligned}
&x = (r\sin\omega\sin\theta\cos\phi, r\sin\omega\sin\theta\sin\phi,r\sin\omega\cos\theta,r\cos\omega)\\
& \omega,\theta \in [0,\pi],\phi \in [0,2\pi)
\end{aligned}
x = ( r sin ω sin θ cos ϕ , r sin ω sin θ sin ϕ , r sin ω cos θ , r cos ω ) ω , θ ∈ [ 0 , π ] , ϕ ∈ [ 0 , 2 π )
积分体积元成为:
d 4 x = r 3 sin 2 ω sin θ d ϕ d θ d ω d r d^4x = r^3\sin^2\omega\sin\theta d\phi d\theta d\omega dr
d 4 x = r 3 sin 2 ω sin θ d ϕ d θ d ω d r
那么上述积分很容易计算:
i ( − 1 ) m ( 2 π ) 4 ∫ d Ω 4 ∫ 0 ∞ d l E l E 3 ∣ l E 2 + Δ ∣ m = i ( − 1 ) m ( 2 π ) 4 ⋅ 2 π 2 ⋅ ∫ 0 ∞ r 3 d r ∣ r 2 + Δ ∣ m = i ( − 1 ) m 8 π 2 ⋅ ∫ 0 ∞ ( r 2 + Δ − Δ ) d ( r 2 + Δ ) 2 ( r 2 + Δ ) m = i ( − 1 ) m 16 π 2 ( − 1 m − 2 1 ( r 2 + Δ ) m − 2 + Δ m − 1 1 ( r 2 + Δ ) m − 1 ) ∣ 0 ∞ = i ( − 1 ) m ( 4 π ) 2 1 ( m − 1 ) ( m − 2 ) 1 Δ m − 2 \begin{aligned}
&\frac{i(-1)^m}{(2\pi)^4}\int d\Omega_4 \int_{0}^{\infty}dl_E \frac{l_E^3}{|l_E^2 + \Delta|^m}\\
&= \frac{i(-1)^m}{(2\pi)^4}\cdot 2\pi^2 \cdot \int_{0}^{\infty} \frac{r^3dr}{|r^2+\Delta|^m}\\
&= \frac{i(-1)^m}{8\pi^2} \cdot \int_{0}^{\infty} \frac{(r^2+\Delta-\Delta)d(r^2+\Delta)}{2(r^2+\Delta)^m}\\
&= \frac{i(-1)^m}{16\pi^2} (\frac{-1}{m-2}\frac{1}{(r^2+\Delta)^{m-2}} + \frac{\Delta}{m-1}\frac{1}{(r^2+\Delta)^{m-1}})|_{0}^{\infty}\\
&= \frac{i(-1)^m}{(4\pi)^2}\frac{1}{(m-1)(m-2)}\frac{1}{\Delta^{m-2}}
\end{aligned}
( 2 π ) 4 i ( − 1 ) m ∫ d Ω 4 ∫ 0 ∞ d l E ∣ l E 2 + Δ ∣ m l E 3 = ( 2 π ) 4 i ( − 1 ) m ⋅ 2 π 2 ⋅ ∫ 0 ∞ ∣ r 2 + Δ ∣ m r 3 d r = 8 π 2 i ( − 1 ) m ⋅ ∫ 0 ∞ 2 ( r 2 + Δ ) m ( r 2 + Δ − Δ ) d ( r 2 + Δ ) = 1 6 π 2 i ( − 1 ) m ( m − 2 − 1 ( r 2 + Δ ) m − 2 1 + m − 1 Δ ( r 2 + Δ ) m − 1 1 ) ∣ 0 ∞ = ( 4 π ) 2 i ( − 1 ) m ( m − 1 ) ( m − 2 ) 1 Δ m − 2 1
类似的有:
∫ d 4 l ( 2 π ) 4 l 2 ∣ l 2 − Δ ∣ m = i ( − 1 ) m − 1 ( 4 π ) 2 2 ( m − 1 ) ( m − 2 ) ( m − 3 ) 1 Δ m − 3 \int \frac{d^4 l}{(2\pi)^4}\frac{l^2}{|l^2-\Delta|^m} = \frac{i(-1)^{m-1}}{(4\pi)^2}\frac{2}{(m-1)(m-2)(m-3)}\frac{1}{\Delta^{m-3}}
∫ ( 2 π ) 4 d 4 l ∣ l 2 − Δ ∣ m l 2 = ( 4 π ) 2 i ( − 1 ) m − 1 ( m − 1 ) ( m − 2 ) ( m − 3 ) 2 Δ m − 3 1
这个积分适用条件是 m > 3 m>3 m > 3 。对于 m = 3 m=3 m = 3 ,wick rotation 不适用,积分将发散。但我们发现这正是我们所考虑的圈积分所对应的情形!为了解决这个问题,我们必须引入某种截断,因为场论本身也只是一个低能时的有效理论,那些高能区的物理将进入到与截断相关的耦合系数中去。我们考虑将其中的光子传播子进行替换:
1 ( k − p ) 2 + i ϵ → 1 ( k − p ) 2 + i ϵ − 1 ( k − p ) 2 − Λ 2 + i ϵ \frac{1}{(k-p)^2 +i\epsilon} \rightarrow \frac{1}{(k-p)^2 +i\epsilon}-\frac{1}{(k-p)^2-\Lambda^2 +i\epsilon}
( k − p ) 2 + i ϵ 1 → ( k − p ) 2 + i ϵ 1 − ( k − p ) 2 − Λ 2 + i ϵ 1
其中 Λ \Lambda Λ 为一个很大的质量。我们添加了一项假想的重光子传播子。那么这个积分对于小动量不会有影响,而会在大动量 k ≳ Λ k\gtrsim \Lambda k ≳ Λ 处截断。
对于包含重光子的圈积分,分子不需要进行修改,分母只需要进行如下替换:
Δ ⟶ Δ Λ = − x y q 2 + ( 1 − z ) 2 m 2 + z Λ 2 \Delta \longrightarrow \Delta_{\Lambda} = -xyq^2 + (1-z)^2m^2 + z\Lambda^2
Δ ⟶ Δ Λ = − x y q 2 + ( 1 − z ) 2 m 2 + z Λ 2
由此:
∫ d 4 l ( 2 π ) 3 ( l 2 ∣ l 2 − Δ ∣ 3 − l 2 ∣ l 2 − Δ Λ ∣ 3 ) = i ( 4 π ) 2 ∫ 0 ∞ d l E 2 ( l E 4 ∣ l E 2 + Δ ∣ 3 − l E 4 ∣ l E 2 + Δ Λ ∣ 3 ) = i ( 4 π ) 2 log ( Δ Λ Δ ) \begin{aligned}
\int \frac{d^4l}{(2\pi)^3}(\frac{l^2}{|l^2-\Delta|^3}-\frac{l^2}{|l^2-\Delta_{\Lambda}|^3}) &= \frac{i}{(4\pi)^2}\int_{0}^{\infty}dl_E^2 (\frac{l_E^4}{|l^2_E+\Delta|^3}-\frac{l_E^4}{|l_E^2+\Delta_{\Lambda}|^3})\\
&= \frac{i}{(4\pi)^2}\log(\frac{\Delta_{\Lambda}}{\Delta})
\end{aligned}
∫ ( 2 π ) 3 d 4 l ( ∣ l 2 − Δ ∣ 3 l 2 − ∣ l 2 − Δ Λ ∣ 3 l 2 ) = ( 4 π ) 2 i ∫ 0 ∞ d l E 2 ( ∣ l E 2 + Δ ∣ 3 l E 4 − ∣ l E 2 + Δ Λ ∣ 3 l E 4 ) = ( 4 π ) 2 i log ( Δ Δ Λ )
这种引入假想的重光子传播子的方法称为 Pauli-Villars 正规化(Pauli-Villars regularization) 。
那么最终得到的一阶圈图修正利用以上公式可以写为:
α 2 π ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) × u ˉ ( p ′ ) ( γ μ [ log z Λ 2 Δ + 1 Δ ( ( 1 − x ) ( 1 − y ) q 2 + ( 1 − 4 z + z 2 ) m 2 ] + i σ μ ν q ν 2 m [ 1 Δ 2 m 2 z ( 1 − z ) ] ) u ( p ) \begin{aligned}
&\frac{\alpha}{2\pi} \int_{0}^{1}dxdydz\delta(x+y+z-1)\\
&\times \bar{u}(p')(\gamma^{\mu}[\log \frac{z\Lambda^2}{\Delta} +\frac{1}{\Delta}((1-x)(1-y)q^2 + (1-4z+z^2)m^2]\\
&+ \frac{i\sigma^{\mu\nu}q_{\nu}}{2m}[\frac{1}{\Delta}2m^2z(1-z)])u(p)\\
\end{aligned}
2 π α ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) × u ˉ ( p ′ ) ( γ μ [ log Δ z Λ 2 + Δ 1 ( ( 1 − x ) ( 1 − y ) q 2 + ( 1 − 4 z + z 2 ) m 2 ] + 2 m i σ μ ν q ν [ Δ 1 2 m 2 z ( 1 − z ) ] ) u ( p )
其中方括号内的就是所求的形状因子。但实际上,这个式子仍然是存在红外发散的(后面会详细说明),我们还要为光子引入一个不为零的小质量以构成红外截断。这意味着,对于 1 / Δ 1/\Delta 1 / Δ 项,需要作替换:
( k − p ) 2 → ( k − p ) 2 − μ 2 (k-p)^2 \rightarrow (k-p)^2 -\mu^2
( k − p ) 2 → ( k − p ) 2 − μ 2
其中 μ \mu μ 是一个小质量。
如此形状因子成为:
F 1 ( q 2 ) = 1 + α 2 π ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) × [ log ( m 2 ( 1 − z ) 2 m 2 ( 1 − z ) 2 − q 2 x y ) + m 2 ( 1 − 4 z + z 2 ) + q 2 ( 1 − x ) ( 1 − y ) m 2 ( 1 − z ) 2 − q 2 x y + μ 2 z ] − m 2 ( 1 − 4 z + z 2 ) m 2 ( 1 − z ) 2 + μ 2 z ) + O ( α 2 ) F 2 ( q 2 ) = α 2 π ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) ( 2 m 2 z ( 1 − z ) m 2 ( ( 1 − z ) 2 − q 2 x y ) ) + O ( α 2 ) \begin{aligned}
F_1(q^2) =& 1 + \frac{\alpha}{2\pi}\int_{0}^{1}dxdydz\delta(x+y+z-1)\\
&\times [\log(\frac{m^2(1-z)^2}{m^2(1-z)^2 - q^2xy}) + \frac{m^2(1-4z+z^2)+q^2(1-x)(1-y)}{m^2(1-z)^2 - q^2xy + \mu^2z}]\\
&- \frac{m^2(1-4z+z^2)}{m^2(1-z)^2+\mu^2z}) + \mathcal{O}(\alpha^2)\\
&\\
F_2(q^2) &= \frac{\alpha}{2\pi}\int_{0}^{1}dxdydz\delta(x+y+z-1)(\frac{2m^2z(1-z)}{m^2((1-z)^2 -q^2xy)}) + \mathcal{O}(\alpha^2)\\
\end{aligned}
F 1 ( q 2 ) = F 2 ( q 2 ) 1 + 2 π α ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) × [ log ( m 2 ( 1 − z ) 2 − q 2 x y m 2 ( 1 − z ) 2 ) + m 2 ( 1 − z ) 2 − q 2 x y + μ 2 z m 2 ( 1 − 4 z + z 2 ) + q 2 ( 1 − x ) ( 1 − y ) ] − m 2 ( 1 − z ) 2 + μ 2 z m 2 ( 1 − 4 z + z 2 ) ) + O ( α 2 ) = 2 π α ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) ( m 2 ( ( 1 − z ) 2 − q 2 x y ) 2 m 2 z ( 1 − z ) ) + O ( α 2 )
注意到红外或紫外发散并不影响 F 2 ( q 2 ) F_2(q^2) F 2 ( q 2 ) ,其中:
F 2 ( q 2 = 0 ) = α 2 π ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) ( 2 m 2 z ( 1 − z ) m 2 ( ( 1 − z ) 2 ) ) = α π ∫ 0 1 d z ∫ 0 1 − z d y z 1 − z = α 2 π \begin{aligned}
F_2(q^2 = 0) &= \frac{\alpha}{2\pi}\int_{0}^{1}dxdydz\delta(x+y+z-1)(\frac{2m^2z(1-z)}{m^2((1-z)^2)})\\
&= \frac{\alpha}{\pi}\int_{0}^{1}dz\int_{0}^{1-z}dy \frac{z}{1-z}\\
&= \frac{\alpha}{2\pi}\\
\end{aligned}
F 2 ( q 2 = 0 ) = 2 π α ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) ( m 2 ( ( 1 − z ) 2 ) 2 m 2 z ( 1 − z ) ) = π α ∫ 0 1 d z ∫ 0 1 − z d y 1 − z z = 2 π α
得到一阶圈图对 g-factor 的修正为:
a e ≡ g − 2 2 = α 2 π ≈ 0.0011614 a_{e} \equiv \frac{g-2}{2} = \frac{\alpha}{2\pi} \approx 0.0011614
a e ≡ 2 g − 2 = 2 π α ≈ 0 . 0 0 1 1 6 1 4
对比实验值 a e = 0.0011597 a_{e} = 0.0011597 a e = 0 . 0 0 1 1 5 9 7 ,两者已经很接近了。我们可以考虑更高阶的圈图,使得计算更为精确。
红外发散
现在来考虑 F 1 ( q 2 ) F_1(q^2) F 1 ( q 2 ) 中产生红外发散的项。
F 1 ( q 2 ) ≈ α 2 π ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) [ m 2 ( 1 − 4 z + z 2 ) + q 2 ( 1 − x ) ( 1 − y ) m 2 ( 1 − z ) 2 − q 2 x y + μ 2 z − m 2 ( 1 − 4 z + z 2 ) m 2 ( 1 − z ) 2 + μ 2 z ) ] (5) \begin{aligned}
F_1(q^2) \approx& \frac{\alpha}{2\pi}\int_{0}^{1}dxdydz\delta(x+y+z-1)\\
&[\frac{m^2(1-4z+z^2)+q^2(1-x)(1-y)}{m^2(1-z)^2 - q^2xy + \mu^2z}- \frac{m^2(1-4z+z^2)}{m^2(1-z)^2+\mu^2z})]\\
\end{aligned}\tag{5}
F 1 ( q 2 ) ≈ 2 π α ∫ 0 1 d x d y d z δ ( x + y + z − 1 ) [ m 2 ( 1 − z ) 2 − q 2 x y + μ 2 z m 2 ( 1 − 4 z + z 2 ) + q 2 ( 1 − x ) ( 1 − y ) − m 2 ( 1 − z ) 2 + μ 2 z m 2 ( 1 − 4 z + z 2 ) ) ] ( 5 )
在 μ → 0 \mu\rightarrow 0 μ → 0 时,分母在 z = 1 , x = y = 0 z = 1,x = y = 0 z = 1 , x = y = 0 时取得极值,该积分主要由这个极点附近的区域贡献。对于分子,我们直接令 z = 1 , x = y = 0 z = 1,x=y=0 z = 1 , x = y = 0 ,并且积掉 δ \delta δ 函数:
F 1 ( q 2 ) = α 2 π ∫ 0 1 d z ∫ 0 1 − z d y [ − 2 m 2 + q 2 m 2 ( 1 − z ) 2 − q 2 x y + μ 2 − − 2 m 2 m 2 ( 1 − z ) 2 + μ 2 ] (6) F_1(q^2) = \frac{\alpha}{2\pi}\int_0^1dz \int_{0}^{1-z}dy[\frac{-2m^2+q^2}{m^2(1-z)^2 - q^2xy + \mu^2}- \frac{-2m^2}{m^2(1-z)^2+\mu^2}]\tag{6}
F 1 ( q 2 ) = 2 π α ∫ 0 1 d z ∫ 0 1 − z d y [ m 2 ( 1 − z ) 2 − q 2 x y + μ 2 − 2 m 2 + q 2 − m 2 ( 1 − z ) 2 + μ 2 − 2 m 2 ] ( 6 )
作代换:
y = ( 1 − z ) ξ , w = 1 − z y = (1-z)\xi,\quad w=1-z
y = ( 1 − z ) ξ , w = 1 − z
( 6 ) (6) ( 6 ) 式成为:
F 1 ( q 2 ) = α 4 π ∫ 0 1 d ξ ∫ 0 1 d ( w 2 ) ( − 2 m 2 + q 2 ( m 2 − q 2 ( 1 − ξ ) ξ ) w 2 + μ 2 − − 2 m 2 m 2 w 2 + μ 2 ) = α 4 π ∫ 0 1 d ξ [ ( − 2 m 2 + q 2 m 2 − q 2 ξ ( 1 − ξ ) ) log ( m 2 − q 2 ξ ( 1 − ξ ) μ 2 ) + 2 log ( m 2 μ 2 ) ] (7) \begin{aligned}
F_1(q^2) &= \frac{\alpha}{4\pi}\int_0^1d\xi \int_{0}^{1}d(w^2)(\frac{-2m^2+q^2}{(m^2- q^2(1-\xi)\xi)w^2 + \mu^2}- \frac{-2m^2}{m^2w^2+\mu^2})\\
&= \frac{\alpha}{4\pi}\int_{0}^{1}d\xi[(\frac{-2m^2+q^2}{m^2-q^2\xi(1-\xi)})\log(\frac{m^2-q^2\xi(1-\xi)}{\mu^2}) + 2\log(\frac{m^2}{\mu^2})]
\end{aligned}\tag{7}
F 1 ( q 2 ) = 4 π α ∫ 0 1 d ξ ∫ 0 1 d ( w 2 ) ( ( m 2 − q 2 ( 1 − ξ ) ξ ) w 2 + μ 2 − 2 m 2 + q 2 − m 2 w 2 + μ 2 − 2 m 2 ) = 4 π α ∫ 0 1 d ξ [ ( m 2 − q 2 ξ ( 1 − ξ ) − 2 m 2 + q 2 ) log ( μ 2 m 2 − q 2 ξ ( 1 − ξ ) ) + 2 log ( μ 2 m 2 ) ] ( 7 )
在 μ → 0 \mu\rightarrow 0 μ → 0 时,我们可以忽略对数中分子的细节。每一个正比于 m 2 m^2 m 2 或 − q 2 -q^2 − q 2 的项都是近似相同的。
F 1 ( q 2 ) = 1 − α 2 π f I R ( q 2 ) log ( − q 2 o r m 2 μ 2 ) + O ( α 2 ) \begin{aligned}
F_1(q^2) &= 1 - \frac{\alpha}{2\pi}f_{IR}(q^2)\log(\frac{-q^2\ or\ m^2}{\mu^2}) + \mathcal{O}(\alpha^2)\\
\end{aligned}
F 1 ( q 2 ) = 1 − 2 π α f I R ( q 2 ) log ( μ 2 − q 2 o r m 2 ) + O ( α 2 )
其中:
f I R ( q 2 ) = ∫ 0 1 ( m 2 − q 2 / 2 m 2 − q 2 ξ ( 1 − ξ ) ) d ξ − 1 f_{IR}(q^2) = \int_{0}^{1}(\frac{m^2 - q^2/2}{m^2 - q^2\xi(1-\xi)})d\xi - 1
f I R ( q 2 ) = ∫ 0 1 ( m 2 − q 2 ξ ( 1 − ξ ) m 2 − q 2 / 2 ) d ξ − 1
由于 ξ ( 1 − ξ ) \xi(1-\xi) ξ ( 1 − ξ ) 的最大值为 1 4 \frac{1}{4} 4 1 ,因此 f I R ( q 2 ) > 0 f_{IR}(q^2) > 0 f I R ( q 2 ) > 0 。
我们在作顶点修正时实际在作如下替换:
− e γ μ → − e γ μ F 1 ( q 2 ) -e\gamma^{\mu} \rightarrow -e\gamma^{\mu} F_1(q^2)
− e γ μ → − e γ μ F 1 ( q 2 )
这等价于将电子电量 e e e 替换成 e F 1 ( q 2 ) eF_1(q^2) e F 1 ( q 2 ) 。因此对散射截面的顶点修正成为:
d σ d Ω ≃ ( d σ d Ω ) 0 ⋅ [ 1 − α π f I R ( q 2 ) log ( − q 2 o r m 2 μ 2 ) + O ( α 2 ) ] (8) \begin{aligned}
\frac{d\sigma}{d\Omega} \simeq (\frac{d\sigma}{d\Omega})_0 \cdot [1 - \frac{\alpha}{\pi}f_{IR}(q^2)\log(\frac{-q^2\ or\ m^2}{\mu^2}) + \mathcal{O}(\alpha^2)]\\
\end{aligned}\tag{8}
d Ω d σ ≃ ( d Ω d σ ) 0 ⋅ [ 1 − π α f I R ( q 2 ) log ( μ 2 − q 2 o r m 2 ) + O ( α 2 ) ] ( 8 )
我们现在对 F 1 ( q 2 ) F_1(q^2) F 1 ( q 2 ) 进行一下估计,在 − q 2 → ∞ -q^2\rightarrow \infty − q 2 → ∞ 时,有:
f I R ( q 2 ) = ∫ 0 1 ( m 2 − q 2 / 2 m 2 − q 2 ξ ( 1 − ξ ) ) d ξ → 1 2 ∫ 0 1 − q 2 m 2 − q 2 ξ ( 1 − ξ ) d ξ ∼ 1 2 ( ∫ 0 − q 2 m 2 − q 2 ξ d ξ + ∫ 1 − q 2 m 2 − q 2 ( 1 − ξ ) d ξ ) ∼ log ( − q 2 m 2 ) \begin{aligned}
f_{IR}(q^2) &= \int_{0}^{1}(\frac{m^2 - q^2/2}{m^2 - q^2\xi(1-\xi)})d\xi\\
&\rightarrow \frac{1}{2}\int_{0}^{1}\frac{- q^2}{m^2 - q^2\xi(1-\xi)}d\xi\\
&\sim \frac{1}{2}(\int_{0}\frac{- q^2}{m^2 - q^2\xi}d\xi + \int^{1}\frac{- q^2}{m^2 - q^2(1-\xi)}d\xi)\\
&\sim \log(\frac{-q^2}{m^2})\\
\end{aligned}
f I R ( q 2 ) = ∫ 0 1 ( m 2 − q 2 ξ ( 1 − ξ ) m 2 − q 2 / 2 ) d ξ → 2 1 ∫ 0 1 m 2 − q 2 ξ ( 1 − ξ ) − q 2 d ξ ∼ 2 1 ( ∫ 0 m 2 − q 2 ξ − q 2 d ξ + ∫ 1 m 2 − q 2 ( 1 − ξ ) − q 2 d ξ ) ∼ log ( m 2 − q 2 )
代入 ( 8 ) (8) ( 8 ) 式,得到:
d σ d Ω ( p → p ′ ) = ( d σ d Ω ) 0 ⋅ [ 1 − α π log ( − q 2 m 2 ) log ( − q 2 μ 2 ) + O ( α 2 ) ] (9) \begin{aligned}
\frac{d\sigma}{d\Omega} (p\rightarrow p') = (\frac{d\sigma}{d\Omega})_0 \cdot [1 - \frac{\alpha}{\pi}\log(\frac{-q^2}{m^2})\log(\frac{-q^2}{\mu^2}) + \mathcal{O}(\alpha^2)]\\
\end{aligned}\tag{9}
d Ω d σ ( p → p ′ ) = ( d Ω d σ ) 0 ⋅ [ 1 − π α log ( m 2 − q 2 ) log ( μ 2 − q 2 ) + O ( α 2 ) ] ( 9 )
考虑上一篇计算出的轫致辐射的微分散射截面:
d σ d Ω ( p → p ′ + γ ) = ( d σ d Ω ) 0 ⋅ [ + α π log ( − q 2 m 2 ) log ( − q 2 μ 2 ) + O ( α 2 ) ] (10) \begin{aligned}
\frac{d\sigma}{d\Omega} (p\rightarrow p' + \gamma) = (\frac{d\sigma}{d\Omega})_0 \cdot [+\frac{\alpha}{\pi}\log(\frac{-q^2}{m^2})\log(\frac{-q^2}{\mu^2}) + \mathcal{O}(\alpha^2)]\\
\end{aligned}\tag{10}
d Ω d σ ( p → p ′ + γ ) = ( d Ω d σ ) 0 ⋅ [ + π α log ( m 2 − q 2 ) log ( μ 2 − q 2 ) + O ( α 2 ) ] ( 1 0 )
单单看顶点修正后的弹性散射微分散射截面或是轫致辐射的微分散射截面,它们都是发散的,因为 μ → 0 \mu\rightarrow 0 μ → 0 。但是它们的总和是收敛的。其中的物理原因是十分深刻的:因为我们不能单独观测到轫致辐射或是弹性散射。对于任何一个探测设备来说,其都存在一定的探测能量的下限 E l E_l E l 。如此,测量得到的弹性散射截面实际上是:
( d σ d Ω ) ( p → p ′ ) + ( d σ d Ω ) ( p → p ′ + γ ( k < E l ) ) ≡ ( d σ d Ω ) m e a s u r e d (\frac{d\sigma}{d\Omega})(p\rightarrow p') + (\frac{d\sigma}{d\Omega})(p\rightarrow p' + \gamma(k<E_l)) \equiv (\frac{d\sigma}{d\Omega})_{measured}
( d Ω d σ ) ( p → p ′ ) + ( d Ω d σ ) ( p → p ′ + γ ( k < E l ) ) ≡ ( d Ω d σ ) m e a s u r e d
这个微分散射截面为:
( d σ d Ω ) m e a s u r e d ≈ ( d σ d Ω ) 0 [ 1 − α π f I R ( q 2 ) log ( − q 2 o r m 2 μ 2 ) + α 2 π I ( v , v ′ ) log ( E l 2 μ 2 ) + O ( α 2 ) ] \begin{aligned}
(\frac{d\sigma}{d\Omega})_{measured} \approx (\frac{d\sigma}{d\Omega})_0[&1-\frac{\alpha}{\pi}f_{IR}(q^2)\log(\frac{-q^2\ or\ m^2}{\mu^2})\\
&+\frac{\alpha}{2\pi}\mathcal{I}(\bm{v},\bm{v}')\log(\frac{E_l^2}{\mu^2}) + \mathcal{O}(\alpha^2)]
\end{aligned}
( d Ω d σ ) m e a s u r e d ≈ ( d Ω d σ ) 0 [ 1 − π α f I R ( q 2 ) log ( μ 2 − q 2 o r m 2 ) + 2 π α I ( v , v ′ ) log ( μ 2 E l 2 ) + O ( α 2 ) ]
在 − q 2 ≫ m 2 -q^2\gg m^2 − q 2 ≫ m 2 时,有 I ( v , v ′ ) = 2 f I R ( q 2 ) \mathcal{I}(\bm{v},\bm{v}') = 2f_{IR}(q^2) I ( v , v ′ ) = 2 f I R ( q 2 ) ,此时上式成为:
( d σ d Ω ) m e a s u r e d ≈ ( d σ d Ω ) 0 [ 1 − α π f I R ( q 2 ) log ( − q 2 o r m 2 E l 2 ) + O ( α 2 ) ] (11) (\frac{d\sigma}{d\Omega})_{measured} \approx (\frac{d\sigma}{d\Omega})_0[1-\frac{\alpha}{\pi}f_{IR}(q^2)\log(\frac{-q^2\ or\ m^2}{E_l^2})+ \mathcal{O}(\alpha^2)]\tag{11}
( d Ω d σ ) m e a s u r e d ≈ ( d Ω d σ ) 0 [ 1 − π α f I R ( q 2 ) log ( E l 2 − q 2 o r m 2 ) + O ( α 2 ) ] ( 1 1 )
这将成为一个有限的项。
这里说明 I ( v , v ′ ) = 2 f I R ( q 2 ) \mathcal{I}(\bm{v},\bm{v}') = 2f_{IR}(q^2) I ( v , v ′ ) = 2 f I R ( q 2 ) :
I ( v , v ′ ) = ∫ d Ω k 4 π ( 2 p ⋅ p ′ ( k ^ ⋅ p ′ ) ( k ^ ⋅ p ) − m 2 ( k ^ ⋅ p ′ ) 2 − m 2 ( k ^ ⋅ p ) 2 ) \mathcal{I}(\bm{v},\bm{v}') = \int \frac{d\Omega_{\bm{k}}}{4\pi}(\frac{2p\cdot p'}{(\hat{k}\cdot p')(\hat{k}\cdot p)} - \frac{m^2}{(\hat{k}\cdot p')^2} - \frac{m^2}{(\hat{k}\cdot p)^2})
I ( v , v ′ ) = ∫ 4 π d Ω k ( ( k ^ ⋅ p ′ ) ( k ^ ⋅ p ) 2 p ⋅ p ′ − ( k ^ ⋅ p ′ ) 2 m 2 − ( k ^ ⋅ p ) 2 m 2 )
对于后两项,我们计算得到:
∫ d Ω k 4 π 1 ( k ^ ⋅ p ) 2 = 1 2 ∫ − 1 1 d cos θ 1 ( p 0 − p cos θ ) 2 = 1 2 p 1 p 0 − p cos θ ∣ − 1 1 = 1 m 2 \begin{aligned}
\int \frac{d\Omega_{\bm{k}}}{4\pi}\frac{1}{(\hat{k}\cdot p)^2} &= \frac{1}{2} \int_{-1}^{1}d\cos\theta \frac{1}{(p^0 - p\cos\theta)^2}\\
&= \frac{1}{2p} \frac{1}{p^0 - p\cos\theta}|_{-1}^{1} = \frac{1}{m^2}\\
\end{aligned}
∫ 4 π d Ω k ( k ^ ⋅ p ) 2 1 = 2 1 ∫ − 1 1 d cos θ ( p 0 − p cos θ ) 2 1 = 2 p 1 p 0 − p cos θ 1 ∣ − 1 1 = m 2 1
对于第一项,通过费曼参数化得到:
∫ d Ω k 4 π 2 p ⋅ p ′ ( k ^ ⋅ p ′ ) ( k ^ ⋅ p ) = ∫ 0 1 d ξ ∫ d Ω k 4 π 1 [ ξ k ^ ⋅ p ′ + ( 1 − ξ ) k ^ ⋅ p ] 2 = ∫ 0 1 d ξ ∫ d Ω k 4 π 1 [ k ^ ⋅ ( ξ p ′ + ( 1 − ξ ) p ) ] 2 = ∫ 0 1 d ξ 1 ( ξ p ′ + ( 1 − ξ ) p ) 2 = ∫ 0 1 d ξ 1 m 2 − ξ ( 1 − ξ ) q 2 \begin{aligned}
\int \frac{d\Omega_{\bm{k}}}{4\pi}\frac{2p\cdot p'}{(\hat{k}\cdot p')(\hat{k}\cdot p)} &= \int_{0}^{1}d\xi \int \frac{d\Omega_{\bm{k}}}{4\pi} \frac{1}{[\xi\hat{k}\cdot p' + (1-\xi)\hat{k}\cdot p]^2}\\
&= \int_{0}^{1}d\xi \int \frac{d\Omega_{\bm{k}}}{4\pi} \frac{1}{[\hat{k}\cdot(\xi p' + (1-\xi)p)]^2}\\
&= \int_{0}^{1}d\xi \frac{1}{(\xi p' + (1-\xi)p)^2}\\
&= \int_{0}^1d\xi \frac{1}{m^2-\xi(1-\xi)q^2}
\end{aligned}
∫ 4 π d Ω k ( k ^ ⋅ p ′ ) ( k ^ ⋅ p ) 2 p ⋅ p ′ = ∫ 0 1 d ξ ∫ 4 π d Ω k [ ξ k ^ ⋅ p ′ + ( 1 − ξ ) k ^ ⋅ p ] 2 1 = ∫ 0 1 d ξ ∫ 4 π d Ω k [ k ^ ⋅ ( ξ p ′ + ( 1 − ξ ) p ) ] 2 1 = ∫ 0 1 d ξ ( ξ p ′ + ( 1 − ξ ) p ) 2 1 = ∫ 0 1 d ξ m 2 − ξ ( 1 − ξ ) q 2 1
将这三项加起来就得到:
I ( v , v ′ ) = 2 f I R ( q 2 ) \mathcal{I}(\bm{v},\bm{v}') = 2f_{IR}(q^2)
I ( v , v ′ ) = 2 f I R ( q 2 )
这里仍然存在问题需要进行说明:
在超过领头阶的范围内,轫致辐射与电子弹性散射的散射截面中的发散能否相消呢?
如何解释出现在 ( 11 ) (11) ( 1 1 ) 式中出现的散射截面对能量下限 E l E_l E l 的依赖?
如何得到轫致辐射的粒子数?
对于这些问题,可以参考 an introduction to Quantum Field Theory,peskin 6.5 小节的讨论。下面仅仅展示最后得到的结果:对于发射 n n n 个 soft photons 的情况,且他们的能量介于 E − E_{-} E − 与 E + E_{+} E + 之间(高于 E + E_+ E + 的成为 hard photon)。那么我们需要将弹性散射截面乘以以下因子:
P r o b ( n γ w i t h E − < E < E + ) = 1 n ! [ α π f I R ( q 2 ) log ( E + 2 E − 2 ) ] 2 exp [ − α π f I R ( q 2 ) log ( E + 2 E − 2 ) ] Prob(n\gamma\ with\ E_{-}<E<E_{+}) = \frac{1}{n!}[\frac{\alpha}{\pi}f_{IR}(q^2)\log(\frac{E_{+}^2}{E_{-}^2})]^2\exp[-\frac{\alpha}{\pi}f_{IR}(q^2)\log(\frac{E_{+}^2}{E_{-}^2})]
P r o b ( n γ w i t h E − < E < E + ) = n ! 1 [ π α f I R ( q 2 ) log ( E − 2 E + 2 ) ] 2 exp [ − π α f I R ( q 2 ) log ( E − 2 E + 2 ) ]
平均辐射光子数为:
⟨ n ⟩ = α π log ( E + E − ) I ( v , v ′ ) \langle n\rangle = \frac{\alpha}{\pi}\log(\frac{E_{+}}{E_{-}})\mathcal{I}(\bm{v},\bm{v}')
⟨ n ⟩ = π α log ( E − E + ) I ( v , v ′ )