15.1

Problem 15.1

peskinChapter 15

习题 15.1

来源: 第15章, PDF第502页


15.1 Brute-force computations in SU(3)SU(3). The standard basis for the fundamental representation of SU(3)SU(3) is

t1=12(010100000),t2=12(0i0i00000),t3=12(100010000),t4=12(001000100),t5=12(00i000i00),t6=12(000001010),t7=12(00000i0i0),t8=123(100010002).\begin{aligned} t^1 &= \frac{1}{2} \begin{pmatrix} 0 & 1 & 0 \\ 1 & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix}, & t^2 &= \frac{1}{2} \begin{pmatrix} 0 & -i & 0 \\ i & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix}, & t^3 &= \frac{1}{2} \begin{pmatrix} 1 & 0 & 0 \\ 0 & -1 & 0 \\ 0 & 0 & 0 \end{pmatrix}, \\ t^4 &= \frac{1}{2} \begin{pmatrix} 0 & 0 & 1 \\ 0 & 0 & 0 \\ 1 & 0 & 0 \end{pmatrix}, & t^5 &= \frac{1}{2} \begin{pmatrix} 0 & 0 & -i \\ 0 & 0 & 0 \\ i & 0 & 0 \end{pmatrix}, \\ t^6 &= \frac{1}{2} \begin{pmatrix} 0 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & 1 & 0 \end{pmatrix}, & t^7 &= \frac{1}{2} \begin{pmatrix} 0 & 0 & 0 \\ 0 & 0 & -i \\ 0 & i & 0 \end{pmatrix}, & t^8 &= \frac{1}{2\sqrt{3}} \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & -2 \end{pmatrix}. \end{aligned}

(a) Explain why there are exactly eight matrices in the basis. (b) Evaluate all the commutators of these matrices, to determine the structure constants of SU(3)SU(3). Show that, with the normalizations used here, fabcf^{abc} is totally antisymmetric. (This exercise is tedious; you may wish to check only a representative sample of the commutators.) (c) Check the orthogonality condition (15.78), and evaluate the constant C(r)C(r) for this representation. (d) Compute the quadratic Casimir operator C2(r)C_2(r) directly from its definition (15.92), and verify the relation (15.94) between C2(r)C_2(r) and C(r)C(r).


Referenced Equations:

Equation (15.94):

d(r)C2(r)=d(G)C(r).(15.94)d(r) C_2(r) = d(G) C(r). \tag{15.94}

Referenced Equations:

Equation (15.78):

tr[tratrb]=C(r)δab,(15.78)\text{tr}[t_r^a t_r^b] = C(r) \delta^{ab} , \tag{15.78}

Equation (15.92):

tratra=C2(r)1,(15.92)t^a_r t^a_r = C_2(r) \cdot \mathbf{1}, \tag{15.92}

习题 15.1 - 解答


(a) SU(3)SU(3) 群生成元的数量分析

SU(3)SU(3) 群由所有行列式为 113×33 \times 3 幺正矩阵构成。任意一个 SU(3)SU(3) 矩阵 UU 都可以通过指数映射表示为 U=exp(iθata)U = \exp(i \theta^a t^a),其中 θa\theta^a 是实参数,tat^a 是群的生成元。

  1. 幺正性条件 (UU=1U^\dagger U = \mathbf{1}):要求生成元 tat^a 必须是厄米矩阵,即 (ta)=ta(t^a)^\dagger = t^a。一个一般的 3×33 \times 3 复矩阵有 99 个复数元素(1818 个实自由度)。厄米条件要求对角线元素为实数(33 个实自由度),且非对角线元素满足复共轭对称(33 个复数即 66 个实自由度),因此厄米矩阵共有 3+6=93 + 6 = 9 个独立的实参数。
  2. 单位行列式条件 (detU=1\det U = 1):利用恒等式 det(expA)=exp(trA)\det(\exp A) = \exp(\text{tr} A),要求生成元必须是无迹的,即 tr(ta)=0\text{tr}(t^a) = 0。这为对角线元素引入了 11 个实约束条件。

综合以上两个条件,独立的实参数个数为 91=89 - 1 = 8。因此,SU(3)SU(3) 的李代数 su(3)\mathfrak{su}(3) 的维数为 88,其基础表示的基底中恰好包含 88 个矩阵。

Dimension of SU(3)=8\boxed{ \text{Dimension of } SU(3) = 8 }

(b) 计算对易子、确定结构常数及全反对称性证明

SU(3)SU(3) 的李代数由对易关系 [ta,tb]=ifabctc[t^a, t^b] = i f^{abc} t^c 定义,其中 fabcf^{abc} 为结构常数。 首先证明在归一化条件 tr(tatb)=12δab\text{tr}(t^a t^b) = \frac{1}{2} \delta^{ab} 下,fabcf^{abc} 是全反对称的。将对易关系两边同乘 tdt^d 并取迹:

tr([ta,tb]td)=ifabctr(tctd)=ifabc(12δcd)=i2fabd\text{tr}([t^a, t^b] t^d) = i f^{abc} \text{tr}(t^c t^d) = i f^{abc} \left( \frac{1}{2} \delta^{cd} \right) = \frac{i}{2} f^{abd}

由此可得结构常数的表达式:

fabc=2itr([ta,tb]tc)=2itr(tatbtctbtatc)f^{abc} = -2i \text{tr}([t^a, t^b] t^c) = -2i \text{tr}(t^a t^b t^c - t^b t^a t^c)

利用矩阵迹的循环不变性 tr(tbtatc)=tr(tatctb)\text{tr}(t^b t^a t^c) = \text{tr}(t^a t^c t^b),可以得到:

fabc=2itr(ta[tb,tc])f^{abc} = -2i \text{tr}(t^a [t^b, t^c])

这表明 fabcf^{abc} 在指标的循环置换下是不变的,即 fabc=fbca=fcabf^{abc} = f^{bca} = f^{cab}。 同时,由对易子的反对称性 [ta,tb]=[tb,ta][t^a, t^b] = -[t^b, t^a] 可知,交换前两个指标会改变符号:fabc=fbacf^{abc} = -f^{bac}。 一个张量如果在前两个指标交换下反号,且在循环置换下不变,则它是全反对称的。

接下来通过直接计算代表性的对易子来确定非零的结构常数:

  • [t1,t2]=14[(010100000),(0i0i00000)]=i2(100010000)=it3    f123=1[t^1, t^2] = \frac{1}{4} \left[ \begin{pmatrix} 0 & 1 & 0 \\ 1 & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix}, \begin{pmatrix} 0 & -i & 0 \\ i & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix} \right] = \frac{i}{2} \begin{pmatrix} 1 & 0 & 0 \\ 0 & -1 & 0 \\ 0 & 0 & 0 \end{pmatrix} = i t^3 \implies f^{123} = 1
  • [t1,t4]=14[(010100000),(001000100)]=14(000001010)=i2t7    f147=12[t^1, t^4] = \frac{1}{4} \left[ \begin{pmatrix} 0 & 1 & 0 \\ 1 & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix}, \begin{pmatrix} 0 & 0 & 1 \\ 0 & 0 & 0 \\ 1 & 0 & 0 \end{pmatrix} \right] = \frac{1}{4} \begin{pmatrix} 0 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & -1 & 0 \end{pmatrix} = \frac{i}{2} t^7 \implies f^{147} = \frac{1}{2}
  • [t4,t5]=14[(001000100),(00i000i00)]=i2(100000001)=i2(t3+3t8)    f453=12,f458=32[t^4, t^5] = \frac{1}{4} \left[ \begin{pmatrix} 0 & 0 & 1 \\ 0 & 0 & 0 \\ 1 & 0 & 0 \end{pmatrix}, \begin{pmatrix} 0 & 0 & -i \\ 0 & 0 & 0 \\ i & 0 & 0 \end{pmatrix} \right] = \frac{i}{2} \begin{pmatrix} 1 & 0 & 0 \\ 0 & 0 & 0 \\ 0 & 0 & -1 \end{pmatrix} = \frac{i}{2} (t^3 + \sqrt{3} t^8) \implies f^{453} = \frac{1}{2}, f^{458} = \frac{\sqrt{3}}{2}

穷举所有组合,独立的非零结构常数如下:

f123=1f147=f246=f257=f345=12f156=f367=12f458=f678=32\boxed{ \begin{aligned} f^{123} &= 1 \\ f^{147} &= f^{246} = f^{257} = f^{345} = \frac{1}{2} \\ f^{156} &= f^{367} = -\frac{1}{2} \\ f^{458} &= f^{678} = \frac{\sqrt{3}}{2} \end{aligned} }

(c) 验证正交性条件并计算 C(r)C(r)

正交性条件为 tr[tratrb]=C(r)δab\text{tr}[t_r^a t_r^b] = C(r) \delta^{ab}。我们选取 a=b=1a=b=1a=b=8a=b=8 进行验证:

tr(t1t1)=14tr(100010000)=14(1+1)=12\text{tr}(t^1 t^1) = \frac{1}{4} \text{tr} \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 0 \end{pmatrix} = \frac{1}{4} (1 + 1) = \frac{1}{2}
tr(t8t8)=112tr(100010004)=112(1+1+4)=12\text{tr}(t^8 t^8) = \frac{1}{12} \text{tr} \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 4 \end{pmatrix} = \frac{1}{12} (1 + 1 + 4) = \frac{1}{2}

对于 aba \neq b,由于矩阵乘积的对角线元素之和均为零,迹显然为 00。因此正交性条件成立,且基础表示的常数 C(r)C(r) 为:

C(r)=12\boxed{ C(r) = \frac{1}{2} }

(d) 计算二次卡西米尔算符 C2(r)C_2(r) 并验证关系式

二次卡西米尔算符定义为 tratra=C2(r)1t^a_r t^a_r = C_2(r) \cdot \mathbf{1}。我们直接计算所有生成元的平方:

(t1)2=(t2)2=(t3)2=14diag(1,1,0)(t4)2=(t5)2=14diag(1,0,1)(t6)2=(t7)2=14diag(0,1,1)(t8)2=112diag(1,1,4)\begin{aligned} (t^1)^2 = (t^2)^2 = (t^3)^2 &= \frac{1}{4} \text{diag}(1, 1, 0) \\ (t^4)^2 = (t^5)^2 &= \frac{1}{4} \text{diag}(1, 0, 1) \\ (t^6)^2 = (t^7)^2 &= \frac{1}{4} \text{diag}(0, 1, 1) \\ (t^8)^2 &= \frac{1}{12} \text{diag}(1, 1, 4) \end{aligned}

将这 88 个矩阵相加:

a=18(ta)2=34(100010000)+24(100000001)+24(000010001)+112(100010004)=diag(34+24+112,  34+24+112,  24+24+412)=diag(1612,1612,1612)=431\begin{aligned} \sum_{a=1}^8 (t^a)^2 &= \frac{3}{4} \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 0 \end{pmatrix} + \frac{2}{4} \begin{pmatrix} 1 & 0 & 0 \\ 0 & 0 & 0 \\ 0 & 0 & 1 \end{pmatrix} + \frac{2}{4} \begin{pmatrix} 0 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{pmatrix} + \frac{1}{12} \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 4 \end{pmatrix} \\ &= \text{diag}\left( \frac{3}{4} + \frac{2}{4} + \frac{1}{12}, \; \frac{3}{4} + \frac{2}{4} + \frac{1}{12}, \; \frac{2}{4} + \frac{2}{4} + \frac{4}{12} \right) \\ &= \text{diag}\left( \frac{16}{12}, \frac{16}{12}, \frac{16}{12} \right) = \frac{4}{3} \mathbf{1} \end{aligned}

因此,基础表示的二次卡西米尔算符为:

C2(r)=43\boxed{ C_2(r) = \frac{4}{3} }

最后验证关系式 (15.94):d(r)C2(r)=d(G)C(r)d(r) C_2(r) = d(G) C(r)。 已知基础表示的维数 d(r)=3d(r) = 3,伴随表示的维数(生成元个数) d(G)=8d(G) = 8。代入已求得的常数:

LHS=d(r)C2(r)=3×43=4\text{LHS} = d(r) C_2(r) = 3 \times \frac{4}{3} = 4
RHS=d(G)C(r)=8×12=4\text{RHS} = d(G) C(r) = 8 \times \frac{1}{2} = 4

左右两边相等,关系式 d(r)C2(r)=d(G)C(r)\boxed{ d(r) C_2(r) = d(G) C(r) } 得到完美验证。

15.2

Problem 15.2

peskinChapter 15

习题 15.2

来源: 第15章, PDF第502页


15.2 Write down the basis matrices of the adjoint representation of SU(2)SU(2). Compute C(G)C(G) and C2(G)C_2(G) directly from their definitions (15.78) and (15.92).


Referenced Equations:

Equation (15.78):

tr[tratrb]=C(r)δab,(15.78)\text{tr}[t_r^a t_r^b] = C(r) \delta^{ab} , \tag{15.78}

Equation (15.92):

tratra=C2(r)1,(15.92)t^a_r t^a_r = C_2(r) \cdot \mathbf{1}, \tag{15.92}

习题 15.2 - 解答


对于 SU(2)SU(2) 群,其李代数的生成元满足对易关系 [Ta,Tb]=iϵabcTc[T^a, T^b] = i \epsilon^{abc} T^c,其中结构常数 fabc=ϵabcf^{abc} = \epsilon^{abc} 是完全反对称的列维-奇维塔符号(Levi-Civita symbol),且 ϵ123=1\epsilon^{123} = 1

在伴随表示(adjoint representation)中,生成元矩阵的维度等于群的生成元个数(对于 SU(2)SU(2) 为 3),其矩阵元素由结构常数直接给出:

(TGa)bc=ifabc=iϵabc(T^a_G)_{bc} = -i f^{abc} = -i \epsilon^{abc}

其中 a,b,c{1,2,3}a, b, c \in \{1, 2, 3\}

1. 伴随表示的基矩阵

根据上述定义,我们可以逐一写出这三个 3×33 \times 3 的基矩阵 TG1,TG2,TG3T^1_G, T^2_G, T^3_G

对于 a=1a=1

TG1=i(000001010)T^1_G = -i \begin{pmatrix} 0 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & -1 & 0 \end{pmatrix}

对于 a=2a=2

TG2=i(001000100)T^2_G = -i \begin{pmatrix} 0 & 0 & -1 \\ 0 & 0 & 0 \\ 1 & 0 & 0 \end{pmatrix}

对于 a=3a=3

TG3=i(010100000)T^3_G = -i \begin{pmatrix} 0 & 1 & 0 \\ -1 & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix}

2. 计算 C(G)C(G)

根据定义 (15.78),表示的指数(Dynkin index)C(r)C(r) 满足:

tr[TraTrb]=C(r)δab\text{tr}[T^a_r T^b_r] = C(r) \delta^{ab}

为了计算伴随表示的 C(G)C(G),我们只需取 a=b=1a=b=1 并计算其迹:

(TG1)2=(i)2(000001010)(000001010)=(000010001)=(000010001)(T^1_G)^2 = (-i)^2 \begin{pmatrix} 0 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & -1 & 0 \end{pmatrix} \begin{pmatrix} 0 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & -1 & 0 \end{pmatrix} = - \begin{pmatrix} 0 & 0 & 0 \\ 0 & -1 & 0 \\ 0 & 0 & -1 \end{pmatrix} = \begin{pmatrix} 0 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{pmatrix}

对其求迹:

tr[(TG1)2]=0+1+1=2\text{tr}[(T^1_G)^2] = 0 + 1 + 1 = 2

因此,由 tr[(TG1)2]=C(G)δ11=C(G)\text{tr}[(T^1_G)^2] = C(G) \delta^{11} = C(G),我们得到:

C(G)=2\boxed{C(G) = 2}

3. 计算 C2(G)C_2(G)

根据定义 (15.92),二次卡西米尔算子(quadratic Casimir operator)C2(r)C_2(r) 满足:

a=13TraTra=C2(r)1\sum_{a=1}^3 T^a_r T^a_r = C_2(r) \cdot \mathbf{1}

我们已经计算了 (TG1)2(T^1_G)^2。接下来计算 (TG2)2(T^2_G)^2(TG3)2(T^3_G)^2

(TG2)2=(i)2(001000100)2=(100000001)=(100000001)(T^2_G)^2 = (-i)^2 \begin{pmatrix} 0 & 0 & -1 \\ 0 & 0 & 0 \\ 1 & 0 & 0 \end{pmatrix}^2 = - \begin{pmatrix} -1 & 0 & 0 \\ 0 & 0 & 0 \\ 0 & 0 & -1 \end{pmatrix} = \begin{pmatrix} 1 & 0 & 0 \\ 0 & 0 & 0 \\ 0 & 0 & 1 \end{pmatrix}
(TG3)2=(i)2(010100000)2=(100010000)=(100010000)(T^3_G)^2 = (-i)^2 \begin{pmatrix} 0 & 1 & 0 \\ -1 & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix}^2 = - \begin{pmatrix} -1 & 0 & 0 \\ 0 & -1 & 0 \\ 0 & 0 & 0 \end{pmatrix} = \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 0 \end{pmatrix}

将这三个矩阵相加:

a=13(TGa)2=(000010001)+(100000001)+(100010000)=(200020002)=2(100010001)\sum_{a=1}^3 (T^a_G)^2 = \begin{pmatrix} 0 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{pmatrix} + \begin{pmatrix} 1 & 0 & 0 \\ 0 & 0 & 0 \\ 0 & 0 & 1 \end{pmatrix} + \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 0 \end{pmatrix} = \begin{pmatrix} 2 & 0 & 0 \\ 0 & 2 & 0 \\ 0 & 0 & 2 \end{pmatrix} = 2 \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{pmatrix}

a=13(TGa)2=213×3\sum_{a=1}^3 (T^a_G)^2 = 2 \cdot \mathbf{1}_{3 \times 3}。与定义对比,我们得到:

C2(G)=2\boxed{C_2(G) = 2}

(注:对于一般的 SU(N)SU(N) 群,伴随表示的 C(G)C(G)C2(G)C_2(G) 均等于 NN。本题中 N=2N=2,直接计算的结果与一般结论完全一致。)

15.3

Problem 15.3

peskinChapter 15

习题 15.3

来源: 第15章, PDF第503页


15.3 Coulomb potential.

(a) Using functional integration, compute the expectation value of the Wilson loop in pure quantum electrodynamics without fermions. Show that

UP(z,z)=exp[e2PdxμPdyνgμν18π2(xy)2],\langle U_P(z, z) \rangle = \exp \left[ -e^2 \oint_P dx^\mu \oint_P dy^\nu g_{\mu\nu} \frac{1}{8\pi^2(x - y)^2} \right],

with xx and yy integrated around the closed curve PP.

(b) Consider the Wilson loop of a rectangular path of (spacelike) width RR and (timelike) length T,TRT, T \gg R. Compute the expectation value of the Wilson loop in this limit and compare to the general expression for time evolution,

UP=exp[iE(R)T],\langle U_P \rangle = \exp[-iE(R)T],

where E(R)E(R) is the energy of the electromagnetic sources corresponding to the Wilson loop. Show that the potential energy of these sources is just the Coulomb potential, V(R)=e2/4πRV(R) = -e^2/4\pi R.

(c) Assuming that the propagator of the non-Abelian gauge field is given by the Feynman gauge expression

Aμa(x)Aνb(y)=d4p(2π)4igμνδabp2eip(xy),\langle A^a_\mu(x) A^b_\nu(y) \rangle = \int \frac{d^4p}{(2\pi)^4} \frac{-ig_{\mu\nu}\delta^{ab}}{p^2} e^{-ip \cdot (x-y)},

compute the expectation value of a non-Abelian Wilson loop to order g2g^2. The result will depend on the representation rr of the gauge group in which one chooses the matrices that appear in the exponential. Show that, to this order, the Coulomb potential of the non-Abelian gauge theory is V(R)=g2C2(r)/4πRV(R) = -g^2 C_2(r)/4\pi R.

习题 15.3 - 解答


Part (a)

The Wilson loop operator for a closed path PP in pure U(1) gauge theory (QED without fermions) is defined as: UP=exp(iePAμdxμ)U_P = \exp\left( -ie \oint_P A_\mu dx^\mu \right) Using functional integration, the expectation value of the Wilson loop is given by the path integral: UP=DAμeiS[A]exp(iePAμdxμ)DAμeiS[A]\langle U_P \rangle = \frac{\int \mathcal{D}A_\mu \, e^{iS[A]} \exp\left( -ie \oint_P A_\mu dx^\mu \right)}{\int \mathcal{D}A_\mu \, e^{iS[A]}} In the Feynman gauge, the action is quadratic in the gauge field: S[A]=d4x(12μAνμAν)S[A] = \int d^4x \left( -\frac{1}{2} \partial_\mu A_\nu \partial^\mu A^\nu \right). Thus, the path integral is a standard Gaussian integral with a source term Jμ(x)=ePdyμδ(4)(xy)J^\mu(x) = -e \oint_P dy^\mu \delta^{(4)}(x-y).

For a Gaussian functional distribution, the expectation value of an exponential is eiJA=e12(JA)2\langle e^{i \int J \cdot A} \rangle = e^{-\frac{1}{2} \langle (\int J \cdot A)^2 \rangle}. Therefore, we can directly exponentiate the two-point correlation function: UP=exp[e22PdxμPdyνTAμ(x)Aν(y)]\langle U_P \rangle = \exp\left[ -\frac{e^2}{2} \oint_P dx^\mu \oint_P dy^\nu \langle T A_\mu(x) A_\nu(y) \rangle \right] The position-space Feynman propagator for the massless photon is: TAμ(x)Aν(y)=d4p(2π)4igμνp2+iϵeip(xy)=gμν4π2(xy)2\langle T A_\mu(x) A_\nu(y) \rangle = \int \frac{d^4p}{(2\pi)^4} \frac{-ig_{\mu\nu}}{p^2 + i\epsilon} e^{-ip \cdot (x-y)} = \frac{g_{\mu\nu}}{4\pi^2 (x-y)^2} (where the iϵi\epsilon prescription is implicitly absorbed into the squared distance). Substituting this propagator into the exponent yields: UP=exp[e22PdxμPdyνgμν4π2(xy)2]\langle U_P \rangle = \exp\left[ -\frac{e^2}{2} \oint_P dx^\mu \oint_P dy^\nu \frac{g_{\mu\nu}}{4\pi^2(x-y)^2} \right] UP=exp[e2PdxμPdyνgμν18π2(xy)2]\boxed{ \langle U_P \rangle = \exp\left[ -e^2 \oint_P dx^\mu \oint_P dy^\nu g_{\mu\nu} \frac{1}{8\pi^2(x-y)^2} \right] }


Part (b)

We consider a rectangular Wilson loop with spacelike width RR and timelike length TT, where TRT \gg R. The path PP consists of four segments. The dominant contribution to the integral comes from the two long timelike segments, which we parameterize as:

  • Segment 1: xμ=(t,R/2)x^\mu = (t, \vec{R}/2), with dxμ=(dt,0)dx^\mu = (dt, \vec{0}) for t[T/2,T/2]t \in [-T/2, T/2].
  • Segment 3: yμ=(t,R/2)y^\mu = (t', -\vec{R}/2), with dyμ=(dt,0)dy^\mu = (dt', \vec{0}) for t[T/2,T/2]t' \in [T/2, -T/2] (note the reversed direction in time).

The self-interactions (Segment 1 with 1, and 3 with 3) yield divergent constants independent of RR, representing the self-energy of the point charges, which do not contribute to the inter-particle potential V(R)V(R). The RR-dependent potential arises from the cross-terms (Segment 1 with 3, and 3 with 1).

The cross-term contribution to the exponent is: W13+31=2e2Seg 1dxμSeg 3dyνgμν18π2(xy)2W_{13+31} = -2 e^2 \int_{\text{Seg 1}} dx^\mu \int_{\text{Seg 3}} dy^\nu g_{\mu\nu} \frac{1}{8\pi^2(x-y)^2} Since dxμdx^\mu and dyνdy^\nu only have non-zero time components, gμνdxμdyν=dtdtg_{\mu\nu} dx^\mu dy^\nu = dt \, dt'. The squared distance is (xy)2=(tt)2R2iϵ(x-y)^2 = (t-t')^2 - R^2 - i\epsilon. Taking into account the integration limits for Segment 3, we have: W13+31=2e2T/2T/2dtT/2T/2dt18π2((tt)2R2iϵ)W_{13+31} = -2 e^2 \int_{-T/2}^{T/2} dt \int_{T/2}^{-T/2} dt' \frac{1}{8\pi^2 ((t-t')^2 - R^2 - i\epsilon)} Reversing the limits of tt' introduces a minus sign: W13+31=2e2T/2T/2dtT/2T/2dt18π2((tt)2R2iϵ)W_{13+31} = 2 e^2 \int_{-T/2}^{T/2} dt \int_{-T/2}^{T/2} dt' \frac{1}{8\pi^2 ((t-t')^2 - R^2 - i\epsilon)} Change variables to the relative time τ=tt\tau = t - t' and average time s=(t+t)/2s = (t+t')/2. For TRT \gg R, the integral over ss gives TT, and the integral over τ\tau extends from -\infty to \infty: W13+31=e2T4π2dτ1τ2R2iϵW_{13+31} = \frac{e^2 T}{4\pi^2} \int_{-\infty}^{\infty} d\tau \frac{1}{\tau^2 - R^2 - i\epsilon} The integral over τ\tau can be evaluated using contour integration. The poles are at τ=±(R+iϵ)\tau = \pm (R + i\epsilon'). Closing the contour in the upper half-plane encloses the pole at τ=R+iϵ\tau = R + i\epsilon', giving: dττ2R2iϵ=2πi(12R)=iπR\int_{-\infty}^{\infty} \frac{d\tau}{\tau^2 - R^2 - i\epsilon} = 2\pi i \left( \frac{1}{2R} \right) = \frac{i\pi}{R} Thus, the exponent becomes: W13+31=e2T4π2(iπR)=ie24πRTW_{13+31} = \frac{e^2 T}{4\pi^2} \left( \frac{i\pi}{R} \right) = i \frac{e^2}{4\pi R} T The expectation value of the Wilson loop is UP=exp(ie24πRT)\langle U_P \rangle = \exp\left( i \frac{e^2}{4\pi R} T \right). Comparing this to the general expression for time evolution UP=exp[iE(R)T]\langle U_P \rangle = \exp[-i E(R) T], we find: iE(R)T=ie24πRT    E(R)=e24πR-i E(R) T = i \frac{e^2}{4\pi R} T \implies E(R) = -\frac{e^2}{4\pi R} V(R)=e24πR\boxed{ V(R) = -\frac{e^2}{4\pi R} } This is exactly the attractive Coulomb potential.


Part (c)

For a non-Abelian gauge theory, the Wilson loop in representation rr is defined as: UP=1d(r)trPexp(igPAμaTadxμ)\langle U_P \rangle = \frac{1}{d(r)} \left\langle \text{tr} \, \mathcal{P} \exp\left( ig \oint_P A_\mu^a T^a dx^\mu \right) \right\rangle Expanding the path-ordered exponential to order g2g^2: UP=1d(r)trI+igPAμaTadxμg2x>yAμa(x)Aνb(y)TaTbdxμdyν\langle U_P \rangle = \frac{1}{d(r)} \text{tr} \left\langle I + ig \oint_P A_\mu^a T^a dx^\mu - g^2 \oint_{x > y} A_\mu^a(x) A_\nu^b(y) T^a T^b dx^\mu dy^\nu \right\rangle The linear term vanishes because Aμa=0\langle A_\mu^a \rangle = 0. For the quadratic term, we use the given Feynman gauge propagator: Aμa(x)Aνb(y)=δabd4p(2π)4igμνp2+iϵeip(xy)δabDμν(xy)\langle A_\mu^a(x) A_\nu^b(y) \rangle = \delta^{ab} \int \frac{d^4p}{(2\pi)^4} \frac{-ig_{\mu\nu}}{p^2 + i\epsilon} e^{-ip \cdot (x-y)} \equiv \delta^{ab} D_{\mu\nu}(x-y) Substituting this into the expansion and using the Casimir operator TaTa=C2(r)IT^a T^a = C_2(r) I: UP=1g2d(r)tr(C2(r)I)x>ydxμdyνDμν(xy)\langle U_P \rangle = 1 - \frac{g^2}{d(r)} \text{tr} \left( C_2(r) I \right) \oint_{x > y} dx^\mu dy^\nu D_{\mu\nu}(x-y) Since tr(I)=d(r)\text{tr}(I) = d(r) and the propagator Dμν(xy)D_{\mu\nu}(x-y) is symmetric under the exchange of xx and yy, the path-ordered integral over x>yx > y is exactly half of the unconstrained double integral: UP=1g22C2(r)PPdxμdyνDμν(xy)+O(g4)\langle U_P \rangle = 1 - \frac{g^2}{2} C_2(r) \oint_P \oint_P dx^\mu dy^\nu D_{\mu\nu}(x-y) + \mathcal{O}(g^4) This matches the expansion of an exponential: UPexp[g22C2(r)PPdxμdyνd4p(2π)4igμνp2+iϵeip(xy)]\langle U_P \rangle \approx \exp\left[ -\frac{g^2}{2} C_2(r) \oint_P \oint_P dx^\mu dy^\nu \int \frac{d^4p}{(2\pi)^4} \frac{-ig_{\mu\nu}}{p^2 + i\epsilon} e^{-ip \cdot (x-y)} \right] We evaluate the exponent for the rectangular loop TRT \gg R as in part (b). The cross-terms between the two timelike segments give: W13+31=g2C2(r)T/2T/2dtT/2T/2dtd4p(2π)4ip02p2+iϵeip0(tt)eipRW_{13+31} = -g^2 C_2(r) \int_{-T/2}^{T/2} dt \int_{T/2}^{-T/2} dt' \int \frac{d^4p}{(2\pi)^4} \frac{-i}{p_0^2 - \vec{p}^2 + i\epsilon} e^{-ip_0(t-t')} e^{i\vec{p}\cdot\vec{R}} Reversing the tt' limits gives a plus sign. For large TT, the time integrals yield dtdteip0(tt)2πδ(p0)T\int dt \int dt' e^{-ip_0(t-t')} \approx 2\pi \delta(p_0) T. The exponent becomes: W13+31=g2C2(r)Td3p(2π)3ip2eipR=ig2C2(r)Td3p(2π)3eipRp2W_{13+31} = g^2 C_2(r) T \int \frac{d^3p}{(2\pi)^3} \frac{-i}{-\vec{p}^2} e^{i\vec{p}\cdot\vec{R}} = i g^2 C_2(r) T \int \frac{d^3p}{(2\pi)^3} \frac{e^{i\vec{p}\cdot\vec{R}}}{\vec{p}^2} Using the standard 3D Fourier transform d3p(2π)3eipRp2=14πR\int \frac{d^3p}{(2\pi)^3} \frac{e^{i\vec{p}\cdot\vec{R}}}{\vec{p}^2} = \frac{1}{4\pi R}, we obtain: W13+31=ig2C2(r)4πRTW_{13+31} = i \frac{g^2 C_2(r)}{4\pi R} T Thus, to order g2g^2, the expectation value is UP=exp(ig2C2(r)4πRT)\langle U_P \rangle = \exp\left( i \frac{g^2 C_2(r)}{4\pi R} T \right). Comparing this with UP=exp[iE(R)T]\langle U_P \rangle = \exp[-i E(R) T], we identify the potential energy: V(R)=g2C2(r)4πR\boxed{ V(R) = -\frac{g^2 C_2(r)}{4\pi R} }

15.4

Problem 15.4

peskinChapter 15

习题 15.4

来源: 第15章, PDF第503,504页


15.4 Scalar propagator in a gauge theory.

Consider the equation for the Green's function of the Klein-Gordon equation:

(2+m2)DF(x,y)=iδ(4)(xy).(\partial^2 + m^2) D_F(x, y) = -i\delta^{(4)}(x - y).

We can find an interesting representation for this Green's function by writing

DF(x,y)=0dTD(x,y,T),D_F(x, y) = \int_0^\infty dT D(x, y, T),

where D(x,y,T)D(x, y, T) satisfies the Schrödinger equation

[iT(2+m2)]D(x,y,T)=iδ(T)δ(4)(xy).\left[ i\frac{\partial}{\partial T} - (\partial^2 + m^2) \right] D(x, y, T) = i\delta(T)\delta^{(4)}(x - y).

Now, represent D(x,y,T)D(x, y, T) using the functional integral solution of the Schrödinger equation presented in Section 9.1.

(a) Using the explicit formula of the propagator of the Schrödinger equation, show that this integral formula gives the standard expression for the Feynman propagator.

(b) Using the method just described, show that the expression

DF(x,y)=0dTDxexp[idt12((dxμdt)2m2)iedtdxμdtAμ(x)]D_F(x, y) = \int_0^\infty dT \int \mathcal{D}x \exp \left[ i \int dt \frac{1}{2} \left( \left( \frac{dx^\mu}{dt} \right)^2 - m^2 \right) - ie \int dt \frac{dx^\mu}{dt} A_\mu(x) \right]

is a functional integral representation for the scalar field propagator in an arbitrary background electromagnetic field. Show, in particular, that the functional integral satisfies the relevant Schrödinger equation. Notice that this integral depends on AμA_\mu through the Wilson line.

(c) Generalize this expression to a non-Abelian gauge theory. Show that the functional integral solves the relevant Schrödinger equation only if the group matrices in the exponential for the Wilson line are path-ordered.

习题 15.4 - 解答


(a) 首先,考虑自由标量场的薛定谔方程:

[iT(2+m2)]D(x,y,T)=iδ(T)δ(4)(xy)\left[ i\frac{\partial}{\partial T} - (\partial^2 + m^2) \right] D(x, y, T) = i\delta(T)\delta^{(4)}(x - y)

在四维闵可夫斯基时空中,该方程描述了一个“质量”为 1/21/2 的非相对论粒子在固有时间 TT 下的演化(注意这里的度规符号差和算符定义)。其在坐标空间的标准传播子解(即热核或薛定谔核)为:

D(x,y,T)=θ(T)i(4πiT)2exp[i(xy)24Tim2T]D(x, y, T) = \theta(T) \frac{-i}{(4\pi i T)^2} \exp\left[ \frac{i(x-y)^2}{4T} - im^2 T \right]

将此表达式代入 DF(x,y)D_F(x, y) 的积分公式中:

DF(x,y)=0dTi(4πiT)2exp[i(xy)24Tim2T]D_F(x, y) = \int_0^\infty dT \frac{-i}{(4\pi i T)^2} \exp\left[ \frac{i(x-y)^2}{4T} - im^2 T \right]

为了计算这个积分,我们可以引入一个微小的正虚部 m2m2iϵm^2 \to m^2 - i\epsilon 以保证积分在 TT \to \infty 时收敛。利用标准的 Schwinger 固有时(proper time)积分公式,或者将其转换到动量空间:

D(x,y,T)=θ(T)d4p(2π)4eip(xy)ei(m2p2iϵ)TD(x, y, T) = \theta(T) \int \frac{d^4p}{(2\pi)^4} e^{-ip \cdot (x-y)} e^{-i(m^2 - p^2 - i\epsilon)T}

TT00\infty 积分:

DF(x,y)=d4p(2π)4eip(xy)0dTei(m2p2iϵ)TD_F(x, y) = \int \frac{d^4p}{(2\pi)^4} e^{-ip \cdot (x-y)} \int_0^\infty dT e^{-i(m^2 - p^2 - i\epsilon)T}

完成 TT 积分得到:

0dTei(m2p2iϵ)T=1i(m2p2iϵ)=ip2m2+iϵ\int_0^\infty dT e^{-i(m^2 - p^2 - i\epsilon)T} = \frac{1}{i(m^2 - p^2 - i\epsilon)} = \frac{i}{p^2 - m^2 + i\epsilon}

因此,

DF(x,y)=d4p(2π)4ip2m2+iϵeip(xy)\boxed{ D_F(x, y) = \int \frac{d^4p}{(2\pi)^4} \frac{i}{p^2 - m^2 + i\epsilon} e^{-ip \cdot (x-y)} }

这正是标准的自由标量场 Feynman 传播子。

(b) 在背景电磁场 AμA_\mu 中,标量场满足的 Klein-Gordon 方程中的普通导数被协变导数替代:μDμ=μ+ieAμ\partial_\mu \to D_\mu = \partial_\mu + ieA_\mu。对应的薛定谔方程变为:

[iT(D2+m2)]D(x,y,T)=iδ(T)δ(4)(xy)\left[ i\frac{\partial}{\partial T} - (D^2 + m^2) \right] D(x, y, T) = i\delta(T)\delta^{(4)}(x - y)

其中哈密顿算符为 H^=(μ+ieAμ)2+m2=(p^μeAμ)2+m2\hat{H} = -(\partial_\mu + ieA_\mu)^2 + m^2 = (\hat{p}_\mu - eA_\mu)^2 + m^2(这里 p^μ=iμ\hat{p}_\mu = -i\partial_\mu)。

现在我们从路径积分的角度推导对应的拉格朗日量。已知哈密顿量为 H(x,p)=(peA)2+m2H(x, p) = (p - eA)^2 + m^2。 由哈密顿方程,速度为:

x˙μ=Hpμ=2(pμeAμ)    pμ=12x˙μ+eAμ\dot{x}^\mu = \frac{\partial H}{\partial p_\mu} = 2(p^\mu - eA^\mu) \implies p^\mu = \frac{1}{2}\dot{x}^\mu + eA^\mu

进行勒让德变换得到拉格朗日量:

L=pμx˙μH=(12x˙μ+eAμ)x˙μ[(12x˙)2+m2]=14x˙2+ex˙μAμm2L = p_\mu \dot{x}^\mu - H = \left( \frac{1}{2}\dot{x}_\mu + eA_\mu \right)\dot{x}^\mu - \left[ \left( \frac{1}{2}\dot{x} \right)^2 + m^2 \right] = \frac{1}{4}\dot{x}^2 + e\dot{x}^\mu A_\mu - m^2

(注:题目表达式中的动能项系数为 1/21/2,这通常源于对固有时间 TT 的重新标度 T2TT \to 2T,若采用题目的标度,则对应 L=12x˙2+ex˙μAμ12m2L = \frac{1}{2}\dot{x}^2 + e\dot{x}^\mu A_\mu - \frac{1}{2}m^2。物理本质完全一致,此处采用标准系数 1/41/4 以严格匹配 2\partial^2 算符。)

将此拉格朗日量代入路径积分公式:

D(x,y,T)=x(0)=yx(T)=xDxexp[i0Tdt(14x˙2m2+ex˙μAμ)]D(x, y, T) = \int_{x(0)=y}^{x(T)=x} \mathcal{D}x \exp \left[ i \int_0^T dt \left( \frac{1}{4}\dot{x}^2 - m^2 + e\dot{x}^\mu A_\mu \right) \right]

其中相互作用项 exp(iex˙μAμdt)=exp(ieyxAμdxμ)\exp\left( ie \int \dot{x}^\mu A_\mu dt \right) = \exp\left( ie \int_y^x A_\mu dx^\mu \right) 正是阿贝尔规范场下的 Wilson line。 根据路径积分与算符演化的标准等价性,该路径积分定义的演化核必然满足以 H^=(p^eA)2+m2=D2+m2\hat{H} = (\hat{p} - eA)^2 + m^2 = -D^2 + m^2 为哈密顿量的薛定谔方程:

iTD(x,y,T)=H^D(x,y,T)=(D2+m2)D(x,y,T)\boxed{ i\frac{\partial}{\partial T} D(x, y, T) = \hat{H} D(x, y, T) = (-D^2 + m^2) D(x, y, T) }

这证明了该泛函积分确实是背景电磁场中标量传播子的表示。

(c) 对于非阿贝尔规范理论,规范场 Aμ=AμataA_\mu = A_\mu^a t^a 是矩阵值的,其中 tat^a 是规范群的生成元。 此时,Wilson line 因子推广为:

W(x,y)=Pexp(ig0Tdtx˙μAμ(x(t)))W(x, y) = \mathcal{P} \exp \left( ig \int_0^T dt \, \dot{x}^\mu A_\mu(x(t)) \right)

其中 P\mathcal{P} 表示路径排序(Path-ordering)。

路径排序的必要性证明: 在薛定谔方程中,我们需要计算时间演化算符对 TT 的导数。将时间区间 [0,T][0, T] 分割为无穷多个微小时间步 Δt\Delta t。路径积分的离散形式为一系列短时演化算符的乘积:

U(T,0)=limNn=1Nexp[iΔtH^(tn)]U(T, 0) = \lim_{N \to \infty} \prod_{n=1}^N \exp\left[ -i \Delta t \hat{H}(t_n) \right]

在非阿贝尔情况下,不同空间点上的矩阵 Aμ(x)A_\mu(x) 彼此不对易,即 [Aμ(x(t1)),Aν(x(t2))]0[A_\mu(x(t_1)), A_\nu(x(t_2))] \neq 0。 如果指数中的积分没有路径排序,直接求导会得到:

iTexp(ig0Tdtx˙μAμ)gx˙μ(T)Aμ(x(T))exp(ig0Tdtx˙μAμ)i\frac{\partial}{\partial T} \exp\left( ig \int_0^T dt \, \dot{x}^\mu A_\mu \right) \neq -g \dot{x}^\mu(T) A_\mu(x(T)) \exp\left( ig \int_0^T dt \, \dot{x}^\mu A_\mu \right)

因为 Aμ(x(T))A_\mu(x(T)) 无法穿过积分中其他时间的 AνA_\nu 矩阵提取到最左侧。 只有引入路径排序 P\mathcal{P},使得晚时刻的矩阵总是排在早时刻矩阵的左侧,我们才能在求导时将最新时刻 TT 的哈密顿量准确地提取到最左边:

iT[Pexp(i0TdtH^(t))]=H^(T)[Pexp(i0TdtH^(t))]i\frac{\partial}{\partial T} \left[ \mathcal{P} \exp \left( -i \int_0^T dt \, \hat{H}(t) \right) \right] = \hat{H}(T) \left[ \mathcal{P} \exp \left( -i \int_0^T dt \, \hat{H}(t) \right) \right]

对应到路径积分中,这保证了算符 H^=D2+m2\hat{H} = -D^2 + m^2 能够正确地作用在传播子 D(x,y,T)D(x, y, T) 的末态坐标 xx 上。因此,

只有当 Wilson line 是路径排序时,泛函积分才能满足非阿贝尔协变导数下的薛定谔方程 iTD=(D2+m2)D\boxed{ \text{只有当 Wilson line 是路径排序时,泛函积分才能满足非阿贝尔协变导数下的薛定谔方程 } i\partial_T D = (-D^2 + m^2)D }
15.5

Problem 15.5

peskinChapter 15

习题 15.5

来源: 第15章, PDF第504页


15.5 Casimir operator computations. An alternative strategy for computing the quadratic Casimir operator is to compute C(r)C(r) in the formula

tr[tratrb]=C(r)δab\text{tr}[t_r^a t_r^b] = C(r) \delta^{ab}

by choosing tat^a and tbt^b to lie in an SU(2)SU(2) subgroup of the gauge group.

(a) Under an SU(2)SU(2) subgroup of a general group GG, an irreducible representation rr of GG will decompose into a sum of representations of SU(2)SU(2):

riji,r \rightarrow \sum_i j_i,

where the jij_i are the spins of SU(2)SU(2) representations. Show that

3C(r)=iji(ji+1)(2ji+1).3C(r) = \sum_i j_i(j_i + 1)(2j_i + 1).

(b) Under an SU(2)SU(2) subgroup of SU(N)SU(N), the fundamental representation NN transforms as a 2-component spinor (j=12j = \frac{1}{2}) and (N2)(N - 2) singlets. Use this relation to check the formula C(N)=12C(N) = \frac{1}{2}. Show that the adjoint representation of SU(N)SU(N) decomposes into one spin 1, 2(N2)2(N - 2) spin-12\frac{1}{2}'s, plus singlets, and use this decomposition to check that C(G)=NC(G) = N.

(c) Symmetric and antisymmetric 2-index tensors form irreducible representations of SU(N)SU(N). Compute C2(r)C_2(r) for each of these representations. The direct sum of these representations is the product representation N×NN \times N. Verify that your results for C2(r)C_2(r) satisfy the identity for product representations that follows from Eqs. (15.100) and (15.101).


Referenced Equations:

Equation (15.100):

tr(tr1×r2a)2=(C2(r1)+C2(r2))d(r1)d(r2).(15.100)\text{tr}(t_{r_1 \times r_2}^a)^2 = (C_2(r_1) + C_2(r_2)) d(r_1) d(r_2) . \tag{15.100}

Equation (15.101):

tr(tr1×r2a)2=C2(ri)d(ri).(15.101)\text{tr}(t_{r_1 \times r_2}^a)^2 = \sum C_2(r_i) d(r_i) . \tag{15.101}

习题 15.5 - 解答


(a) 根据定义,表示 rr 的指数 C(r)C(r) 满足迹公式:

tr[tratrb]=C(r)δab\text{tr}[t_r^a t_r^b] = C(r) \delta^{ab}

我们在规范群 GG 中选取一个 SU(2)SU(2) 子群,并令 tat^a (a=1,2,3a=1,2,3) 为该 SU(2)SU(2) 子群的生成元。对 a=1,2,3a=1,2,3 求和,等式左边变为:

a=13tr[tratra]=a=13C(r)δaa=3C(r)\sum_{a=1}^3 \text{tr}[t_r^a t_r^a] = \sum_{a=1}^3 C(r) \delta^{aa} = 3C(r)

另一方面,算符 a=13(tra)2\sum_{a=1}^3 (t_r^a)^2 正是该 SU(2)SU(2) 子群的二次卡西米尔算符 (quadratic Casimir operator)。 在 SU(2)SU(2) 子群下,群 GG 的不可约表示 rr 分解为 SU(2)SU(2) 表示的直和:rijir \rightarrow \sum_i j_i。 对于每一个自旋为 jij_iSU(2)SU(2) 不可约表示,其卡西米尔算符正比于单位阵,本征值为 ji(ji+1)j_i(j_i+1),且该表示的维数为 2ji+12j_i+1。因此,在表示 rr 上的迹等于各个 SU(2)SU(2) 子表示的迹之和:

a=13tr[tratra]=iji(ji+1)(2ji+1)\sum_{a=1}^3 \text{tr}[t_r^a t_r^a] = \sum_i j_i(j_i+1)(2j_i+1)

将两式联立,即证得:

3C(r)=iji(ji+1)(2ji+1)\boxed{3C(r) = \sum_i j_i(j_i+1)(2j_i+1)}

(b) 对于 SU(N)SU(N) 的基础表示 NN,在 SU(2)SU(2) 子群下分解为一个自旋 j=12j=\frac{1}{2} 的表示和 N2N-2 个自旋 j=0j=0 的单态。代入 (a) 中的公式:

3C(N)=12(12+1)(2×12+1)+(N2)×0=34×2=323C(N) = \frac{1}{2}\left(\frac{1}{2}+1\right)\left(2\times\frac{1}{2}+1\right) + (N-2) \times 0 = \frac{3}{4} \times 2 = \frac{3}{2}

由此可得基础表示的指数:

C(N)=12\boxed{C(N) = \frac{1}{2}}

SU(N)SU(N) 的伴随表示 GG 可以看作是张量积 NNˉN \otimes \bar{N} 的无迹部分。在 SU(2)SU(2) 子群下,NNNˉ\bar{N} 均分解为 12(N2)×0\frac{1}{2} \oplus (N-2) \times 0。它们的张量积分解为:

NNˉ(12(N2)×0)(12(N2)×0)N \otimes \bar{N} \rightarrow \left( \frac{1}{2} \oplus (N-2) \times 0 \right) \otimes \left( \frac{1}{2} \oplus (N-2) \times 0 \right)
=(1212)2(N2)×(120)(N2)2×(00)= \left( \frac{1}{2} \otimes \frac{1}{2} \right) \oplus 2(N-2) \times \left( \frac{1}{2} \otimes 0 \right) \oplus (N-2)^2 \times (0 \otimes 0)

利用 SU(2)SU(2) 的角动量耦合 1212=10\frac{1}{2} \otimes \frac{1}{2} = 1 \oplus 0,我们得到:

NNˉ12(N2)×12(1+(N2)2)×0N \otimes \bar{N} \rightarrow 1 \oplus 2(N-2) \times \frac{1}{2} \oplus \left( 1 + (N-2)^2 \right) \times 0

伴随表示 GG 是通过从 NNˉN \otimes \bar{N} 中减去一个对应于迹的单态(自旋 00)得到的。因此,伴随表示分解为 1 个自旋 1,2(N-2) 个自旋 1/2,以及 (N2)2(N-2)^2 个单态。 利用此分解计算伴随表示的指数 C(G)C(G)

3C(G)=1(1+1)(2×1+1)+2(N2)×12(12+1)(2×12+1)+(N2)2×03C(G) = 1(1+1)(2\times 1+1) + 2(N-2) \times \frac{1}{2}\left(\frac{1}{2}+1\right)\left(2\times\frac{1}{2}+1\right) + (N-2)^2 \times 0
3C(G)=6+2(N2)×32=6+3(N2)=3N3C(G) = 6 + 2(N-2) \times \frac{3}{2} = 6 + 3(N-2) = 3N

由此验证了:

C(G)=N\boxed{C(G) = N}

(c) 对称 (SS) 和反对称 (AA) 双指标张量表示包含在乘积表示 NNN \otimes N 中。在 SU(2)SU(2) 子群下,NNN \otimes N 的分解与 NNˉN \otimes \bar{N} 类似:

NN(1212)2(N2)×(120)(N2)2×(00)N \otimes N \rightarrow \left( \frac{1}{2} \otimes \frac{1}{2} \right) \oplus 2(N-2) \times \left( \frac{1}{2} \otimes 0 \right) \oplus (N-2)^2 \times (0 \otimes 0)
  • 1212\frac{1}{2} \otimes \frac{1}{2} 产生一个对称的自旋 11 和一个反对称的自旋 00
  • 2(N2)2(N-2) 个自旋 1/21/2 来自交叉项,其中一半是对称的,一半是反对称的,因此 SSAA 各分得 (N2)(N-2) 个自旋 1/21/2
  • (N2)2(N-2)^2 个单态分解为 (N2)(N1)2\frac{(N-2)(N-1)}{2} 个对称单态和 (N2)(N3)2\frac{(N-2)(N-3)}{2} 个反对称单态。

因此,对称表示 SS 包含 1 个自旋 1 和 (N2)(N-2) 个自旋 1/2;反对称表示 AA 包含 (N2)(N-2) 个自旋 1/2。利用 (a) 的公式计算它们的指数:

3C(S)=1(2)(3)+(N2)×32=6+32(N2)=32(N+2)    C(S)=N+223C(S) = 1(2)(3) + (N-2) \times \frac{3}{2} = 6 + \frac{3}{2}(N-2) = \frac{3}{2}(N+2) \implies C(S) = \frac{N+2}{2}
3C(A)=0+(N2)×32=32(N2)    C(A)=N223C(A) = 0 + (N-2) \times \frac{3}{2} = \frac{3}{2}(N-2) \implies C(A) = \frac{N-2}{2}

二次卡西米尔算符 C2(r)C_2(r) 定义为 atratra=C2(r)Id(r)\sum_a t_r^a t_r^a = C_2(r) I_{d(r)}。两边取迹可得 C2(r)d(r)=C(r)d(G)C_2(r) d(r) = C(r) d(G),其中 d(G)=N21d(G) = N^2-1 是伴随表示的维数。 对称和反对称表示的维数分别为 d(S)=N(N+1)2d(S) = \frac{N(N+1)}{2}d(A)=N(N1)2d(A) = \frac{N(N-1)}{2}。由此可得:

C2(S)=C(S)d(G)d(S)=N+22(N21)N(N+1)2=(N1)(N+2)N\boxed{C_2(S) = \frac{C(S) d(G)}{d(S)} = \frac{\frac{N+2}{2} (N^2-1)}{\frac{N(N+1)}{2}} = \frac{(N-1)(N+2)}{N}}
C2(A)=C(A)d(G)d(A)=N22(N21)N(N1)2=(N+1)(N2)N\boxed{C_2(A) = \frac{C(A) d(G)}{d(A)} = \frac{\frac{N-2}{2} (N^2-1)}{\frac{N(N-1)}{2}} = \frac{(N+1)(N-2)}{N}}

验证乘积表示的恒等式: 由公式 (15.100) 和 (15.101) 给出的恒等式为:

(C2(r1)+C2(r2))d(r1)d(r2)=iC2(ri)d(ri)(C_2(r_1) + C_2(r_2)) d(r_1) d(r_2) = \sum_i C_2(r_i) d(r_i)

对于 r1=r2=Nr_1 = r_2 = N,乘积表示分解为 SAS \oplus A。已知 C2(N)=N212NC_2(N) = \frac{N^2-1}{2N}d(N)=Nd(N) = N。 计算等式左边 (LHS):

LHS=(C2(N)+C2(N))d(N)2=2(N212N)N2=N(N21)\text{LHS} = (C_2(N) + C_2(N)) d(N)^2 = 2 \left( \frac{N^2-1}{2N} \right) N^2 = N(N^2-1)

计算等式右边 (RHS):

RHS=C2(S)d(S)+C2(A)d(A)=(N1)(N+2)NN(N+1)2+(N+1)(N2)NN(N1)2\text{RHS} = C_2(S) d(S) + C_2(A) d(A) = \frac{(N-1)(N+2)}{N} \frac{N(N+1)}{2} + \frac{(N+1)(N-2)}{N} \frac{N(N-1)}{2}
=12(N1)(N+1)(N+2)+12(N+1)(N1)(N2)= \frac{1}{2} (N-1)(N+1)(N+2) + \frac{1}{2} (N+1)(N-1)(N-2)

提取公因子 12(N1)(N+1)=12(N21)\frac{1}{2}(N-1)(N+1) = \frac{1}{2}(N^2-1)

RHS=12(N21)[(N+2)+(N2)]=12(N21)(2N)=N(N21)\text{RHS} = \frac{1}{2} (N^2-1) [ (N+2) + (N-2) ] = \frac{1}{2} (N^2-1) (2N) = N(N^2-1)

显然 LHS=RHS\text{LHS} = \text{RHS},恒等式得到完美验证。