This appendix collects the results from the representation theory of SU(2){\mathrm{SU}}(2) that are used in the main text: irreducible representations, the Peter–Weyl theorem, intertwiners, and the 3j3j and 6j6j symbols. The conventions are those of the Notation and Conventions page.

A.1 The group SU(2) and Its Lie Algebra

The group SU(2){\mathrm{SU}}(2) consists of the unitary complex 2×22\times2 matrices of unit determinant; it is diffeomorphic to the sphere S3S^3 and is the double cover of SO(3){\mathrm{SO}}(3). Its Lie algebra su(2){\mathfrak{su}}(2) is generated by the anti-Hermitian matrices τi=−i2σi\tau_i = -\frac{\mathrm{i}}2\sigma_i, with commutation relations [τi,τj]=ϵijkτk[\tau_i,\tau_j] = \epsilon_{ij}{}^k\tau_k. Every element can be written as h=exp⁡(θ n^iτi)h = \exp(\theta\,\hat n^i\tau_i), where θ∈[0,2π]\theta\in[0,2\pi] is the rotation angle and n^\hat n is a unit vector. In terms of the Euler angles (ϕ,ϑ,ψ)∈[0,2π)×[0,π]×[0,4π)(\phi,\vartheta,\psi)\in[0,2\pi)\times[0,\pi]\times[0,4\pi), the normalized Haar measure, which is both left- and right-invariant, reads

dh=116π2 sin⁡ϑ  dϕ dϑ dψ,∫SU(2)dh=1.(A.1){\mathrm{d}} h = \frac{1}{16\pi^2}\,\sin\vartheta\;{\mathrm{d}}\phi\,{\mathrm{d}}\vartheta\,{\mathrm{d}}\psi, \qquad \int_{{\mathrm{SU}}(2)}{\mathrm{d}} h = 1 . \tag{A.1}

A.2 Irreducible Representations

The irreducible unitary representations of SU(2){\mathrm{SU}}(2) are labeled by a spin j∈N/2j\in\mathbb N/2. The carrier space VjV_j of the spin-jj representation has dimension dj=2j+1d_j = 2j+1 and admits the orthonormal basis {∣j,m⟩}m=−j,…,j\{{\left|{j,m}\right\rangle}\}_{m=-j,\dots,j} of eigenvectors of J3=iτ3J_3 = {\mathrm{i}}\tau_3:

J⃗2∣j,m⟩=j(j+1)∣j,m⟩,J3∣j,m⟩=m∣j,m⟩,J±∣j,m⟩=j(j+1)−m(m±1)  ∣j,m±1⟩,(A.2)\begin{gathered} {\vec{{J}}}^2{\left|{j,m}\right\rangle} = j(j+1){\left|{j,m}\right\rangle}, \qquad J_3{\left|{j,m}\right\rangle} = m{\left|{j,m}\right\rangle}, \\ J_\pm{\left|{j,m}\right\rangle} = \sqrt{j(j+1)-m(m\pm1)}\;{\left|{j,m\pm1}\right\rangle} , \end{gathered} \tag{A.2}

with J±=J1±iJ2J_\pm = J_1\pm{\mathrm{i}} J_2. The Wigner matrices Dmn(j)(h)=⟨j,m|h|j,n⟩D^{(j)}_{mn}(h) = {\left\langle{j,m}\middle|{h}\middle|{j,n}\right\rangle} represent the group elements, and their trace defines the character

χj(h)=Tr⁡D(j)(h)=sin⁡((2j+1)θ/2)sin⁡(θ/2),(A.3)\chi^j(h) = {\operatorname{Tr}} D^{(j)}(h) = \frac{\sin\big((2j+1)\theta/2\big)}{\sin(\theta/2)}, \tag{A.3}

which depends only on the conjugacy class of hh, that is, on the angle θ\theta. For j=12j = \tfrac12, we recover χ1/2=Tr⁡h=2cos⁡(θ/2)\chi^{1/2} = {\operatorname{Tr}} h = 2\cos(\theta/2), the trace that appears in Wilson loops.

A.3 The Peter–Weyl Theorem

The functions 2j+1 Dmn(j)\sqrt{2j+1}\,D^{(j)}_{mn} form an orthonormal basis of L2(SU(2),dh)L^2({\mathrm{SU}}(2),{\mathrm{d}} h); the orthogonality relations are given by (3.20). This yields the decomposition

L2(SU(2),dh)=⨁j∈N/2Vj⊗Vj∗,(A.4)L^2\big({\mathrm{SU}}(2),{\mathrm{d}} h\big) = \bigoplus_{j\in\mathbb N/2} V_j\otimes V_j^* , \tag{A.4}

and, for class functions, the orthonormality of the characters, ∫dh  χj(h)‾ χj′(h)=δjj′\int{\mathrm{d}} h\;\overline{\chi^j(h)}\,\chi^{j'}(h) = \delta_{jj'}. The delta distribution on the group, defined with respect to the Haar measure, admits the expansion

δ(h)=∑j∈N/2(2j+1) χj(h),(A.5)\delta(h) = \sum_{j\in\mathbb N/2}(2j+1)\,\chi^j(h), \tag{A.5}

which is used in Chapter 5. Finally, the integral of a tensor product of representations is the projector onto the invariant subspace:

∫SU(2)dh  D(j1)(h)⊗⋯⊗D(jn)(h)=PInv=∑ι∣ι⟩⟨ι∣,(A.6)\int_{{\mathrm{SU}}(2)}{\mathrm{d}} h\;D^{(j_1)}(h)\otimes\cdots\otimes D^{(j_n)}(h) = \mathbb P_{\mathrm{Inv}} = \sum_{\iota}{\left|{\iota}\right\rangle}{\left\langle{\iota}\right|}, \tag{A.6}

where the sum runs over an orthonormal basis of Inv(Vj1⊗⋯⊗Vjn)\mathrm{Inv}(V_{j_1}\otimes\cdots\otimes V_{j_n}).

A.4 Coupling of Angular Momenta and Intertwiners

The tensor product of two irreducible representations decomposes according to the Clebsch–Gordan rule

Vj1⊗Vj2=⨁j=∣j1−j2∣j1+j2Vj,(A.7)V_{j_1}\otimes V_{j_2} = \bigoplus_{j={\left|{j_1-j_2}\right|}}^{j_1+j_2} V_j , \tag{A.7}

where jj varies in integer steps. From this rule one obtains the dimension of the intertwiner spaces, which appear at the nodes of spin networks.

Trivalent nodes. The space Inv(Vj1⊗Vj2⊗Vj3)\mathrm{Inv}(V_{j_1}\otimes V_{j_2}\otimes V_{j_3}) is one-dimensional if ∣j1−j2∣≤j3≤j1+j2{\left|{j_1-j_2}\right|}\le j_3\le j_1+j_2 and j1+j2+j3∈Nj_1+j_2+j_3\in\mathbb N, and trivial otherwise. The normalized intertwiner is given by the Wigner 3j3j symbol, which is related to the Clebsch–Gordan coefficients by

(j1j2j3m1m2m3)=(−1)j1−j2−m32j3+1  ⟨j1,m1; j2,m2|j3,−m3⟩.(A.8)\begin{pmatrix} j_1 & j_2 & j_3\\ m_1 & m_2 & m_3 \end{pmatrix} = \frac{(-1)^{j_1-j_2-m_3}}{\sqrt{2j_3+1}}\;{\left\langle{j_1,m_1;\,j_2,m_2}\middle|{j_3,-m_3}\right\rangle} . \tag{A.8}

It is nonzero only if m1+m2+m3=0m_1+m_2+m_3 = 0, and it satisfies ∑mi∣j1j2j3m1m2m3∣2=1\sum_{m_i}\big|\begin{smallmatrix} j_1 & j_2 & j_3\\ m_1 & m_2 & m_3\end{smallmatrix}\big|^2 = 1.

Four-valent nodes. A basis of Inv(Vj1⊗⋯⊗Vj4)\mathrm{Inv}(V_{j_1}\otimes\cdots\otimes V_{j_4}) is obtained by first coupling j1j_1 and j2j_2 into a “virtual” spin kk, and then coupling j3j_3 and j4j_4 into the same spin: the intertwiner ιk\iota_k is the contraction of two 3j3j symbols along a virtual link of spin kk, normalized by the factor 2k+1\sqrt{2k+1}. The virtual spin ranges over

max⁡(∣j1−j2∣,∣j3−j4∣)≤k≤min⁡(j1+j2, j3+j4),(A.9)\max\big({\left|{j_1-j_2}\right|},{\left|{j_3-j_4}\right|}\big) \le k \le \min\big(j_1+j_2,\, j_3+j_4\big), \tag{A.9}

in integer steps, and the number of allowed values is the dimension of the intertwiner space. For four spins 12\tfrac12, we find k∈{0,1}k\in\{0,1\}: the space is two-dimensional, a fact used in the volume computation (3.36). The change of coupling channel, for instance from (j1j2)(j3j4)(j_1j_2)(j_3j_4) to (j1j3)(j2j4)(j_1j_3)(j_2j_4), is given by the 6j6j symbols.

A.5 The 6j Symbol

The 6j6j symbol is the contraction of four 3j3j symbols following the structure of a tetrahedron:

{j1j2j3j4j5j6}=∑m1,…,m6(−1)∑k=16(jk−mk)  (j1j2j3−m1−m2−m3)(j1j5j6m1−m5m6)×  (j4j2j6m4m2−m6)(j4j5j3−m4m5m3).\begin{align} {\begin{Bmatrix}{j_1}&{j_2}&{j_3}\\{j_4}&{j_5}&{j_6}\end{Bmatrix}} = \sum_{m_1,\dots,m_6} (-1)^{\sum_{k=1}^6 (j_k-m_k)}\; &\begin{pmatrix} j_1 & j_2 & j_3\\ -m_1 & -m_2 & -m_3\end{pmatrix} \begin{pmatrix} j_1 & j_5 & j_6\\ m_1 & -m_5 & m_6\end{pmatrix} \\ \times\; &\begin{pmatrix} j_4 & j_2 & j_6\\ m_4 & m_2 & -m_6\end{pmatrix} \begin{pmatrix} j_4 & j_5 & j_3\\ -m_4 & m_5 & m_3\end{pmatrix} . \tag{A.10} \end{align}

Each of the four triples (j1,j2,j3)(j_1,j_2,j_3), (j1,j5,j6)(j_1,j_5,j_6), (j4,j2,j6)(j_4,j_2,j_6) and (j4,j5,j3)(j_4,j_5,j_3) must satisfy the triangle conditions; they correspond to the four faces of the tetrahedron in Fig. 5.5. The 6j6j symbol is invariant under the 2424 symmetry operations of the tetrahedron, namely permutations of its columns and the interchange of the upper and lower entries in any two columns. It gives the recoupling coefficients of three angular momenta,

⟨(j1j2)j12,j3;J|j1,(j2j3)j23;J⟩=(−1)j1+j2+j3+J(2j12+1)(2j23+1)  {j1j2j12j3Jj23},(A.11){\left\langle{(j_1j_2)j_{12},j_3;J}\middle|{j_1,(j_2j_3)j_{23};J}\right\rangle} = (-1)^{j_1+j_2+j_3+J}\sqrt{(2j_{12}+1)(2j_{23}+1)}\;{\begin{Bmatrix}{j_1}&{j_2}&{j_{12}}\\{j_3}&{J}&{j_{23}}\end{Bmatrix}}, \tag{A.11}

and satisfies the orthogonality relation

∑x(2x+1)(2y+1){abxcdy}{abxcdy′}=δyy′.(A.12)\sum_x (2x+1)(2y+1){\begin{Bmatrix}{a}&{b}&{x}\\{c}&{d}&{y}\end{Bmatrix}}{\begin{Bmatrix}{a}&{b}&{x}\\{c}&{d}&{y'}\end{Bmatrix}} = \delta_{yy'} . \tag{A.12}

For example, {111111}={1212112121}=16{\begin{Bmatrix}{1}&{1}&{1}\\{1}&{1}&{1}\end{Bmatrix}} = {\begin{Bmatrix}{\frac12}&{\frac12}&{1}\\{\frac12}&{\frac12}&{1}\end{Bmatrix}} = \frac16 and {222222}=−370{\begin{Bmatrix}{2}&{2}&{2}\\{2}&{2}&{2}\end{Bmatrix}} = -\frac{3}{70}. We have verified the asymptotic formula (5.22) numerically for the regular tetrahedron: for j=20j = 20, the exact value {6j}≃−5.029×10−3\{6j\}\simeq-5.029\times10^{-3} is reproduced to within 0.1 %0.1\,\% by the Ponzano–Regge formula, which gives −5.034×10−3-5.034\times10^{-3}.