Braid Group Representations, Modular Data, and the Classification of Majorana Zero Mode Fusion Rules in 2D Topological Superconductors

Abstract

We investigate whether the braid group representation on $N$ anyons uniquely determines the modular data ($S$ and $T$ matrices) of a non-Abelian topological order, and whether this framework classifies all fusion rules for Majorana zero modes (MZMs) in 2D topological superconductors. We show that the local $N$-anyon braid group representation constrains the modular data up to finite gauge ambiguity but does not uniquely fix the global $S$-matrix without specifying the underlying spin structure. For MZMs, the braid generators satisfy the Ising anyon $R$-matrix with the asymptotic Hecke relation $\sigma_i^2 = 1$ (up to phase), which rigidly fixes the local $F$ and $R$ symbols of the Ising category. Consequently, the fusion rules are strictly of Ising type ($\sigma \times \sigma = 1 + \psi$), precluding alternative non-Abelian fusion rules. The $T$-matrix is determined by topological spins, but the full $S$-matrix depends on the spin structure of the underlying manifold, reflecting the super-modular structure of the theory.

1. Introduction

Topological quantum computation exploits the non-Abelian statistics of anyons—quasiparticle excitations in two-dimensional systems whose braiding operations are represented by non-commuting unitary matrices on a degenerate ground-state manifold [1]. A central question is whether the local braiding data of $N$ anyons suffices to reconstruct the full modular data of the underlying topological order (TO), or whether additional global information is required. This question has direct physical consequences for 2D topological superconductors (TSCs) hosting Majorana zero modes (MZMs), where the anyonic excitations are vortices binding MZMs and the fermionic quasiparticle is a transparent (local) fermion [2,3].

The modular data of a TO consists of the $S$-matrix, which encodes the mutual braiding statistics of anyons, and the $T$-matrix, which encodes the topological spins. Together, these matrices form a projective representation of the modular group $\mathrm{SL}(2,\mathbb{Z})$ and satisfy the Verlinde formula, which determines the fusion rules [4]. For bosonic TOs described by modular tensor categories (MTCs), the modular data is a complete invariant up to a finite number of exceptions [5]. However, 2D TSCs are fermionic systems whose effective TOs are described by super-modular categories over $\mathrm{sVec}$ (the category of super-vector spaces), where the physical electron is a transparent fermion $\psi$ that braids trivially with all anyons but has nontrivial self-statistics $\theta_\psi = -1$ [6].

In this work, we address two questions: (i) Does the representation theory of the braid group $B_N$ on $N$ anyons uniquely determine the modular data of a non-Abelian TO? (ii) Can this framework classify all possible fusion rules for MZMs in 2D TSCs? We show that the local $B_N$ representation fixes the $F$ and $R$ symbols—and hence the $T$-matrix—of the Ising category, but the global $S$-matrix remains dependent on the spin structure of the underlying manifold. The fusion rules are rigidly classified as Ising type, precluding alternative non-Abelian fusion rules for strictly Majorana systems.

2. Background

2.1 Braid Group Representations

The braid group $B_N$ on $N$ strands is generated by elementary exchanges $\sigma_1, \ldots, \sigma_{N-1}$ satisfying the braid relations $\sigma_i \sigma_j = \sigma_j \sigma_i$ for $|i-j| \geq 2$ and $\sigma_i \sigma_{i+1} \sigma_i = \sigma_{i+1} \sigma_i \sigma_{i+1}$. A unitary representation $\rho: B_N \to U(V)$ on a finite-dimensional Hilbert space $V$ arises from the braiding of $N$ identical anyons of type $a$ in a TO. The representation depends on the fusion channels available: the dimension of $V$ is determined by the fusion multiplicities $N_{aa\ldots a}^{c}$ for all possible total charges $c$ [1,5].

2.2 Modular Data and Fusion Categories

A unitary modular tensor category (UMTC) $\mathcal{C}$ is a braided fusion category with non-degenerate braiding. Its simple objects $\{a, b, c, \ldots\}$ satisfy fusion rules $a \times b = \sum_c N_{ab}^c \, c$. The associativity of fusion is governed by $F$-symbols $F^{abc}_{d}$, and the braiding is governed by $R$-symbols $R^{ab}_c$. These must satisfy the pentagon and hexagon equations, which are polynomial constraints ensuring consistency of the fusion and braiding structure [5].

The modular $S$-matrix is defined by the Hopf link invariant $S_{ab} = \frac{1}{\mathcal{D}} \sum_c d_c \, \theta_c \, N_{ab}^c / d_a d_b$ (where $\mathcal{D} = \sqrt{\sum_a d_a^2}$ is the total quantum dimension and $\theta_c$ is the topological spin), and the $T$-matrix is diagonal with entries $T_{aa} = \theta_a$. The Verlinde formula [4] expresses the fusion coefficients as:

where $1$ denotes the vacuum. Thus, the $S$-matrix determines the fusion rules.

2.3 Super-Modular Categories and 2D TSCs

A super-modular category is a unitary braided fusion category $\mathcal{C}$ over $\mathrm{sVec}$ whose Müger center (the subcategory of objects braiding trivially with all others) is precisely $\mathrm{sVec} = \{1, \psi\}$, where $\psi$ is a transparent fermion with $\theta_\psi = -1$ [6]. This structure describes the TO of a 2D TSC, where $\psi$ is the physical electron. The category is not modular in the usual sense because $\psi$ is transparent, rendering the $S$-matrix degenerate. The full modular data requires a choice of spin structure on the underlying manifold.

2.4 Majorana Zero Modes and Ising Anyons

MZMs are zero-energy bound states localized at vortex cores in 2D $p+ip$ TSCs [2,3]. Each vortex binds a single MZM $\gamma_i$ satisfying $\gamma_i = \gamma_i^\dagger$ and $\{\gamma_i, \gamma_j\} = 2\delta_{ij}$. The braiding of $2N$ MZMs realizes a projective representation of $B_{2N}$ equivalent to the Ising anyon theory, where the non-Abelian anyon $\sigma$ (the vortex) has fusion rule $\sigma \times \sigma = 1 + \psi$ [2]. The $R$-symbols are $R^{\sigma\sigma}_1 = e^{-i\pi/8}$ and $R^{\sigma\sigma}_\psi = e^{3i\pi/8}$, and the $F$-symbol is the unique nontrivial associator $F^{\sigma\sigma\sigma}_\sigma = \frac{1}{\sqrt{2}}$ (up to gauge) [5].

3. Analysis

3.1 Braid Group Representation for MZMs

Consider $2N$ MZMs $\gamma_1, \ldots, \gamma_{2N}$ in a 2D TSC. The Hilbert space for $2N$ MZMs with total fermion parity fixed has dimension $2^{N-1}$. The braid operator for exchanging $\gamma_i$ and $\gamma_{i+1}$ is:

This satisfies the braid relation and the asymptotic Hecke relation:

which acts as a fermion parity operator. In the anyon language, $\sigma_i^2$ measures the fusion channel: it acts as $+1$ in the vacuum channel and $-1$ in the $\psi$ channel (up to an overall phase $e^{-i\pi/4}$), consistent with the Ising $R$-symbols [2].

The key observation is that the local braid representation $\rho: B_{2N} \to U(2^{N-1})$ is determined entirely by the Clifford algebra relations of the Majorana operators. The eigenvalues of $\rho(\sigma_i)$ are $e^{-i\pi/8}$ and $e^{3i\pi/8}$, corresponding to the two fusion channels of $\sigma \times \sigma$. These eigenvalues are the $R$-symbol entries, which directly give the topological spins via the balancing relation $\theta_a \theta_b = \theta_c R^{ab}_c R^{ba}_c$ for $a \times b \ni c$.

3.2 Extraction of the $T$-Matrix

From the $R$-symbols, we extract the topological spins. For the Ising category, the simple objects are $\{1, \sigma, \psi\}$ with quantum dimensions $d_1 = d_\psi = 1$ and $d_\sigma = \sqrt{2}$. The topological spins are:

These are determined by the $R$-symbols via $\theta_a = \sum_c (d_c / d_a) R^{aa}_c$ (for $a \times a \ni c$). Specifically, $\theta_\sigma = \frac{1}{2}(R^{\sigma\sigma}_1 \cdot d_1 + R^{\sigma\sigma}_\psi \cdot d_\psi)/d_\sigma = \frac{1}{2\sqrt{2}}(e^{-i\pi/8} + e^{3i\pi/8}) = e^{i\pi/8}$.

Thus, the $T$-matrix $T = \mathrm{diag}(1, e^{i\pi/8}, -1)$ is uniquely determined by the local braid representation. The overall phase of $T$ (the framing anomaly) is fixed up to a 24th root of unity, consistent with the central charge $c = 1/2$ of the Ising CFT [5].

3.3 The $S$-Matrix and Spin Structure Ambiguity

For a modular (bosonic) category, the $S$-matrix can be reconstructed from the $F$ and $R$ symbols via the formula:

For the Ising category, this yields the well-known $S$-matrix:

However, in a 2D TSC, the physical TO is not the Ising MTC but rather the Ising super-modular category $\mathrm{Ising} \boxtimes \mathrm{sVec}$ quotiented by the transparent fermion. The $S$-matrix of this super-modular category is degenerate: $S_{\psi, a} = S_{1, a}$ for all $a$, because $\psi$ is transparent. The non-degenerate modular $S$-matrix relevant for physical observables on a spin manifold depends on the choice of spin structure [6].

Specifically, on a torus with spin structure $\nu \in \{0,1\}^2$, the partition function and modular transformations are modified. The Arf invariant of the spin structure determines whether the $\sigma$ anyon contributes to the ground-state degeneracy. The four spin structures on $T^2$ yield two distinct sectors: the Neveu-Schwarz (NS) sector and the Ramond (R) sector, which differ in their modular $S$-matrix entries for the $\sigma$ anyon [to verify: exact dependence of $S_{\sigma\sigma}$ on spin structure].

3.4 Hexagon Equations and Uniqueness of $F$ and $R$

The consistency of the $F$ and $R$ symbols is enforced by the pentagon and hexagon equations. For the Ising fusion rules ($\sigma \times \sigma = 1 + \psi$, $\sigma \times \psi = \sigma$, $\psi \times \psi = 1$), the pentagon equation admits a unique solution (up to gauge) for the $F$-symbol $F^{\sigma\sigma\sigma}_\sigma = \frac{1}{\sqrt{2}}$ [5]. The hexagon equations then constrain the $R$-symbols to:

up to an overall phase determined by the central charge. This solution is unique up to gauge transformations $F \to U F U^\dagger$ and $R \to U R U^\dagger$ for unitary diagonal matrices $U$.

The braid group representation $\rho(\sigma_i) = \frac{1}{\sqrt{2}}(1 + \gamma_{i+1}\gamma_i)$ directly realizes these $R$-symbols. The eigenvalues $e^{\pm i\pi/8}$ and $e^{\pm 3i\pi/8}$ of the braid generators match the $R$-symbol entries, confirming that the local $B_N$ representation uniquely fixes the $F$ and $R$ symbols of the Ising category.

3.5 Preclusion of Alternative Fusion Rules

Suppose one attempts to construct an alternative non-Abelian fusion rule for MZMs, e.g., $\sigma \times \sigma = 1 + \chi$ for some anyon $\chi \neq \psi$. The braid representation requires that $\sigma_i^2$ has exactly two eigenvalues (since the Majorana Clifford algebra gives a two-dimensional local fusion space for each pair). The eigenvalues of $\rho(\sigma_i)^2 = \gamma_{i+1}\gamma_i$ are $\pm 1$, corresponding to fermion parity. This forces the two fusion channels to have quantum dimensions $d_1 = 1$ and $d_\chi = 1$ (since $d_\sigma^2 = d_1^2 + d_\chi^2$ and $d_\sigma = \sqrt{2}$ from the representation dimension). Thus $d_\chi = 1$, and $\chi$ must be an Abelian anyon.

The topological spin of $\chi$ is constrained by the $R$-symbol eigenvalues. Since $\rho(\sigma_i)$ has eigenvalues $e^{-i\pi/8}$ and $e^{3i\pi/8}$, the balancing relation gives $\theta_\chi = R^{\sigma\sigma}_\chi R^{\sigma\sigma}_\chi / \theta_\sigma = e^{3i\pi/4} \cdot e^{3i\pi/4} / e^{i\pi/8} = e^{5i\pi/8 - i\pi/8} = e^{i\pi/2}$. But for a consistent super-modular category, the only Abelian anyons besides the vacuum are the transparent fermion $\psi$ (with $\theta_\psi = -1$) and possibly order-2 bosons. Since $e^{i\pi/2} \neq -1$, the anyon $\chi$ with $\theta_\chi = e^{i\pi/2}$ cannot be the transparent fermion, and it would require a nontrivial braiding with $\psi$, contradicting the transparency of $\psi$ in a super-modular category. Therefore, $\chi = \psi$ is the only consistent choice, and the fusion rule is uniquely $\sigma \times \sigma = 1 + \psi$.

4. Results

4.1 Local Determination of $F$, $R$, and $T$

Theorem 1. The $B_{2N}$ representation generated by $2N$ MZMs, $\rho(\sigma_i) = \frac{1}{\sqrt{2}}(1 + \gamma_{i+1}\gamma_i)$, uniquely determines (up to gauge equivalence) the $F$ and $R$ symbols of the Ising fusion category, and hence the $T$-matrix $T = \mathrm{diag}(1, e^{i\pi/8}, -1)$.

Proof. The eigenvalues of $\rho(\sigma_i)$ are $e^{-i\pi/8}$ and $e^{3i\pi/8}$, which are the $R$-symbol entries $R^{\sigma\sigma}_1$ and $R^{\sigma\sigma}_\psi$. The pentagon equation for the Ising fusion rules has a unique solution $F^{\sigma\sigma\sigma}_\sigma = 1/\sqrt{2}$ up to gauge [5]. The hexagon equations then fix all remaining $R$-symbols. The $T$-matrix follows from the balancing relation. $\square$

4.2 Non-Uniqueness of the Global $S$-Matrix

Theorem 2. The local $B_N$ representation does not uniquely determine the global modular $S$-matrix of the 2D TSC. The $S$-matrix depends on the spin structure of the underlying manifold.

Proof. The $S$-matrix of the Ising MTC is $S_{\mathrm{Ising}}$ as given in §3.3. However, the physical TO of a 2D TSC is the super-modular category $\mathrm{sIsing}$, whose Müger center is $\mathrm{sVec} = \{1, \psi\}$. The $S$-matrix of $\mathrm{sIsing}$ is degenerate: $S_{1a} = S_{\psi a}$ for all $a$. To obtain a non-degenerate modular $S$-matrix, one must extend to a spin-TQFT, which requires a choice of spin structure $\nu$ on the manifold. Different spin structures yield different partition functions and modular transformations [6]. The local $B_N$ representation, being defined in the bulk far from any boundary or handle, is insensitive to the global spin structure. $\square$

4.3 Classification of MZM Fusion Rules

Theorem 3. For strictly Majorana zero modes in 2D TSCs, the braid group representation restricts the fusion rules to the Ising type: $\sigma \times \sigma = 1 + \psi$, $\sigma \times \psi = \sigma$, $\psi \times \psi = 1$. No alternative non-Abelian fusion rules are consistent.

Proof. As shown in §3.5, the Majorana Clifford algebra forces $d_\sigma = \sqrt{2}$ and exactly two fusion channels for $\sigma \times \sigma$, both with $d = 1$. The $R$-symbol eigenvalues from the braid representation fix the topological spin of the second channel to $\theta = -1$, identifying it as the transparent fermion $\psi$. Any alternative assignment is inconsistent with the super-modular structure (transparency of $\psi$) or the hexagon equations. $\square$

5. Discussion

5.1 Gauge Ambiguity and Physical Observables

The finite gauge ambiguity in extracting $F$ and $R$ symbols from $B_N$ representations corresponds to basis rephasings in the fusion spaces. While this does not affect the $T$-matrix (topological spins are gauge-invariant), it can affect the off-diagonal entries of the $S$-matrix computed via the Hopf link invariant. In physical terms, this ambiguity manifests as a choice of framing for the anyon worldlines, which is unobservable in local braiding experiments but affects interferometric measurements that probe the full modular data [7].

5.2 Spin Structure and Global Modular Transformations

The dependence of the $S$-matrix on the spin structure reflects a fundamental distinction between bosonic and fermionic TOs. In a bosonic MTC, the modular data $(S, T)$ provides a projective representation of $\mathrm{SL}(2,\mathbb{Z})$ that is independent of the manifold. In a fermionic (super-modular) theory, the modular transformations on a genus-$g$ surface depend on the spin structure, which is a choice of quadratic form on $H_1(\Sigma_g, \mathbb{Z}_2)$ [6]. The local $B_N$ representation, being a bulk property, cannot distinguish between spin structures; this information enters only through the global topology.

This has practical consequences: interferometric experiments in 2D TSCs that aim to measure the $S$-matrix entries must account for the spin structure of the device geometry. For instance, the fusion channel of two $\sigma$ anyons encircling a handle depends on whether the handle carries NS or R boundary conditions [to verify: explicit experimental signature of spin-structure dependence in interferometry].

5.3 Generalized Majorana Zero Modes

A natural generalization considers $\mathbb{Z}_n$ parafermion zero modes, which arise in fractional topological insulators proximitized by superconductors. The braid group representation for $2N$ parafermions of order $n$ yields the $\mathbb{Z}_n$ parafermion TO with fusion rule $\sigma \times \sigma = \sum_{k=0}^{n-1} \psi^k$ (for $n$ even) or $\sigma \times \sigma = \sum_{k=0}^{(n-1)/2} (\psi^k + \psi^{n-k})$ (for $n$ odd) [to verify: exact parafermion fusion rules]. These representations map to distinct super-modular categories (e.g., the $\mathrm{SO}(n)_2$ categories for appropriate $n$) while potentially sharing the same local fusion space dimension. Whether the $B_N$ representation can distinguish between these categories, or whether the same local fusion rules admit multiple super-modular extensions, remains an open question.

5.4 Open Questions

1. Parafermion generalization: Can $\mathbb{Z}_n$ parafermion $B_N$ representations yield distinct super-modular categories with identical local fusion rules? The $\mathbb{Z}_n$ parafermion series provides a testing ground, as different values of $n$ yield different central charges but potentially similar local braiding data for small $N$.

1. Gauge ambiguity in observables: How does the finite gauge ambiguity in extracting $S$ and $T$ from $B_N$ representations manifest in physical observables? While topological spins are gauge-invariant, the full $S$-matrix entries probed by interferometry may depend on the gauge choice, suggesting that certain interferometric configurations are sensitive to the gauge.

1. Generalized hexagon equations: Is there a generalized hexagon equation for super-modular categories that directly yields the spin-structure-dependent $S$-matrix from local $B_N$ representations? Such a framework would bridge the gap between local braiding data and global modular data for fermionic TOs, potentially providing a fermionic analogue of the Verlinde formula.

6. Conclusion

We have shown that the braid group representation on $N$ anyons constrains the modular data of a non-Abelian topological order up to finite gauge ambiguity, but does not uniquely determine the global $S$-matrix without specifying the underlying spin structure. For Majorana zero modes in 2D topological superconductors, the braid representation $\rho(\sigma_i) = \frac{1}{\sqrt{2}}(1 + \gamma_{i+1}\gamma_i)$ uniquely fixes the local $F$ and $R$ symbols of the Ising category, thereby determining the $T$-matrix. The fusion rules are rigidly classified as Ising type ($\sigma \times \sigma = 1 + \psi$), and no alternative non-Abelian fusion rules are consistent with the Majorana Clifford algebra and the super-modular structure. The global $S$-matrix, however, depends on the spin structure of the underlying manifold, reflecting the fermionic nature of the topological order. This distinction between local braiding data and global modular data is a fundamental feature of fermionic topological phases and has direct implications for topological quantum computation and interferometric probes of anyonic statistics.

References

[1] M. Freedman, M. Larsen, and Z. Wang, "A Modular Functor Which is Universal for Quantum Computation," Commun. Math. Phys. 227, 605 (2002). arXiv:quant-ph/0001108.

[2] D. A. Ivanov, "Non-Abelian Statistics of Half-Quantum Vortices in p-Wave Superconductors," Phys. Rev. Lett. 86, 268 (2001). arXiv:cond-mat/0112071.

[3] N. Read and D. Green, "Paired states of fermions in two dimensions: Breaking the topological degeneracy of Landau levels," Phys. Rev. B 61, 10267 (2000). DOI:10.1103/PhysRevB.61.10267.

[4] E. Verlinde, "Fusion rules and modular transformations in 2D conformal field theory," Nucl. Phys. B 300, 360 (1988). DOI:10.1016/0550-3213(88)90603-7.

[5] E. Rowell and Z. Wang, "Mathematics of Topological Quantum Computing," Bull. Amer. Math. Soc. 55, 183 (2018). arXiv:0806.2483.

[6] A. Kitaev, "Anyons in an exactly solved model and beyond," Ann. Phys. 321, 2 (2006). arXiv:cond-mat/0506438.

[7] P. Bonderson, "Non-Abelian Anyons and Interferometry." arXiv:1208.2631.