---
title: Near-Group Fusion Algebras
url: https://www.emergentmind.com/topics/near-group-fusion-algebras
type: topic
---

# Near-Group Fusion Algebras

Searching arXiv for recent and foundational papers on near-group fusion rings/categories and related VOA examples.
Near-group fusion algebras are fusion rings obtained by adjoining a single non-invertible basis element to an integral group ring. In the standard notation \(R(G,\ell)\), where \(G\) is a finite group and \(\ell\in \mathbb{Z}_{\ge 0}\), the distinguished basis is \(G\cup\{\rho\}\) and the fusion rules are
\[
g h = gh,\qquad g\rho=\rho g=\rho,\qquad \rho^2=\ell\rho+\sum_{g\in G} g.
\]
Equivalent notation \(K(G,\alpha)\) is also used, with the same ring-theoretic content and with \(X^*=X\) for the non-invertible basis element. These rings are the simplest extensions of integral group rings by one non-invertible element, and they occupy a central place in the theory of fusion rings, fusion categories, modular tensor categories, and related VOA and subfactor constructions [2208.07319][2312.13796][2407.15955].

## 1. Classical form and basic invariants

A fusion ring is a unital based \(\mathbb{Z}_{\ge 0}\)-ring of finite rank. Concretely, it is a finitely generated free \(\mathbb{Z}\)-module with a distinguished basis, nonnegative integer structure constants, and an anti-involution compatible with duality. The integral group ring \(\mathbb{Z}G\) of a finite group \(G\) is the basic pointed example: its basis elements are all invertible and have Frobenius–Perron dimension \(1\). A near-group fusion ring is the corresponding “group plus one” extension, with basis \(G\cup\{\rho\}\), group multiplication on \(G\), trivial action of \(G\) on \(\rho\), and the single nontrivial rule \(\rho^2=\ell\rho+\sum_{g\in G}g\) [2208.07319].

The Frobenius–Perron dimension of the distinguished non-invertible basis element is the positive root of
\[
x^2-\ell x-|G|=0.
\]
Thus, if \(d=\mathrm{FPdim}(\rho)\), then
\[
d^2-\ell d-|G|=0.
\]
Near-group rings are therefore special cases of two-dimension fusion rings: all basis elements have Frobenius–Perron dimension in \(\{1,d\}\) for some \(d>1\) [2208.07319]. In the language of fusion categories, a near-group fusion category is a fusion category whose Grothendieck ring is \(R(G,\ell)\); its simple objects are the invertible objects indexed by \(G\) together with one distinguished non-invertible simple object.

The case \(\ell=0\) is the Tambara–Yamagami case. Here
\[
\rho^2=\sum_{g\in G} g,
\]
so the non-invertible object has Frobenius–Perron dimension \(\sqrt{|G|}\). Tambara–Yamagami categories are the most tractable non-pointed examples in this class and already exhibit many of the basic structural phenomena of near-group fusion algebras [2312.13796].

A useful broader perspective is that near-group rings are exactly the near-integral fusion rings \(R(\mathbb{Z}G,\kappa)\) for which the maximal-rank integral subring is pointed. In that formulation the basis is \(G\cup\{\rho\}\), the integral part is \(\mathbb{Z}G\), and the quadratic relation becomes \(\rho^2=\kappa\rho+\sum_{g\in G}g\). This recasts near-group rings as a pointed special case of a larger class characterized by a maximal-rank proper fusion subring [2407.15955].

The commutativity of a near-group fusion algebra is controlled entirely by the group part. The based ring \(NG(G,m)\) is commutative if and only if \(G\) is abelian; if \(G\) is non-abelian, then the group subring \(\mathbb{Z}G\) is already noncommutative, so the full near-group ring is noncommutative as well [1908.01655].

## 2. Structural constraints and categorifiability

Near-group fusion algebras are easy to write down, but categorifiability is highly constrained. For general two-dimension fusion rings with quadratic relation
\[
d^2-rd-s=0,
\]
one basic dichotomy is that if \(d\in\mathbb{Z}\), then \(r<s\), while \(r\ge s\) forces \(d\) irrational. Applied to near-group rings, this yields: if \(\ell\ge |G|\), then \(\mathrm{FPdim}(\rho)\) is irrational. Moreover, if a categorifiable near-group ring has irrational Frobenius–Perron dimension, then \(|G|\mid \ell\) [2208.07319].

This divisibility condition is one of the main structural restrictions. For near-group rings \(R(G,\ell)\), irrational categorifiable cases must have
\[
\ell=k|G|,\qquad k\in\mathbb{Z}_{\ge 1}.
\]
The paper on categorification of integral group rings extended by one dimension develops this in the broader setting of two-dimension and two-orbit fusion rings, using pseudounitarity, formal codegrees, induction to the Drinfeld center, Galois action, and number-theoretic bounds on sums of roots of unity [2208.07319].

A complete classification is obtained there for elementary abelian \(2\)-groups. If \(G=C_2^m\), then \(R(G,\ell)\) is categorifiable if and only if
\[
\ell=0;\qquad G=C_2\text{ and }\ell=1\text{ or }\ell=2;\qquad G=C_2^2\text{ and }\ell=4.
\]
Equivalently, for \(m\ge 3\), the only categorifiable near-group ring is the Tambara–Yamagami case \(\ell=0\). This gives an infinite family in which the near-group ansatz exists ring-theoretically but almost never admits a fusion-category realization [2208.07319].

An older classification for abelian \(G\) in the \(C^*\)-setting imposes a different but related dichotomy. For a near-group \(C^*\)-category of type \(G+n'\) with \(H^2(G;\mathbb{T})=1\), either
\[
n'=|G|-1\qquad\text{or}\qquad n'\in |G|\mathbb{Z}.
\]
Evans and Gannon call such systems of type \(G+n'\). In the case \(n'=|G|-1\), \(G\) must be cyclic, and for most \(n\) there is exactly one \(C^*\)-category of type \(\mathbb{Z}_n+(n-1)\), namely \(\mathrm{Mod}(\mathrm{Aff}_1(\mathbb{F}_q))\) with \(q=n+1\), except for \(n=1,2,3,7\), which have additional exceptional categories [1208.1500].

The same work studies the second class \(n'=kn\), especially \(n'=n\), and constructs over 40 new finite depth subfactors with Jones index ranging from around 6.85 to around 14.93. In that regime, the near-group fusion algebra is no longer a sporadic phenomenon but part of a substantial infinite landscape tied to \(2^n1\) principal graphs and Izumi-type systems [1208.1500].

## 3. Centers, modular data, and Frobenius–Schur indicators

The Drinfeld center of a near-group fusion category is a modular tensor category, and its modular data often admit unexpectedly explicit descriptions. For near-group \(C^*\)-categories of type \(G+n'\), Evans and Gannon compute the tube algebra and the corresponding doubles, obtaining modular data relevant to conformal field theory. For type \(G+n\) with \(G\) of odd order, they conjecture and verify in many cases that the modular data of the double are controlled by quadratic forms on \(G\) and on a second group \(G'\) of order \(|G|+4\) [1208.1500].

A concrete modern example is the near-group fusion category \(\mathcal{A}\) of type \(\mathbb{Z}_3+6\). Its simple objects are the invertible objects of \(\mathbb{Z}_3\) together with a unique non-invertible object \(X\), and
\[
X\otimes X = e\oplus g\oplus g^2\oplus 6X.
\]
Yu proves that its center is
\[
\mathcal{Z}(\mathcal{A})\cong\mathcal{C}(\mathbb{Z}_3,\eta)\boxtimes\mathcal{C}(\mathfrak{sl}_3,9)_{\mathbb{Z}_3}^0,
\]
and also constructs two non-trivial faithful extensions of \(\mathcal{A}\) explicitly, whose Drinfeld centers can also be obtained from representation categories of quantum groups at root of unity [2312.12195]. This is a particularly transparent instance in which a non-pointed near-group category has a center decomposing into a pointed factor and a quantum-group factor.

Higher Frobenius–Schur indicators supply another layer of structure. For near-group categories with \(m=|G|-1\), indicators of the non-invertible simple object can be written in terms of \(\theta_k^G(e)\), the number of \(k\)-th roots of the identity in \(G\), together with small correction terms coming from the center. Tucker proves that the near-group fusion rings \(NG(G,|G|-1)\) exhibit Frobenius–Schur indicator rigidity: all categorifications are distinguished by their Frobenius–Schur indicators [1510.05696].

For near-group categories with \(m=|G|\) and \(|G|\) odd, the same paper derives indicator formulas controlled by quadratic Gauss sums associated to a metric group \((G,q)\) and a second metric group \((G',q')\) of order \(|G|+4\). In all known examples with \(m=|G|\), the non-invertible object has Frobenius–Schur indicators given by quadratic Gauss sums. This ties the center, modular data, and arithmetic of quadratic forms directly to the intrinsic fusion algebra [1510.05696].

Minimal modular extensions sharpen this picture. Schopieray studies weakly integral braided fusion categories with elementary fusion rules and classifies near-group braided fusion categories satisfying the minimal modular extension conjecture. A braided near-group fusion category possesses a minimal modular extension if and only if it is braided equivalent to one of four types: \(\mathcal{Z}(\mathrm{Vec}_{Q_8}^\gamma)_\mathrm{ad}\) or its reverse, a braided Tambara–Yamagami category \(\mathcal{C}(\chi_n^1,\tau)\), a symmetrically braided near-group fusion category, or one of seven nonsymmetrically braided near-group fusion categories described explicitly in the paper [2109.04011]. The remaining Tambara–Yamagami braided fusion categories then yield arbitrarily large families of braided fusion categories with identical fusion rules violating the minimal modular extension conjecture.

## 4. Generalized near-group and near-integral frameworks

The classical near-group condition allows exactly one non-invertible simple object. A systematic generalization keeps the group action but permits multiple non-invertibles, provided the group of invertible objects acts transitively on them. In Dong’s formulation, a generalized near-group fusion category is a fusion category \(\mathcal{C}\) such that \(G(\mathcal{C})\) acts transitively on the set of non-isomorphic non-invertible simple objects. If \(X_1\) is a chosen non-invertible simple object and \(T=G[X_1]\) its stabilizer, then
\[
X_1\otimes X_1^*=\bigoplus_{h\in T} h\oplus k_1X_1\oplus\cdots\oplus k_nX_n,
\]
and the category is said to be of type \((G,T,k_1,\dots,k_n)\) [1903.08791].

The classical near-group case is recovered when there is exactly one non-invertible simple object: then \(G=T\), and the defining rule reduces to
\[
X\otimes X^*=\bigoplus_{g\in G} g\oplus k_1X,
\]
which matches the usual near-group formula \(X\otimes X=\bigoplus_{g\in G} g\oplus mX\) [1903.08791]. The generalized Tambara–Yamagami case corresponds to \((k_1,\dots,k_n)=(0,\dots,0)\), so all products of non-invertibles decompose purely into invertibles.

This broader framework leads to a structural classification in the slightly degenerate braided setting. Slightly degenerate generalized near-group fusion categories are exactly one of the following: \(\mathcal{I}\boxtimes\mathcal{B}\), where \(\mathcal{I}\) is an Ising category and \(\mathcal{B}\) is slightly degenerate pointed; \(\mathcal{C}_{\mathrm{ad}}\boxtimes\mathcal{C}_{\mathrm{pt}}\), where \(\mathcal{C}_{\mathrm{ad}}\) is a Yang–Lee category; \(\mathcal{C}_{\mathrm{ad}}\boxtimes\mathcal{B}\), where \(\mathcal{C}_{\mathrm{ad}}\) is a slightly degenerate fusion category of the form \(\mathcal{C}(\mathfrak{psl}_2,q^t,8)\) and \(\mathcal{B}\) is non-degenerate pointed; or a prime category generated by a \(\sqrt{2}\)-dimensional simple object [1903.08791]. A companion paper develops the same theory further, including an explicit example arising from an extension of a Fibonacci category and a classification of braided generalized Tambara–Yamagami categories of dimension \(2^m\) [2004.08735].

A complementary unification comes from near-integral fusion. There a near-integral fusion ring is one with a proper fusion subring of maximal rank. Such a ring is determined by an integral fusion subring \(S\) and one extra self-dual basis element \(\rho\), with
\[
\rho^2=\kappa\rho+\sum_{x\in S}\mathrm{FPdim}(x)\,x.
\]
Near-group fusion rings are exactly the case \(S=\mathbb{Z}G\). The character-theoretic description is especially sharp: a fusion ring is near-integral if and only if there exists a ring homomorphism whose kernel is a maximal-rank proper fusion subring, and near-group rings are the pointed instances of this phenomenon [2407.15955].

## 5. NIM-representations, algebra objects, and noncommutative realizations

The near-group relation \(X^2=\sum_{g\in G}g+\alpha X\) is also an efficient starting point for the study of module categories and algebra objects. In the NIM-rep approach, a near-group fusion ring \(K(G,\alpha)\) acts on a \(\mathbb{Z}_+\)-module with basis \(M\), and the action of the non-invertible element \(X\) is encoded by a symmetric matrix \(C=(c_{i,j})\), while the \(G\)-action decomposes \(M\) into \(G\)-orbits governed by subgroups \(H_i\subseteq G\). The basic constraints are
\[
CBC=\alpha C+|G|B^{-1},
\]
where \(B=\mathrm{diag}(|G:H_1|,\dots,|G:H_p|)\) [2312.13796].

This matrix equation completely controls the low-orbit cases. For one orbit, NIM-reps are parametrized by pairs \((H,c_{1,1})\) satisfying
\[
\alpha=c_{1,1}|G:H|-\frac{|H|}{c_{1,1}},
\]
with divisibility and positivity constraints. For two orbits, they are parametrized by tuples \((H_1,H_2,c_{1,1},c_{2,2})\) satisfying
\[
\alpha=c_{1,1}|G:H_1|+c_{2,2}|G:H_2|,
\]
\[
|G|\mid |H_1||H_2|,
\]
and
\[
\frac{|H_1||H_2|}{|G|}+c_{1,1}c_{2,2}
\]
a perfect square [2312.13796]. In the Tambara–Yamagami case \(\alpha=0\), every irreducible NIM-rep has at most two group orbits.

Admissible NIM-reps reconstruct algebra objects. In the one-orbit case, the associated algebra object has the form
\[
A\cong \left(\bigoplus_{h\in H} b_h\right)\oplus c_{1,1}X.
\]
In the two-orbit admissible case \((H_1,H_2,0,c_{2,2})\), the algebra object is purely group-like:
\[
A\cong \bigoplus_{h\in H_1} b_h.
\]
This recovers known classifications of algebra objects in near-group, especially Tambara–Yamagami, categories from a fusion-ring and NIM-rep viewpoint [2312.13796].

The noncommutative side is far more rigid. For a non-abelian finite group \(G\), a unitary near-group fusion category with Grothendieck ring \(NG(G,m)\) exists if and only if \(G\) is an extra-special \(2\)-group. In that case \(|G|=2^{2n+1}\), \(m=2^n\), and \(\mathrm{FPdim}(\rho)=2^{n+1}\). Moreover, for each extra-special \(2\)-group \(G\) there are exactly three inequivalent unitary near-group categories, distinguished by the third Frobenius–Schur indicator \(\nu_3(\rho)\) [1908.01655].

Izumi and Tucker give a purely algebraic construction of all these noncommutative near-group fusion categories from pointed categories categorically Morita equivalent to them. Their categories are of the form
\[
\mathcal{C}_{n,l}=(G_n,\omega_n^l,H_n,1),
\]
with \(G_n\) built from \(\mathbb{F}_2^{2n}\rtimes S_3\), \(H_n\cong \mathbb{Z}_2^{\,n-1}\times \mathbb{Z}_4\), and \(\omega_n^l\) inflated from a generator of \(H^3(S_3,\mathbb{C}^\times)\cong\mathbb{Z}_6\). The group of invertible objects in \(\mathcal{C}_{n,l}\) is the central product of \(n\) copies of \(D_8\) when \(l\) is even and the central product of one copy of \(Q_8\) and \(n-1\) copies of \(D_8\) when \(l\) is odd, while the unique non-invertible simple object satisfies
\[
X_{\gamma_2,1}\otimes X_{\gamma_2,1}\cong 2^nX_{\gamma_2,1}\oplus \bigoplus_{Y\in\Gamma(\mathcal{C}_{n,l})}Y.
\]
This realizes the full noncommutative near-group classification group-theoretically [1908.01655].

## 6. Pointed limits, VOA realizations, and adjacent phenomena

Near-group fusion algebras are best understood against the backdrop of pointed fusion rings. A pointed fusion ring is a fusion ring in which every simple basis element is invertible, equivalently an integral group ring \(\mathbb{Z}G\). Several papers in the area make this contrast explicit: near-group rings are “near” the pointed case because they add exactly one non-invertible simple object to a group ring, while preserving a large invertible sector [2208.07319].

A clean VOA example of the pointed limit is the even part \(F^+\) of the symplectic fermion vertex operator superalgebra. Its simple modules are
\[
F^+,\quad F^-,\quad F_t^+,\quad F_t^-,
\]
and their fusion algebra is isomorphic to the group algebra of the Klein four group:
\[
\mathcal{F}(F^+)\cong \mathbb{Z}[\mathbb{Z}_2\times\mathbb{Z}_2].
\]
All simple modules are invertible, so this is not a genuine near-group fusion algebra but rather a pointed baseline from the VOA side [1108.1823]. This example is useful precisely because it isolates the group-like sector without any extra non-invertible simple object.

A more general pointed-orbifold criterion appears for cyclic orbifolds of lattice VOAs. If \(L\) is an even positive-definite lattice and \(g\in O(L)\), then the orbifold VOA \(V_L^{\hat g}\) has group-like fusion if and only if \(g\) acts trivially on the discriminant group \(L^*/L\), equivalently
\[
(1-g)L^*\subset L.
\]
When this holds and \(g\) is fixed point free, the representation category is a pointed modular tensor category completely determined by its quadratic space structure [1805.10778]. This again identifies a supply of explicitly computable group-like fusion rings that can serve as the invertible sector in broader near-group constructions.

A recent physics-oriented development studies near-group fusion algebras as commutative non-invertible fusion algebras of type \(G+n'\), where \(G\) is finite Abelian and
\[
g_i\rho=\rho g_i=\rho,\qquad \rho^2=n'\rho+\sum_{g_i\in G} g_i.
\]
The paper emphasizes Fibonacci, Ising, and \(\mathrm{Rep}(S_3)\) fusion rules as concrete instances, develops a spurion-labeling scheme for coupling constants directly at the level of the near-group fusion algebra, and interprets radiative violations of tree-level non-invertible selection rules as “loop-induced groupification” [2508.14970]. This does not alter the algebraic definition, but it shows that near-group fusion algebras now function as organizing principles well beyond their original categorical setting.

Taken together, these developments place near-group fusion algebras at a nexus of pointed fusion theory, categorifiability, module-category reconstruction, Drinfeld-center computations, subfactor and quantum-group realizations, and recent applications of non-invertible fusion rules. The pointed case supplies the ambient metric-group data, while the single non-invertible generator encodes the first genuinely non-pointed deformation of the group-ring paradigm.

Source: https://www.emergentmind.com/topics/near-group-fusion-algebras