---
title: Isaacs Fusion Subcategory
url: https://www.emergentmind.com/topics/isaacs-fusion-subcategory
type: topic
---

# Isaacs Fusion Subcategory

An **Isaacs fusion subcategory** is, in the usage of the recent literature, the fusion subcategory generated by a simple object \(X\) when that generated subcategory satisfies the Isaacs integrality condition; more generally one speaks of an **\(s\)-Isaacs fusion subcategory** when it satisfies the \(s\)-Isaacs condition [2507.07329]. The notion is therefore local rather than ambient: it does not name a canonical construction such as a pointed, adjoint, or centralizer subcategory, but instead records an arithmetic property of the smallest fusion subcategory \(\langle X\rangle\) containing a given simple object. Its significance lies in the fact that this local Isaacs condition yields divisibility constraints on \(\dim(X)\) and \(\operatorname{FPdim}(X)\), with especially strong consequences in ribbon and modular settings [2507.07329].

## 1. Definition and formal setting

The operative definition is formulated for a fusion category \(\mathcal C\) through the subcategory generated by a simple object \(X\). The notation
\[
\langle X\rangle
\]
denotes the smallest fusion subcategory containing \(X\), equivalently the full tensor subcategory generated by tensor powers of \(X\) and \(X^*\), closed under subobjects and finite direct sums [2507.07329].

For a fusion category with commutative Grothendieck ring \(K(\mathcal C)=K_0(\mathcal C)\otimes_{\mathbb Z}\mathbb C\), the paper [2507.07329] recalls the \(s\)-Isaacs condition from earlier work. Writing \(\mu_j:K(\mathcal C)\to \mathbb C\) for a linear character, \(\mathcal C^j\) for the corresponding conjugacy class, and \(\mathbb A\) for the ring of algebraic integers, \(\mathcal C\) is called **\(s\)-Isaacs** when
\[
\lambda_s(\mu_j,X):= (\dim \mathcal C)^s \, \dim(\mathcal C^j)^{\,1-s}\, \frac{\mu_j([X])}{\dim(X)} \in \mathbb A
\]
for every linear character \(\mu_j\) and every simple object \(X\) [2507.07329]. The case \(s=0\) is the ordinary **Isaacs property**.

In this language, an **Isaacs fusion subcategory** is simply a generated subcategory \(\langle X\rangle\) that is \(0\)-Isaacs, and an **\(s\)-Isaacs fusion subcategory** is a generated subcategory \(\langle X\rangle\) that is \(s\)-Isaacs [2507.07329]. Thus the term does not introduce a new intrinsic type of subcategory construction; it specifies that a singly generated fusion subcategory satisfies a particular integrality condition.

## 2. Isaacs property as the underlying arithmetic condition

The subcategory notion rests on a broader theory of the Isaacs property for fusion rings and fusion categories. In the fusion-ring formulation, a fusion ring \((R,B)\) with basis \(B=\{x_0,\dots,x_m\}\) is called **Isaacs** if
\[
\omega_i(S^\rho)\in \mathbb A
\]
for all basis elements \(x_i\in B\) and all irreducible representations \(\rho\in \operatorname{Irr}(R_{\mathbb C})\), where \(S^\rho\) is the abstract matrix class sum obtained from the Fourier transform and \(\omega_i\) is the associated central character [2210.13936]. A central explicit formula is
\[
\omega_i( S^\rho)=\frac{\operatorname{FP}(R)}{\operatorname{FP}(x_i)} \frac{\rho(x_i)}{c_\rho},
\]
with \(c_\rho\) the formal codegree [2210.13936].

For pseudo-unitary fusion categories, this ring-theoretic condition coincides with the categorical Isaacs property. More precisely, if \(R=K_0(\mathcal C)\), then
\[
I_0(\rho, X_i)=\omega_i(S^\rho),
\]
so the categorical \(0\)-Isaacs condition is exactly the abstract Isaacs condition on the Grothendieck ring [2210.13936]. In this sense, the arithmetic content of an Isaacs fusion subcategory is already encoded by the Grothendieck ring of \(\langle X\rangle\) whenever pseudo-unitarity is available.

In the commutative case, the Isaacs property occupies a specific place in an arithmetic hierarchy:
\[
\text{integrality of structure constants} \;\Longrightarrow\; \text{Isaacs} \;\Longrightarrow\; 1\text{-Frobenius},
\]
and both implications are strict [2210.13936]. This makes the Isaacs condition stronger than basic Frobenius-type divisibility but weaker than full algebraic integrality of matrix-class-sum structure constants.

The property is also restrictive. The Extended Haagerup fusion categories \(\mathcal{EH}_i\) do not satisfy the Isaacs property, giving a negative answer to a question of Etingof–Nikshych–Cuadra and recovering that \(\mathcal{EH}_1\) has no braiding [2210.13936]. Moreover, Isaacs is not Morita invariant [2210.13936]. Consequently, calling a generated subcategory “Isaacs” is a genuine extra condition, not a formal consequence of exoticity, Morita class, or general fusion-ring behavior.

## 3. Generated subcategories and the associated character data

The theory in [2507.07329] attaches several character-theoretic invariants to a simple object \(X\). For a fusion category \(\mathcal C\) with commutative Grothendieck ring, the **kernel** of \(X\) is
\[
\ker_{\widehat{K(\mathcal C)}}(X):=\{\psi\in \widehat{K(\mathcal C)}\mid \psi([X])=\operatorname{FPdim}(X)\},
\]
and the **center** of \(X\) is
\[
Z_{\mathcal C}(X)=\{\psi\in \widehat{K(\mathcal C)}\mid |\psi([X])|=\operatorname{FPdim}(X)\}.
\]
The set \(Z_{\mathcal C}(X)\) plays the role of an object-wise center and is the quantity that appears in the main divisibility theorems [2507.07329].

When the ambient category is the generated category itself, this center simplifies sharply:
\[
Z_{\langle X\rangle}(X)=G(\widehat{K(\langle X\rangle)}),
\]
where \(G(\widehat{K(\langle X\rangle)})\) denotes the group-like elements of the dual hypergroup [2507.07329]. Since
\[
G(\widehat{K(\mathcal D)})\cong U(\mathcal D),
\]
this identifies the center-set of the generator with the universal grading group of the generated subcategory [2507.07329]. In this way, an Isaacs fusion subcategory is closely tied to the grading-theoretic structure of \(\langle X\rangle\).

There is also a relation to the adjoint subcategory. Under the hypothesis \(\operatorname{FPdim}(X)=\dim(X)\), one has
\[
Z_{\mathcal C}(X)\subseteq (\langle X\rangle_{ad})^\perp,
\]
and for simple \(X\),
\[
Z_{\langle X\rangle}(X)=\big(\langle X\rangle_{ad}\big)^\perp
\]
[2507.07329]. Thus the center-set of a generator is controlled by the adjoint part of the generated subcategory. This is a structural description rather than a merely numerical one: it places Isaacs fusion subcategories at the intersection of character theory, universal grading, and adjoint decomposition.

A further identity links the ambient category \(\mathcal C\) and the generated subcategory \(\mathcal D=\langle X\rangle\):
\[
\frac{\dim(\mathcal C)}{n_{\dim}(Z_{\mathcal C}(X))} = \frac{\dim(\mathcal D)}{|U(\mathcal D)|}.
\]
This Bantay-type formula is one of the key bridges used to transfer divisibility information from the generated subcategory back to the ambient category [2507.07329].

## 4. Divisibility theorems for Isaacs fusion subcategories

The main purpose of the notion is arithmetic. If \(X\) is simple and \(\langle X\rangle\) is Isaacs, then the dimension of \(X\) is forced to divide a controlled quotient of the ambient dimension. In the spherical setting, the refined theorem states:
\[
\dim(X)\mid \frac{\operatorname{FPdim}(\mathcal C)}{n(Z_{\mathcal C}(X))},
\]
where \(n(Z_{\mathcal C}(X))\) is the order attached to the center-set of \(X\) [2507.07329].

Under the additional hypothesis that \(\widehat{K(\mathcal C)}\) is **real non-negative (RN)**, this simplifies to
\[
\dim(X)\mid \frac{\operatorname{FPdim}(\mathcal C)}{|U(\mathcal C)|}
\]
[2507.07329]. The RN condition is available in important classes, including pseudo-unitary braided fusion categories and unitary fusion categories [2507.07329].

The same paper proves broader \(s\)-Isaacs refinements for \(s\ge 0\) and stronger square-divisibility-type statements for \(s\ge \tfrac12\), again expressed through \(Z_{\mathcal C}(X)\), \(n_{\dim}(Z_{\mathcal C}(X))\), and \(U(\mathcal C)\) [2507.07329]. The ordinary Isaacs case \(s=0\) is the basic specialization relevant to the terminology.

One important consequence is global. If every fusion subcategory generated by a simple object is Isaacs, then \(\mathcal C\) is Frobenius type [2507.07329]. A plausible implication is that Isaacs fusion subcategories provide a local mechanism for Frobenius-type divisibility phenomena, complementing the strong Frobenius property established for weakly group-theoretical fusion categories [0809.3031].

## 5. Ribbon and modular specializations

The broadest source of examples comes from braiding. The paper [2507.07329] recalls that every ribbon fusion category is Isaacs. Therefore, in a pseudo-unitary ribbon category, every simple object \(X\) automatically generates an Isaacs fusion subcategory, and one obtains the divisibility theorem
\[
\operatorname{FPdim}(X)\mid \frac{\operatorname{FPdim}(\mathcal C)}{|U(\mathcal C)|}
\]
[2507.07329]. This is presented as a categorical analogue of Isaacs’ classical theorem that the degree of an irreducible character of a finite group divides the index of the center.

The modular case is stronger. If \(\mathcal C\) is modular and generated by a simple object \(X\), so that
\[
\mathcal C=\langle X\rangle,
\]
then
\[
\dim(X)^2\mid \frac{\dim(\mathcal C)}{|U(\mathcal C)|}
\]
[2507.07329]. The proof uses modular data: for simple objects \(X_i\), one has
\[
\dim(\mathcal C^{i})=d_i^2,\qquad d_i=\dim(X_i),
\]
and
\[
\mu_{i'}([X_i])=\frac{s_{ii'}}{d_{i'}},
\]
with \(S=(s_{ij})\) the modular \(S\)-matrix [2507.07329]. The square-divisibility statement is therefore genuinely modular rather than merely spherical.

This stronger modular theorem is used to obtain the converse direction of an Ito–Michler-type result for weakly integral modular categories. If
\[
\operatorname{FPdim}(\mathcal C)=p^\alpha N,\qquad \alpha\ge 1,\quad (p,N)=1,
\]
then
\[
p \text{ does not divide the Frobenius–Perron dimension of any simple object}
\]
if and only if
\[
p^\alpha \mid |U(\mathcal C)|
\]
[2507.07329]. In this sense, Isaacs fusion subcategories are not merely a local curiosity: they feed directly into prime-divisibility criteria for modular categories.

## 6. Terminological scope and relation to other subcategory notions

The expression **Isaacs fusion subcategory** is not a long-established standard term across the fusion-category literature. The paper that develops the abstract and categorical Isaacs property for fusion rings and pseudo-unitary fusion categories does not define an “Isaacs fusion subcategory”; it studies the Isaacs property itself, not a special class of subcategories [2210.13936]. Likewise, works on other major subcategory formalisms use different organizing principles.

For equivariantizations, fusion subcategories are parameterized by triples \((\mathcal E,H,\eta)\) consisting of a \(G\)-invariant fusion subcategory, a normal subgroup, and a \(G\)-equivariant trivialization [2111.09116]. For \(SU(N)_k\)-type braided categories, the canonical non-pointed factor is
\[
\mathcal{M}SU(N)_k=\bigoplus_{i=0}^{n-1}(SU(N)_k)^{im},
\]
equivalently the centralizer of a pointed modular factor; this is a grading- and centralizer-defined construction, not an Isaacs subcategory [2202.00121]. In the modular-extension literature, a different notion is prominent: a **Galois-closed** or **Galois-stable** fusion subcategory, characterized by stability under the Galois action on modular data or, equivalently, by integrality of the relative centralizer in a modular extension [2601.23192]. In the Haagerup and Extended Haagerup settings, the relevant “subgroup-like” structures are often simple module categories or Morita-equivalent categories rather than fusion subcategories with Isaacs-type arithmetic constraints [1102.2631; 1810.06076].

Accordingly, an Isaacs fusion subcategory should be understood narrowly and precisely. It is not a canonical subcategory extracted functorially from a fusion category, nor is it synonymous with a pointed, adjoint, Müger-centralizer, equivariant, or Galois-closed subcategory. It is a **generated fusion subcategory \(\langle X\rangle\) carrying the Isaacs integrality property**, introduced to convert local character-theoretic integrality into explicit divisibility theorems for the generating simple object [2507.07329].

Source: https://www.emergentmind.com/topics/isaacs-fusion-subcategory