---
title: Irreducible Character Immanants in Matrix Theory
url: https://www.emergentmind.com/topics/irreducible-character-immanants
type: topic
---

# Irreducible Character Immanants in Matrix Theory

Searching arXiv for the cited papers to ground the article in current records.
Irreducible character immanants are matrix functions attached to irreducible characters of symmetric groups. For a partition $\lambda \vdash n$, with irreducible character $\chi^\lambda$ of $S_n$, the associated immanant of an $n\times n$ matrix $A=[a_{ij}]$ is
\[
\operatorname{Imm}^{\lambda}(A)=\sum_{\sigma\in S_n}\chi^\lambda(\sigma)\prod_{i=1}^n a_{i,\sigma(i)}.
\]
This construction interpolates between the determinant and the permanent: when $\lambda=(1^n)$, $\chi^\lambda$ is the sign character and the immanant becomes $\det(A)$, while when $\lambda=(n)$, $\chi^\lambda$ is the trivial character and the immanant becomes $\operatorname{perm}(A)$ [2309.16647]. Recent work places irreducible character immanants at the intersection of symmetric-group representation theory, Young-diagram combinatorics, partition-algebra generalizations, quasisymmetric refinements, and quantum/Capelli analogues [2402.05710].

## 1. Classical definition and special cases

The classical immanant is defined by weighting permutation monomials with irreducible character values of $S_n$:
\[
\operatorname{Imm}^{\lambda}(A) = \sum_{\sigma\in S_n} \chi^\lambda(\operatorname{ctype}(\sigma)) \prod_{i=1}^n a_{i,\sigma(i)},
\]
where $\operatorname{ctype}(\sigma)$ denotes the cycle type of $\sigma$ [2501.15667]. Because irreducible characters are constant on conjugacy classes, the coefficient depends only on cycle type, even though the sum itself is indexed by permutations.

The determinant and permanent are the extremal examples recorded in the recent literature. For $\lambda=(1^n)$ one obtains the determinant, and for $\lambda=(n)$ one obtains the permanent [2309.16647]. In this sense, irreducible character immanants form a family of permutation-indexed matrix invariants that interpolate between alternating and totally symmetric coefficient systems.

This formulation is the one retained across several extensions. The partition-algebra paper writes the classical object as
\[
\operatorname{Imm}^\lambda(A) = \sum_{\sigma\in S_n}\chi^\lambda(\sigma)\prod_{i=1}^n a_{i,\sigma(i)},
\]
and uses it as the specialization target for a broader construction [2309.16647]. The quasi-immanant paper likewise takes the irreducible character immanant as its starting point before refining the coefficient system from cycle types to cycle compositions [2501.15667].

## 2. Indexing by partitions, Young diagrams, and symmetric functions

Irreducible characters $\chi^\lambda$ of $S_n$ are indexed by partitions $\lambda\vdash n$, equivalently by Young diagrams $\boldsymbol{\lambda}$ [2402.05710]. This indexing is not merely notational: the combinatorics of the partition controls concrete properties of the corresponding immanant, including vanishing phenomena on structured matrix spaces.

The symmetric-function organization of these coefficients is expressed through the Frobenius characteristic map
\[
\mathrm{Frob}(\theta)=\frac{1}{n!}\sum_{\sigma\in S_n}\theta(\sigma)\,p_{\operatorname{ctype}(\sigma)},
\]
under which
\[
\chi^\lambda \longmapsto s_\lambda,\qquad
s_\lambda=\frac{1}{n!}\sum_{\sigma\in S_n}\chi^\lambda(\operatorname{ctype}(\sigma))\,p_{\operatorname{ctype}(\sigma)}.
\]
Thus the irreducible character coefficients appearing in immanants are exactly the power-sum coefficients of Schur functions [2501.15667]. This point is central in later generalizations: some constructions replace $\chi^\lambda(\operatorname{ctype}(\sigma))$ by coefficients attached to other symmetric or quasisymmetric bases, while preserving the permutation-monomial part of the definition.

A recurrent theme is that the partition $\lambda$ should be viewed simultaneously in three ways: as the label of an irreducible $S_n$-character, as a Young diagram governing rim-hook combinatorics, and as the Schur-function index under Frobenius. This suggests that irreducible character immanants are best understood as representation-theoretic matrix functions whose analytic behavior is often encoded diagrammatically.

## 3. Vanishing on alternating matrices

A recent classification determines exactly which irreducible character immanants vanish identically on the space of alternating complex matrices [2402.05710]. For
\[
\mathbb{A}_n(\mathbb{C})=\{A=(a_{ij})\in \mathbb{M}_n(\mathbb{C}) : a_{ij}=-a_{ji}\ \text{for } i<j,\ a_{ii}=0\},
\]
and for an irreducible character $\chi$ of $S_n$,
\[
d_\chi(A)=\sum_{\sigma\in S_n}\chi(\sigma)\prod_{i=1}^n a_{i\sigma(i)},
\]
the identities
\[
d_\chi(A^T)=d_\chi(A),\qquad A^T=-A,
\]
imply
\[
d_\chi(A)=d_\chi(A^T)=d_\chi(-A)=(-1)^n d_\chi(A).
\]
Hence if $n$ is odd, then $d_\chi(A)=0$ for every alternating matrix $A$ [2402.05710].

The even-dimensional case is the substantive one. A reduction due to Duffner et al. shows that, for even $n$, only permutations with no odd cycles matter on alternating matrices. Writing $P_n\subseteq S_n$ for the set of permutations whose cycle type has no odd cycles, one has
\[
d_\chi(A)=\sum_{\sigma\in P_n}\chi(\sigma)\prod_{i=1}^n a_{i\sigma(i)}
\]
for alternating $A$ [2402.05710]. Therefore the problem of deciding whether an irreducible immanant vanishes identically on $\mathbb{A}_n(\mathbb{C})$ reduces to deciding whether the corresponding character vanishes on all of $P_n$.

The main classification theorem states that for an irreducible character $\chi$ of $S_{2n}$,
\[
\chi(\rho)=0\ \forall\,\rho\in P_{2n}
\quad\Longleftrightarrow\quad
\chi \text{ is induced by an indestructible diagram},
\]
and consequently
\[
d_\chi\equiv 0 \text{ on }\mathbb{A}_n(\mathbb{C})
\]
if and only if either $n$ is odd, or $n$ is even and $\chi$ is induced by an indestructible diagram [2402.05710]. The same paper also shows that, for even $n$, the condition
\[
d_\chi(J\oplus\cdots\oplus J)=0,\qquad
J=\begin{pmatrix}0&1\\-1&0\end{pmatrix},
\]
is equivalent to $\chi$ being induced by an indestructible diagram, giving a concrete test matrix for the vanishing property [2402.05710].

A technical addendum in the same work corrects an earlier statement on immanant-converting maps. The automatic linearity and bijectivity of a map $\Phi$ satisfying
\[
d_\chi(A+\lambda B)=d_{\chi'}(\Phi(A)+\lambda\Phi(B))
\]
requires an extra hypothesis on $\chi$, namely that for every $i,j$ there exists $\sigma\in S_n$ with $\sigma(i)=j$ and $\chi(\sigma)\neq 0$; the paper proves this holds automatically when $\operatorname{char}(\mathbb{F})\neq 2$, but may fail in characteristic $2$ [2402.05710].

## 4. Domino rim-hooks, tilings, and the partition criterion

The classification of vanishing immanants on alternating matrices is combinatorial. A domino is a $2\times 1$ or $1\times 2$ pair of edge-adjacent cells, and a domino rim-hook is a rim-hook of length $2$, equivalently a removable horizontal or vertical domino that leaves a valid Young diagram. A Young diagram is called destructible if one can recursively remove domino rim-hooks until nothing remains; otherwise it is indestructible [2402.05710].

Two structural lemmas organize this recursive removal process. First, if a diagram has two disjoint domino rim-hooks $D$ and $E$, then $E$ remains a domino rim-hook after removing $D$. Second, for a diagram of even size,
\[
\boldsymbol{\lambda}\ \text{destructible} \iff \boldsymbol{\lambda}\setminus D\ \text{destructible}
\]
for any domino rim-hook $D$ [2402.05710]. These facts make domino removal stable under order and reduce the global question of complete dismantling to local removability.

The key obstruction is the triangular diagram
\[
\bigtriangledown_1^m=(m,m-1,\dots,1).
\]
A diagram is triangular if and only if it has no domino rim-hook, and therefore every triangular diagram is indestructible. More generally, a diagram is indestructible exactly when every domino-removal process eventually ends in a triangular diagram [2402.05710]. This yields a structural picture in which destructible diagrams reduce to the empty diagram, while indestructible diagrams terminate at a triangular core.

The representation-theoretic mechanism behind the classification is the Murnaghan–Nakayama rule, together with the auxiliary fact that a rim-hook of length $pk$ can be peeled off in $k$ successive rim-hook removals each of length $p$ [2402.05710]. This permits a reduction from character values on permutations with even cycle lengths to repeated domino-removal.

For destructible diagrams, the paper derives a precise formula for the value on cycle type $(2,2,\dots,2)$:
\[
\chi^{\boldsymbol{\lambda}}(\rho)=\pi(\boldsymbol{\lambda})\sum_{T\in\mathcal{T}} |T|,
\]
where $\mathcal{T}$ is the set of domino tilings of $\boldsymbol{\lambda}$ and
\[
\pi(\boldsymbol{\lambda})=\prod_{i=1}^{t-1}(-1)^{(t-i)\lambda_i}.
\]
A further lemma shows that if $v$ is the number of vertical dominoes in any tiling, then
\[
(-1)^v=\pi(\boldsymbol{\lambda}).
\]
Thus the sign contribution from every tiling is uniform, and the character value is a signed count of tilings [2402.05710]. From this, one obtains the dichotomy: indestructible diagrams force vanishing on all $P_{2n}$, while destructible diagrams do not vanish identically there.

## 5. Partition-algebra generalization: recombinants

One recent generalization replaces symmetric-group characters by partition-algebra characters and replaces permutations by the full diagram basis of the partition algebra [2309.16647]. For a square matrix $A=(a_{ij})_{n\times n}$ and an irreducible character $\chi^{\mathbf{p}}$ of the partition algebra $P_n(r)$, the associated function is the recombinant:
\[
\operatorname{Rec}^{\mathbf{p}}(A) = \sum_{d\in \mathcal{D}_n} \chi^{\mathbf{p}}(d)\,\prod_{a_{ij}\in d} a_{ij},
\]
where $\mathcal{D}_n$ is the full diagram basis. The product is defined diagrammatically by propagating blocks, and if a diagram has propagation number $0$, the product is declared to be $0$ [2309.16647].

This is a genuine enlargement of the irreducible character immanant paradigm. In the classical case, the indexing set is $S_n$, so every summand corresponds to a perfect matching between top and bottom indices. In the recombinant, the indexing set is the whole diagram basis of $P_n(r)$, including diagrams with propagating and non-propagating blocks [2309.16647]. The weight system is therefore no longer a class function on $S_n$.

The relation to irreducible character immanants is nevertheless exact in the “top” part of the partition-algebra Bratteli diagram. If the vacillating tableau data end at an ordinary partition of size $n$, then
\[
\operatorname{Rec}^{\mathbf{p}}(A)=\operatorname{Imm}^{\lambda}(A),
\]
where $\lambda$ is the corresponding symmetric-group partition [2309.16647]. The proof uses the fact that non-propagating diagrams contribute $0$ in this case, so the sum collapses to propagating diagrams, ანუ to permutation diagrams, and the partition-algebra character values become the symmetric-group character values.

The same paper explicitly distinguishes recombinants from $f$-immanants, Temperley–Lieb immanants, and Kazhdan–Lusztig families. The distinction is structural rather than terminological: the recombinant is not a permutation-indexed sum with an arbitrary weight function on $S_n$, but a partition-algebraic character sum over all partition diagrams [2309.16647].

## 6. Quasisymmetric refinements and quasi-immanants

A different line of development refines irreducible character immanants by replacing cycle types with cycle compositions and replacing symmetric functions with quasisymmetric functions [2501.15667]. For a permutation $\sigma$, the cycle composition $\operatorname{ccomp}(\sigma)$ is obtained by writing each cycle in increasing order, sorting the cycles lexicographically, and then taking the sequence of their lengths. This refines $\operatorname{ctype}(\sigma)$: two permutations with the same cycle type may have different cycle compositions [2501.15667].

Using the quasisymmetric power-sum bases $\{\Psi_\alpha\}$ and $\{\Phi_\alpha\}$, the paper defines quasi-immanants
\[
\mathrm{QImm}_{\Psi}^{Q}(A)
=
\sum_{\sigma\in S_n}
\Bigl(\text{coefficient in }n!Q\text{ of }
\begin{cases}
p_{\operatorname{ctype}(\sigma)} & \text{if }Q\in \mathrm{Sym},\\
\Psi_{\operatorname{ccomp}(\sigma)} & \text{otherwise}
\end{cases}
\Bigr)
\prod_{i=1}^n a_{i,\sigma(i)},
\]
and analogously $\mathrm{QImm}_{\Phi}^{Q}(A)$ [2501.15667]. When $Q$ is symmetric, these constructions recover the classical immanant. In particular,
\[
\mathrm{QImm}_{\Psi}^{s_\lambda}(A)=\mathrm{QImm}_{\Phi}^{s_\lambda}(A)=\mathrm{Imm}^\lambda(A),
\]
because the Schur function $s_\lambda$ has the usual power-sum expansion with coefficients $\chi^\lambda(\operatorname{ctype}(\sigma))$ [2501.15667].

The refinement becomes genuinely new for quasisymmetric inputs. The paper studies the quasisymmetric Schur function $\mathcal{S}_{(2,1^{n-2})}$ as an analogue of the classical second immanant and proves an explicit coefficient formula for
\[
\mathrm{QImm}_{\Psi}^{\mathcal{S}_{(2,1^{n-2})}}(A)
=
\sum_{\sigma\in S_n} c_\sigma \prod_{i=1}^n a_{i,\sigma(i)}.
\]
For $n\ge 3$, the coefficient $c_\sigma$ depends on the first part of $\operatorname{ccomp}(\sigma)$: it is nonzero only when that first part is $1$ or $2$, and in those cases is given by a sign times the number of $n$-permutations with the same cycle type as $\sigma$ [2501.15667]. This is a composition-level refinement of classical character-immanant combinatorics.

The $n=3$ example makes the distinction concrete. The quasi-immanant
\[
\mathrm{QImm}_{\Psi}^{\mathcal{S}_{(2,1)}}(A)
=
a_{11}a_{22}a_{33} -3a_{11}a_{23}a_{32} +3a_{12}a_{21}a_{33} +3a_{13}a_{22}a_{31},
\]
differs from the classical second immanant
\[
d_2(A)=2a_{11}a_{22}a_{33}-a_{12}a_{23}a_{31}-a_{13}a_{21}a_{32}
\]
[2501.15667]. This shows that quasi-immanants do not merely repackage irreducible character immanants; they refine them by retaining ordered cycle-decomposition data that ordinary cycle type forgets.

## 7. Quantum immanants and Capelli analogues

In the setting of $U(\mathfrak{gl}(n))$, the objects called quantum immanants are realized as central elements $\mathbf{S}_\lambda(n)$, the Schur elements, defined by
\[
\mathbf{S}_\lambda(n)=\frac{1}{H(\lambda)}\sum_{S}[S\mid S],
\]
where the sum runs over row strictly increasing Young tableaux $S$ of shape $\lambda$, $H(\lambda)$ is the hook-number, and $[S\mid S]$ is a double Young-Capelli bitableau [2107.10205]. The paper emphasizes that this presentation is character-free: it does not define the objects through irreducible characters of symmetric groups.

The same work proves that these Schur elements are the same as the Okounkov quantum immanants and that under the Harish-Chandra isomorphism
\[
X_n: Z(U(\mathfrak{gl}(n)))\longrightarrow \Lambda^*(n),
\]
one has
\[
X_n(\mathbf{S}_\lambda(n))=s_\lambda^*(x_1,\dots,x_n),
\]
the shifted Schur polynomial [2107.10205]. Their representation-theoretic characterization is triangular: if $\mu$ is a partition with $|\mu|<|\lambda|$, then $\mathbf{S}_\lambda(n)\cdot v_\mu=0$; if $\mu=\lambda$, then $\mathbf{S}_\lambda(n)\cdot v_\lambda=H(\lambda)\,v_\lambda$; and if $\mu\neq\lambda$ but $|\mu|=|\lambda|$, the action also vanishes [2107.10205].

The connection back to irreducible character immanants is explicit but secondary. Proposition 4.8 in the paper writes $\mathbf{S}_\lambda(n)$ as a linear combination of diagonal Capelli immanants:
\[
\mathbf{S}_\lambda(n)
=
(-1)^{|\lambda|}
\sum_{h_1+\cdots+h_n=h}
\frac{1}{h_1!\cdots h_n!}\,
\operatorname{Cimm}_\lambda[1^{h_1}\cdots n^{h_n};\,1^{h_1}\cdots n^{h_n}],
\]
so the character-theoretic Capelli immanants appear as an alternative expansion of the same central element [2107.10205]. In this sense, irreducible character immanants persist inside the quantum theory not as the primary definition, but as one coordinate system for tableau-defined central elements.

Special Schur elements recover two classical central families:
\[
\mathbf{H}_k(n)=\mathbf{S}_{(1^k)}(n),\qquad
\mathbf{I}_k(n)=\mathbf{S}_{(k)}(n),
\]
the Capelli elements and the Nazarov-Umeda elements, respectively [2107.10205]. Their eigenvalues are described by horizontal-strip and vertical-strip combinatorics, and the involution $W_n$ on the center satisfies
\[
W_n(\mathbf{H}_k(n))=\mathbf{I}_k(n),\qquad
W_n(\mathbf{S}_\lambda(n))=\mathbf{S}_{\lambda'}(n),
\]
so partition conjugation becomes an internal duality of the quantum-immanant framework [2107.10205]. This suggests that the classical irreducible character immanant sits inside a broader network in which Young-diagram combinatorics governs both matrix functions and central elements of enveloping algebras.

Source: https://www.emergentmind.com/topics/irreducible-character-immanants