Chromatic Symmetric Function

Updated 3 December 2025

Chromatic symmetric function is a symmetric invariant that encodes proper graph colorings into algebraic sums, capturing key combinatorial and geometric information.
It expands in the power-sum basis with combinatorial interpretations that assist in distinguishing non-isomorphic trees through derivative and enumeration techniques.
Extensions to weighted graphs, hypergraphs, and noncommutative analogues bridge combinatorics, representation theory, and algebraic geometry, driving ongoing research.

The chromatic symmetric function (CSF) is a symmetric function invariant of finite graphs, introduced by Stanley as a far-reaching extension and refinement of the chromatic polynomial. It encodes proper colorings of graphs as symmetric functions, carrying combinatorial, algebraic, and geometric information that has deep connections to graph isomorphism, symmetric function theory, algebraic geometry, and complexity theory. The CSF is definable for both ordinary graphs and certain classes of hypergraphs, with various extensions to weighted, directed, or structured graphs. This entry provides an advanced account of its definition, structural properties, distinguishing power, basis expansions, and key generalizations.

1. Definition and Power-Sum Expansion

Let $G = (V,E)$ be a finite simple graph with $n = |V|$ . A proper coloring is a map $K: V \to \mathbb{Z}^+$ such that $K(u)\ne K(v)$ whenever $uv \in E$ . The chromatic symmetric function is defined as

$X_G(x_1, x_2, \ldots) = \sum_{K:\ \text{proper}} \prod_{v\in V} x_{K(v)},$

a homogeneous symmetric function of degree $n$ (Wang et al., 2023).

Stanley's canonical expansion in the power-sum basis is

$X_G = \sum_{\lambda\vdash n} c_\lambda(G) p_\lambda,$

where for $\lambda = (\lambda_1,\ldots,\lambda_\ell)$ , $p_\lambda = \prod_{i=1}^\ell p_{\lambda_i}$ , and $p_k = \sum_i x_i^k$ .

For any $A\subseteq E(G)$ , let $\operatorname{type}(A)$ be the partition of $n$ given by the sizes of the connected components of $(V,A)$ . Then Stanley's inclusion-exclusion formula reads

$X_G = \sum_{A\subseteq E(G)} (-1)^{|A|} p_{\operatorname{type}(A)},$

so $c_\lambda(G) = \sum_{A\subseteq E(G):\ \operatorname{type}(A)=\lambda} (-1)^{|A|}$ .

If $G$ is a forest, there is no cancellation in this sum, and

$c_\lambda(G) = (-1)^{n-\ell(\lambda)}\, \big|\{A\subseteq E(G):\ \operatorname{type}(A)=\lambda\}\big|.$

This explicit combinatorial interpretation is central in distinguishing trees or extracting refined combinatorial invariants (Wang et al., 2023, Aliste-Prieto et al., 2024).

2. Tree Isomorphism and Distinguishing Capacity

A major impetus in CSF theory is Stanley's conjecture: the chromatic symmetric function distinguishes non-isomorphic trees. Significant progress has been made for specific families:

Spiders and 2-spiders: For trees with a single (spiders) or two vertices of degree $\geq 3$ (2-spiders), $X_T$ determines the isomorphism class of $T$ by recursively isolating legs/twigs from the power-sum data (Gerling, 2017, Huryn, 2019, Crew, 2021, Wang et al., 2023). For trees with exactly two high-degree vertices, applying $\partial/\partial p_k$ isolates trunk and twig data, allowing a reconstruction of the attachment structure (see Section 3 below).
Caterpillars: The subtree polynomial and half-generalized degree polynomial associated to caterpillars can coincide for large non-isomorphic families, so up to these invariants, the CSF may fail to distinguish all trees, but to date, no counterexample is known for $X_T$ itself (Aliste-Prieto et al., 2024).
Star connections and other special classes: Explicit formulas exist for the maximal independent sets in star connections and related forms, showing $X_T$ distinguishes these classes (Gerling, 2017).

Despite these results, the full injectivity of $T \mapsto X_T$ remains an open and central problem; it is known to be true for spiders, 2-spiders, and trees with exactly two high-degree vertices (Wang et al., 2023, Huryn, 2019, Crew, 2021), and there is strong evidence for much broader classes (Aliste-Prieto et al., 2024).

3. Differentiation Techniques and Subtree Enumeration

The CSF's power-sum expansion enables sophisticated "peeling" operations via formal differentiation. Consider

$\frac{\partial}{\partial p_k} X_G,$

interpreted as a formal derivative with respect to $p_k$ . Stanley's result shows

$\frac{\partial X_G}{\partial p_j} = \sum_H P_H X_{G\setminus V(H)},$

with the sum over all connected induced subgraphs $H$ of order $j$ , and $P_H$ the leading coefficient of the chromatic polynomial of $H$ .

When $F$ is a forest, every connected induced subgraph is a tree and $P_H = (-1)^{j-1}$ : $\frac{\partial X_F}{\partial p_j} = (-1)^{j-1} \sum_{H \subseteq F,\ |V(H)|=j,\ H\text{ tree}} X_{F\setminus V(H)}.$ Successive differentiation thus provides a way to extract enumerative data on subtrees, trunk sizes, and attachment patterns. In particular, these techniques form the core of the arguments distinguishing trees with two high-degree vertices and reconstructing combinatorial invariants associated to vertex-deletion or subtree attachment (Wang et al., 2023).

4. Generalized Degree Sequences, Subtree and Degree Polynomials

CSFs determine rich enumerative invariants such as the generalized degree polynomial (GDP) and related symmetric functions: $D_T(x,y,z) = \sum_{A \subseteq V(T)} x^{|A|} y^{d(A)} z^{e(A)} = \sum g_T(a,b,c) x^a y^b z^c,$ where $d(A)$ is the number of edges with exactly one endpoint in $A$ and $e(A)$ the number with both endpoints in $A$ .

Crew's Conjecture, now a theorem (Wang et al., 2023, Aliste-Prieto et al., 2024), asserts that $X_T$ determines the full generalized degree sequence, i.e., the multiset of all triples $(|A|, e(A), d(A))$ as $A$ ranges over all subsets, via

$g_T(a,b,c) = \sum_{\lambda\vdash n} c_\lambda(T)\, \omega(\lambda;a,b,c),$

where $\omega$ is a computable combinatorial kernel.

Further, restricting $A$ to subtrees relates $X_T$ to the subtree polynomial

$S_T(q,r) = \sum_{S \subseteq T\atop\text{subtree}} q^{e(S)} r^{\ell(S)},$

enumerating subtrees by numbers of edges and leaves. Linear relations exist between the GDP, restricted GDP, and $S_T$ . For caterpillars, equivalence classes can be large, but the full GDP refines these partitions, making it a strictly stronger invariant (Aliste-Prieto et al., 2024).

5. Positive Bases and Geometric Connections

The expansion of $X_G$ in classical symmetric-function bases exposes deep combinatorics and positivity phenomena:

Elementary basis: $X_{K_k} = k! e_k$ shows that only products of elementary functions (up to scaling) correspond to CSFs of disjoint unions of cliques (Cho et al., 2016).
Schur positivity and hook coefficients: For arbitrary graphs, the Schur hook coefficients $c_{(k,1^{n-k})}$ are always nonnegative and count acyclic orientations with prescribed numbers of sinks (Kaliszewski, 2014).
Geometric realization and $e$ -positivity: For certain graph classes (notably unit-interval or incomparability graphs of $(3+1)$ -free posets), $X_G$ is realized as the graded character of the cohomology of specific Hessenberg varieties or their compactifications. Strong conjectures (Stanley–Stembridge, Shareshian–Wachs) and geometric reformulations postulate $e$ -positivity of $X_G$ in these classes (Kato, 2024).
Noncommutative and extended analogues: $Y_G$ (CSF in noncommuting variables) has explicit combinatorial and algebraic control over positivity for these analogues (Dahlberg et al., 2019). MacMahon-type CSFs capture further invariants in weighted or multi-alphabet settings (Martin et al., 31 Jul 2025).

6. Structural, Algebraic, and Complexity Aspects

The chromatic symmetric function is at the confluence of combinatorics, representation theory, and geometry, and exhibits notable structural phenomena:

Weyl denominator and Lie theory: For graphs derived from Borcherds–Kac–Moody algebras, $X_G$ can be recast as a Lie-theoretic object, with explicit formulas in terms of root multiplicities and connections to colored Weyl denominators (Arunkumar, 2019).
Basis and computation: Symmetric function bases derived from complete multipartite or clique structures ( $r$ -basis, $e$ -basis) allow efficient expansion, and triangulate the space of degree- $n$ symmetric functions. Explicit combinatorial formulas exist for all basis changes between standard bases and chromatic bases associated to forests and multipartite graphs (Mobaraki et al., 2024, Crew et al., 2020).
Newton polytopes and Lorentzian property: For specific graph classes (e.g., indifference graphs of Dyck paths or incomparability graphs of $(3+1)$ -free posets), the Newton polytope of $X_G$ equals a permutahedron determined by greedy coloring, and the support satisfies the SNP property. In special cases, $X_G$ is an example of a Lorentzian polynomial, directly linked to discrete convexity and log-concavity (Matherne et al., 2022).
Categorification: There exists a homological categorification of $X_G$ via bigraded $S_n$ -modules, whose graded Euler characteristic recovers $X_G$ . These categorifications specialize to $X_G$ at $q=t=1$ , lift all classical combinatorial recurrences to long exact sequences, and are conjectured to sharpen the distinguishing power and positivity properties further (Sazdanovic et al., 2015).

7. Extensions, Generalizations, and Open Directions

Weighted and multi-alphabet analogues: The chromatic MacMahon symmetric function encodes vertex-weights and records finer enumerative statistics by lifting the theory to bialphabets, determining, for instance, the generating function for all vertex-subsets by multiple indexes (Martin et al., 31 Jul 2025).
Hypergraphs: For prime-edge hypertrees, $X_H$ is always $F$ -positive, providing a complete combinatorial prescription for all quasisymmetric coefficients (Taylor, 2015).
Constrained coloring and $H$ -chromatic symmetric functions: Introducing coloring models with respect to a constraint graph $H$ ( $H$ -chromatic symmetric functions) generates a unifying framework for the realization of numerous classical symmetric function bases, expansion formulas, and their isomorphism properties (Eagles et al., 2020).
Open problems: The tree distinguishability question remains unresolved in full generality, with stronger invariants (e.g., the GDP or augmented subtree polynomial) distinguishing all trees up to at least 18 vertices (Aliste-Prieto et al., 2024). The $e$ -positivity conjecture for unit-interval graphs (Stanley–Stembridge) is reduced to geometric representation-theoretic statements and remains open (Kato, 2024).

The chromatic symmetric function thus serves as a central object connecting combinatorial graph invariants, representation theory, symmetric function theory, and geometry, with ongoing research at the boundary of computer algebra, combinatorial categorification, and algebraic geometry.

References: