---
title: Proper Caterpillars in Graph Theory
url: https://www.emergentmind.com/topics/proper-caterpillars
type: topic
---

# Proper Caterpillars in Graph Theory

A proper caterpillar is a combinatorial structure within graph theory, specifically a tree whose non-leaf vertices induce a simple path—called the spine—and in which every vertex of the spine is adjacent to at least one leaf, so that no spine vertex has degree 2 in the full tree. Recent work by Aliste-Prieto and Zamora has established that the symmetric chromatic function is a complete invariant for proper caterpillars: two such trees are isomorphic if and only if they have the same symmetric chromatic function [1208.2267].

## 1. Formal Definition and Structural Properties

A tree $T$ is a *caterpillar* if the induced subgraph on all non-leaf vertices forms a simple path, called the *spine* of $T$. If $P(T)\subset E(T)$ is the subset of internal (i.e., non-leaf) edges, $T$ is a caterpillar if and only if $P(T)$ forms a path. A caterpillar is *proper* if each spine vertex has degree at least 3 in $T$, i.e., each is incident to at least one leaf. Formally, a caterpillar $T$ with $k$ spine vertices $v_1,\dots,v_k$ is proper if and only if every $v_i$ is adjacent to at least one leaf.

This properness criterion distinguishes proper caterpillars from more general caterpillars by the absence of spine vertices of degree 2, enforcing that in any proper caterpillar, the spine is "supported" by a minimum number of leaves at each spine vertex.

## 2. The Symmetric Chromatic Function

The symmetric chromatic function $X_G(x)$, introduced by Stanley, generalizes the chromatic polynomial. For a (finite) graph $G$ on vertex set $V(G)$, define
\[
X_G(x) = \sum_{\substack{\kappa: V(G) \to \mathbb{P} \\ \kappa(u)\neq \kappa(v)~\forall \{u,v\} \in E(G)}} \prod_{v\in V(G)} x_{\kappa(v)}
\]
where $\mathbb{P} = \{1,2,3,\dots\}$ and $x=(x_1,x_2,\dots)$ are commuting indeterminates.

Alternatively, $X_G(x)$ can be expressed in the power–sum basis indexed by integer partitions $\lambda$:
\[
X_G(x) = \sum_{A\subseteq E(G)}(-1)^{|A|} p_{\lambda(A)}
\]
where $\lambda(A)$ is the partition of $|V(G)|$ giving the sizes of components of the spanning subgraph $(V(G),A)$. For trees $T$, $X_T$ contains the same information as the Noble–Welsh $U$-polynomial, defined
\[
U_T(x,y) = \sum_{A\subseteq E(T)} x_{\lambda(A)}(y-1)^{|A| - (|V|-k(A))}
\]
where $k(A)$ is the number of connected components of $(V,A)$. Specializing $y = 1$ yields
\[
U_T(x) := U_T(x; 1) = \sum_{A\subseteq E(T)} x_{\lambda(A)}
\]
and $U_T$ can be recovered from $X_T$ and vice versa.

## 3. Bijection Between Proper Caterpillars and Integer Compositions

A key observation is the correspondence between proper caterpillars and integer compositions up to reversal (reverse classes). Let $\mathcal{T}^+$ denote the set of proper caterpillars and $\mathcal{C}$ the set of all compositions (ordered sequences of positive integers). Two compositions $\alpha, \beta \in \mathcal{C}$ lie in the same *reverse class* if $\beta$ is either $\alpha$ or its reverse $\alpha^*$. The reverse-class of $\alpha$ is denoted $[\alpha]^*$.

Given a proper caterpillar $T$ with ordered spine $(v_1,\dots,v_k)$ and $B_i$ leaves adjacent to $v_i$ (with $B_i \geq 2$ for all $i$), define
\[
\Phi: T \longmapsto [B_1, B_2, \dots, B_k]^*
\]
This map is a bijection from $\mathcal{T}^+$ to reverse-classes of compositions with each entry $\geq 2$.

Conversely, any reverse-class $[\alpha_1, \dots, \alpha_k]^*$ with all $\alpha_i \geq 2$ uniquely determines a proper caterpillar constructed by forming a path on $k$ vertices and appending $\alpha_i - 1$ leaves to the $i$th vertex. Aliste-Prieto & Zamora establish the bijection in Lemma 2.2 [1208.2267].

Furthermore, the restricted $U$-polynomial of such a caterpillar, taken over subsets of edges including all leaf-edges, coincides with the composition-lattice polynomial $L(\alpha, x)$ for the associated composition $\alpha$, yielding
\[
U_T(x) \leftrightarrow L(\Phi(T), x)
\]

## 4. Main Theorem and Distinguishability

The central result of Aliste-Prieto & Zamora is the following: for proper caterpillars $T, T'$, if $X_T(x) = X_{T'}(x)$, then $T \cong T'$. That is, the symmetric chromatic function (equivalently, the $U$-polynomial) uniquely determines the isomorphism type of any proper caterpillar [1208.2267, Thm 1.1].

This establishes the symmetric chromatic function as a complete isomorphism invariant for proper caterpillars, resolving Stanley's conjecture within this class.

## 5. Proof Approach: Composition Polynomials and Factorization

The proof proceeds as follows:

- **Reduction to Compositions**: By the established bijection, the $U$-polynomial of a proper caterpillar coincides with the composition-lattice polynomial $L(\alpha, x)$ of its associated composition $\alpha$. Thus, $U_T = U_{T'}$ implies $L(\alpha, \cdot) = L(\beta, \cdot)$ for the corresponding compositions.

- **Classification of Compositions**: Billera–Thomas–van Willigenburg show that two compositions $\alpha,\beta$ have the same $L$-polynomial if and only if they are in the same symmetry class generated by independently reversing the irreducible factors in their composition-monoid factorization: every $\alpha$ admits a unique irreducible factorization $\gamma_1 \circ \cdots \circ \gamma_r$, and all words in the symmetry class are formed by independently reversing any of the $\gamma_i$.

- **Palindromic Case**: If the composition $\alpha$ is a palindrome, its $L$-polynomial uniquely determines it. Therefore, the corresponding caterpillar is $U$-unique (and thus $X_T$-unique).

- **General Case**: If $\alpha$ is not a palindrome, the proof identifies a monomial in $U_T$ whose coefficient differs from that in $L(\alpha^*,x)$, using combinatorial arguments on leaf counts and subset counting (see Lemma 3.4 and Theorem 3.3 of [1208.2267]). As a result, no two distinct reverse-classes yield the same $U$-polynomial, confirming the main theorem.

## 6. Illustrative Examples

- *Palindromic Sequence*: For $\alpha = (3,2,3)$, any proper caterpillar with this leaf-sequence is uniquely identified by $X_T$.
- *Non-Palindromic Sequence and Isomorphism*: The sequences $(2,3,4)$ and its reverse $(4,3,2)$ define non-isomorphic caterpillars, but these are isomorphic after reversing the spine, which is consistent with the identification by reverse-class.
- *Non-Proper Sequences*: For $(1,2,3)$, where at least one spine vertex has a single adjacent vertex, the properness condition fails, and the result does not apply—there exist non-proper caterpillars with the same $X_T$ function.

## 7. Implications for Invariants and Stanley’s Conjecture

This work verifies Stanley’s conjecture regarding the expressive power of the symmetric chromatic function for the expansive class of proper caterpillars. The $U$-polynomial (and hence $X_G$) is shown to encode sufficient structure to recover the full isomorphism class in this setting [1208.2267].

The combinatorial translation between caterpillars and integer compositions allows the application of deep enumerative results on compositions, such as symmetry class analysis and monoid factorization. This machinery offers a pathway to extend distinguishing power arguments beyond caterpillars to broader families of trees.

More generally, the analysis exemplifies how algebraic invariants of graphs—symmetric functions, $U$-polynomials—can be interpreted and leveraged through classical combinatorial structures (integer compositions, Boolean lattices) equipped with robust factorization theory, providing interpretable and computationally tractable invariants for graph isomorphism problems.

Source: https://www.emergentmind.com/topics/proper-caterpillars