Isomeric Tree Factorization

• Isomeric tree factorization is the unique, commutative decomposition of rooted trees into atomic stumps, forming equivalence classes that simplify order conditions.
• It underpins efficient generating functions and enumeration schemes in combinatorics, providing a tractable approach to count non-isomorphic trees.
• The method is crucial in optimizing Runge–Kutta order conditions and rational number factorizations by reducing the scalar order requirements compared to vector conditions.

Isomeric tree factorization is a structural and combinatorial phenomenon arising in the study of rooted trees, with foundational applications in the order theory of Runge–Kutta methods, rational number factorization, and enumerative combinatorics. At its core, it refers to a unique, commutative decomposition of trees into irreducible building blocks ("atomic stumps"), inducing equivalence classes of trees—called isomeric classes—that play a critical role in simplifying order conditions, generating functions, and enumeration schemes in both applied and pure mathematical contexts (Butcher et al., 2021, Serena et al., 4 Mar 2025, Samuels et al., 2014).

1. Foundations: Rooted Trees and Isomeric Equivalence

Let $T = \bigcup_{p \geq 1} T_p$ denote the set of all non-planar rooted trees, where a tree $t\in T_p$ has $p$ vertices. Trees are constructed recursively in prefix notation: $t = \tau_m \, t_1 t_2 \ldots t_m$, encoding a root with $m$ child subtrees. Each tree corresponds to an elementary differential in $B$-series expansions for numerical integration, with significant combinatorial structure (Butcher et al., 2021).

An equivalence relation, called "isomeric," is defined by the capacity to permute atomic building blocks within the tree. Two trees $t_1$ and $t_2$ are isomeric ($t_1 \sim t_2$) if and only if their atomic stump factorizations are identical as multisets. The equivalence classes under $\sim$ are the isomeric classes. This structure underpins the reduction of scalar-order conditions in Runge–Kutta theory and appears in the enumeration of non-isomorphic trees in combinatorics (Butcher et al., 2021, Serena et al., 4 Mar 2025).

2. Atomic Stumps: Factorization Theorem and Operations

A "stump" is a tree in which some leaves are replaced by valencies (visible empty slots for grafting additional stumps). The atomic stumps, denoted $t\in T_p$0, are the minimal nontrivial stumps, consisting of a root with $t\in T_p$1 filled children and $t\in T_p$2 valencies:

$t\in T_p$3

If $t\in T_p$4 and $t\in T_p$5 are stumps, their product $t\in T_p$6 is formed by grafting $t\in T_p$7 into any valency of $t\in T_p$8. This product is commutative, reflecting the independence of attachment order.

The factorization theorem asserts: every rooted tree $t\in T_p$9 admits a unique factorization (up to permutation of factors) as a commutative product of atomic stumps:

$p$0

The multiset $p$1 is the atomic stump factorization of $p$2. Two trees are isomeric if their multisets of $p$3-stumps coincide (Butcher et al., 2021).

3. Enumeration of Isomeric Classes

The enumeration of isomeric tree classes diverges sharply from the enumeration of all rooted trees as order increases. Denote:

• $p$4: count of rooted trees of order $p$5,
• $p$6: count of isomeric classes (i.e., distinct multisets of atomic stumps) in $p$7,
• $p$8: cumulative vector-order conditions,
• $p$9: cumulative scalar-order conditions.

$t = \tau_m \, t_1 t_2 \ldots t_m$0	$t = \tau_m \, t_1 t_2 \ldots t_m$1	$t = \tau_m \, t_1 t_2 \ldots t_m$2	$t = \tau_m \, t_1 t_2 \ldots t_m$3	$t = \tau_m \, t_1 t_2 \ldots t_m$4
1	1	1	1	1
2	1	2	1	2
3	2	4	2	4
4	4	8	4	8
5	9	17	8	16
6	20	37	15	31
...	...	...	...	...
20	12,826,228	20,247,374	23,824	63,338

For $t = \tau_m \, t_1 t_2 \ldots t_m$5, $t = \tau_m \, t_1 t_2 \ldots t_m$6, so scalar order conditions (one per isomeric class) are a proper subset of the vector order conditions. No explicit generating function for $t = \tau_m \, t_1 t_2 \ldots t_m$7 is provided—the values are computed by generating trees, factorizing each into stumps, and collating identical multisets (Butcher et al., 2021).

4. Applications in Runge–Kutta Order Theory

In Runge–Kutta methods, order conditions up to order $t = \tau_m \, t_1 t_2 \ldots t_m$8 for vector problems require $t = \tau_m \, t_1 t_2 \ldots t_m$9 for all $m$0 with $m$1 (one per tree). For scalar (non-autonomous) problems, each isomeric class $m$2 generates the same elementary differential, and only one condition per class is required:

$m$3

Here, $m$4 is the symmetry factor. The reduction from $m$5 to $m$6 conditions allows for the construction of methods with "ambiguous order": scalar methods of order $m$7 that fail some vector conditions, thus exhibiting only vector-order $m$8 (Butcher et al., 2021).

A specialized routine, "test$m$9", systematically increases the degree $B$0 (internal node count) to build up a solvable subset of the scalar-order equations before vector-order conditions must be addressed. Explicit construction yields, for example, 6-stage methods satisfying all scalar order-5 conditions but failing exactly two vector conditions, and similar phenomena for order 6 (Butcher et al., 2021).

5. Enumerative and Generating Function Connections

Isomeric (flip-equivalence) tree classification is central to the enumeration of binary trees under various symmetries. The Wedderburn–Etherington numbers count non-isomorphic binary rooted trees under flip-isomorphism (children unordered). The generating function for their count $B$1 satisfies:

$B$2

Marking pairs of non-isomorphic siblings (parameter $B$3) or color classes (parameter $B$4), the bivariate and trivariate GFs factor into "product" and "dilated argument" terms, always reflecting a split into independent products or paired duplicates:

$B$5

$B$6

This factorization directly parallels the isomeric decomposition: tree symmetries yield multiplicative splits in the GF recursion, reflecting the underlying atomic stump structure. Such generating functions facilitate efficient coefficient extraction for large-scale enumerative tasks (Serena et al., 4 Mar 2025).

6. Factorization Trees for Rational Numbers

A distinct but related notion of tree factorization is employed in the optimal decomposition of rational numbers $B$7 into products of elements minimizing Mahler measure. Here, a factorization tree encodes all possible sequences of primitive factorizations, ordered by increasing partial product complexity. The canonical optimal tree $B$8 is constructed inductively, pruning away branches that cannot yield global optima. The binary tree structure is especially pronounced when $B$9 is square-free—resulting in at most $t_1$0 leaves for $t_1$1 distinct prime factors (Samuels et al., 2014).

While these rational factorization trees differ in purpose from atomic stump factorization, both derive their tractability from the underlying commutative and recursive structure of tree decompositions.

7. Examples and Illustrations

Low-Order Isomeric Classes in Runge–Kutta

Order 4 (no isomeric overlap)

• All four trees yield unique atomic stump multisets; thus, no isomeric pairs arise.

Order 5 (single isomeric class)

• Nine trees split into eight unique and one isomeric pair (e.g., $t_1$2 and $t_1$3 both with factorization $t_1$4).
• Only one scalar condition arises for the class $t_1$5, in contrast to two vector conditions.

Order 6 (multiple isomeric classes)

• Four of the 15 isomeric classes contain pairs or triplets, reflecting the divergence between scalar and vector order requirements for tree-based order conditions.

This pattern extends with increasing tree order, amplifying the effectiveness of isomeric classification in reducing condition counts.

8. Broader Implications

Isomeric tree factorization provides a unifying algebraic and combinatorial framework across disparate domains. In numerical analysis, it underlies the fundamental distinction between scalar and vector order within Runge–Kutta schemes and enables the systematic construction of ambiguous-order methods. In enumerative combinatorics, it explains the multiplicative separation of generating functions for non-isomorphic tree counting with various structural statistics. In rational arithmetic, factorization trees leverage atomic substructures for optimal metric representations.

No generating function or closed formula for the enumeration of isomeric classes is currently available, and open questions remain regarding extensions to algebraic numbers and more general combinatorial structures (Butcher et al., 2021, Samuels et al., 2014, Serena et al., 4 Mar 2025).