Cycle-Joining Method

• Cycle-Joining method is a combinatorial and algebraic technique that merges disjoint cycles using precise joining operations to form comprehensive cycles.
• It employs approaches like conjugate-pair splicing, concatenation trees, and greedy state graph traversals to efficiently generate universal cycles.
• Its applications span de Bruijn sequences, algebraic constructions, topological data analysis, and subgraph sampling under cyclic constraints.

The cycle-joining method comprises a family of combinatorial, algebraic, and topological constructions wherein multiple disjoint cycles—often arising from naturally partitioned structures—are joined into a single cycle or higher-dimensional analog, according to precisely defined joining operations. This paradigm is central to the generation of universal cycles (such as de Bruijn sequences), the construction of symmetric cycles in digraphs via Hajós-type moves, the characterization of subgraph sampling joins under cyclic constraints, and the algebraic-topological interpretation of joined cycles in persistent homology. Its implementations span discrete mathematics, computational algebra, graph theory, and topological data analysis, with rigorous algorithmic and structural foundations.

1. Formal Definitions and General Setting

Given a collection of disjoint cycles (or chains) on a set of objects—such as bitstrings, permutations, graphs, or simplices—the aim of cycle-joining is to merge them into a larger cycle that exhaustively and nonredundantly traverses the objects of interest. In the classical context, for an alphabet $\Sigma$ and length $n$, the universal cycle for a set $\mathbf S\subseteq\Sigma^n$ is a cyclic word $U$ in which every length-$n$ substring appears exactly once, a de Bruijn sequence being the case $\mathbf S=\Sigma^n$.

The underlying cycles may be orbits under shift registers (as in PCR: $f_{\rm PCR}(a_1 a_2 \cdots a_n)=a_1$, producing rotation classes), orbit classes under graph automorphisms, or cycles in the adjacency graphs of combinatorial or algebraic structures. The key operation is the merging (“splicing”) of two cyclic classes $C_1, C_2$ at a pair of “conjugate” or otherwise suitably matched representatives, producing a new cycle covering their union (Sawada et al., 2023, Chang et al., 2017, Amram et al., 2021).

2. The Classical Cycle-Joining Paradigm

Cycle-joining dates to the study of de Bruijn sequences, where cycles under an n-stage feedback shift register partition the state space into necklace classes. The standard join operation (“Cycle-Join Lemma,” (Sawada et al., 2023)) is performed whenever two cycles contain conjugate pairs $$(\alpha = x\beta \in C_1,\, \hat{\alpha} = y\beta \in C_2)$$ with $x\ne y$ and a shared suffix $n$0. Concatenating sequences ending in these representatives yields a universal cycle for their union.

Procedurally, this is captured by the construction of a PCR-based cycle-joining tree: an (unordered) rooted tree whose nodes are necklace-cycles and whose edges are labeled by the corresponding conjugate-pairs. Iteratively contracting edges recursively joins cycles until a single global cycle remains. This construction manifests in various forms—including greedy traversal, explicit enumeration of conjugate pairs (computed via Zech’s logarithms (Chang et al., 2017)), or graph-based joinings in FSR state graphs (Chang et al., 2020).

3. Extensions: Efficient Algorithms and Structural Generalization

Cycle-joining methods, while conceptually transparent, may be algorithmically suboptimal in their bare form, requiring $n$1 time per output symbol due to the need for repeated conjugacy testing and cycle manipulation. Recent frameworks have systematized the structure of these joins to enable more efficient traversal and generation:

• Concatenation Trees (Sawada et al., 2023): Express any PCR-based cycle-joining construction as a bifurcated ordered tree (BOT), where a carefully labeled in-order (right-current-left, RCL) traversal concatenates the aperiodic prefixes of necklace-words in a manner that reconstructs the desired universal cycle. This yields amortized $n$2 time per symbol and $n$3 or better space for a wide variety of combinatorial families (including de Bruijn sequences, weak orders, permutations, orientable sequences).
• Graph-Joining under Greedy Algorithms (Chang et al., 2020): The Generalized Prefer-Opposite (GPO) and its graph-joining extension (GJPO) systematically detect companion-pair “jump points” between disjoint FSR state graph components, embedding the classical cycle-join logic within a bit-by-bit greedy algorithm. These approaches exploit the structure of the state graph and its leaves/cycles to enforce the correct joining behavior.

4. Cycle-Joining in Algebraic and Topological Contexts

The cycle-joining concept generalizes beyond word cycles to algebraic and topological settings:

• Concurrence Topology (Ellis, 2015): In topological data analysis, the join operation merges persistent homology classes from two statistically independent groups of binary variables. The chain-level join constructs a nontrivial $n$4-cycle from representatives $n$5 (in $n$6 of group $n$7) and $n$8 (in $n$9 of group $\mathbf S\subseteq\Sigma^n$0), yielding a new persistent homology class in the join complex, which encodes higher-order independence (or its failure) in the data.
• Spectral Graph Theory (Doan et al., 2021): The join (in the sense of graph theory) of circulant (including cycle) graphs is analyzed via block matrix spectral decompositions, with internal and mixing eigenvalues derived from the joined cycles' adjacency matrices.

5. Advanced Applications: Directed and Symmetric Cycles via Hajós Constructions

A specialized domain of cycle-joining arises in the iterative construction of odd symmetric cycles (bidirected cycles $\mathbf S\subseteq\Sigma^n$1) in digraphs using Hajós-type operations, as formalized by Bang-Jensen et al. (García-Altamirano et al., 2023, García-Altamirano et al., 2022). The essential atomic moves are:

• Directed Hajós Join: Given two digraphs and marked arcs, their directed Hajós join removes the arcs, identifies head vertices, and inserts a new cross-arc.
• Identification of Vertices: Merging two independent vertices to combine neighborhoods.

By recursively splicing copies of smaller symmetric cycles (starting with the triangle $\mathbf S\subseteq\Sigma^n$2) and collapsing redundant vertices, one builds up $\mathbf S\subseteq\Sigma^n$3 for any $\mathbf S\subseteq\Sigma^n$4, with the cycle-joining steps matching the necessary structural properties for $\mathbf S\subseteq\Sigma^n$5-critical digraphs. Automated heuristic search (adapted rank-based genetic algorithms) has been employed to optimize the sequence of joins/identifications, notably giving a 16-step atomic construction for $\mathbf S\subseteq\Sigma^n$6 from $\mathbf S\subseteq\Sigma^n$7 (García-Altamirano et al., 2022).

6. Generalizations: Joins, Independence, and Sampling in Broader Structures

• Join Sampling with Cyclic Constraints: In database and subgraph sampling theory, “cycle-joining” refers to joins under cyclic degree constraints, as in subgraph isomorphism queries for cyclic patterns. Techniques for efficiently reducing these to acyclic constraints (without inflating key bounds) enable uniform sampling and query evaluation (Wang et al., 2023).
• Cycle-Joining and Spanning Tree Enumeration: In algebraic and combinatorial frameworks, the choice of joining sequence (spanning tree in the cycle adjacency graph) encodes the structure of resulting objects (e.g., de Bruijn sequences), with enumeration formulas (via Laplacian determinants) reflecting the number of inequivalent outcomes (Chang et al., 2017, Sawada et al., 2023).

7. Representative Examples and Connections

The table below summarizes principal applications of the cycle-joining method:

Area	Cycle Objects	Join Operation
de Bruijn sequences	PCR necklace-cycles	Conjugate-pair splicing (Sawada et al., 2023)
Greedy FSR-based construction	FSR state cycles	Companion-pair jump (Chang et al., 2020)
Algebraic field constructions	LFSR cycles, Zech logs	Cycle adjacency via exponent arithmetic (Chang et al., 2017)
Symmetric cycles in digraphs	Bidirected cycles	Directed Hajós join, vertex identification (García-Altamirano et al., 2023, García-Altamirano et al., 2022)
Concurrence topology	Persistent homology cycles	Simplicial join at chain level (Ellis, 2015)
Database/subgraph sampling	Join result cycles	Acyclic reduction, polymatroid bounds (Wang et al., 2023)

These applications demonstrate the unifying role of the cycle-joining paradigm across combinatorics, algebra, discrete algorithms, and applied topology.

---

References:

• Concatenation trees and universal cycle constructions (Sawada et al., 2023)
• Graph-joining greedy approaches (Chang et al., 2020)
• Zech’s logarithms and de Bruijn cycles (Chang et al., 2017)
• Prefer-max de Bruijn sequence cycle-joining (Amram et al., 2021)
• Hajós construction of symmetric cycles (García-Altamirano et al., 2023, García-Altamirano et al., 2022)
• Subgraph sampling with cyclic constraints (Wang et al., 2023)
• Spectral theory of cyclic graph joins (Doan et al., 2021)
• Concurrence topology and joined cycles (Ellis, 2015)