Overlap-Free Permutation Decomposition

• Overlap-free permutation decomposition is a framework that partitions permutations into non-overlapping segments based on specified pattern constraints.
• It involves constructing 0-overlap cycles, non-self-overlapping block decompositions, and inversion set splits using graph-theoretic and algebraic methods.
• Recent studies reveal enumerative and asymptotic properties that underscore its significance in combinatorial analysis and algorithm design.

Overlap-free permutation decomposition encompasses a spectrum of notions where permutations, or their combinatorially significant features (such as inversion sets or pattern structure), admit decomposition without prescribed overlaps or interactions of certain types. In the contemporary literature, this area covers (1) enumerative and algorithmic results for cycles and orderings where consecutive permutations have zero required overlap, (2) structural decompositions into blocks with no self-overlap patterns, and (3) graph- and algebraic-based decomposition of permutation inversion sets into independent non-overlapping components.

1. Definitions and Notions of Overlap-Freeness

An overlap-free permutation decomposition can refer to multiple, context-dependent structures, connected by the property that constituent pieces have no overlap in the sense of prescribed patterns, shared inversions, or contiguous repetitions.

s-Overlap Cycles

An s-overlap cycle (or s-ocycle) is an ordering of a set of combinatorial objects, each represented as a string of length $n$, such that a string $a = a_0 a_1 \ldots a_{n-1}$ is immediately followed by $b = b_0 b_1 \ldots b_{n-1}$ if and only if $$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$; that is, the last $s$ symbols of $a$ coincide with the first $s$ symbols of $b$ (Horan, 2013).

The case $s=0$ yields overlap-free cycles, i.e., orderings where no overlap constraint is required—any cyclic arrangement of all $n!$ permutations suffices.

Self-Overlapping and Non-Self-Overlapping Permutations

A permutation $a = a_0 a_1 \ldots a_{n-1}$0 is self-overlapping if there exists $a = a_0 a_1 \ldots a_{n-1}$1 such that the first $a = a_0 a_1 \ldots a_{n-1}$2 entries and last $a = a_0 a_1 \ldots a_{n-1}$3 entries induce order-isomorphic patterns on two respective invariant intervals. Otherwise, $a = a_0 a_1 \ldots a_{n-1}$4 is non-self-overlapping. Every permutation admits a unique decomposition into non-self-overlapping summands via a sequence of direct sums (“⊕”) of non-self-overlapping blocks and a central possibly empty block (Kirgizov et al., 2023).

Inversion Set Decomposition (inv-decomposition)

Given $a = a_0 a_1 \ldots a_{n-1}$5, its inversion set $a = a_0 a_1 \ldots a_{n-1}$6 can sometimes be partitioned into two disjoint subsets, each of which forms the inversion set of a permutation: $a = a_0 a_1 \ldots a_{n-1}$7. We call $a = a_0 a_1 \ldots a_{n-1}$8 an inv-decomposition, or overlap-free inversion set decomposition. Not every permutation admits such a decomposition (Katthän, 2011).

2. Existence Conditions and Constructions

Existence of Overlap-Free Cycles

For permutations, the transition digraph construction (de Bruijn–type) underlies the overlap-free (s=0) case. The digraph $a = a_0 a_1 \ldots a_{n-1}$9 has one vertex (the empty string) and $b = b_0 b_1 \ldots b_{n-1}$0 loop edges, one for each permutation; it is trivially balanced and connected, so an Euler tour exists. Any cycle ordering of the $b = b_0 b_1 \ldots b_{n-1}$1 permutations is a 0-overlap cycle, and such cycles always exist for every $b = b_0 b_1 \ldots b_{n-1}$2 (Horan, 2013).

Non-Self-Overlapping Block Decomposition

Every permutation has a unique decomposition into non-self-overlapping blocks: $b = b_0 b_1 \ldots b_{n-1}$3 where each $b = b_0 b_1 \ldots b_{n-1}$4 is non-self-overlapping, and $b = b_0 b_1 \ldots b_{n-1}$5 is arbitrary (possibly empty). This is achieved by repeatedly locating the minimal overlapping range and splitting off the corresponding non-self-overlapping block, then recursing on the remainder (Kirgizov et al., 2023).

Inversion Set Decomposition: Modular and Block Structure

A permutation’s inversion set $b = b_0 b_1 \ldots b_{n-1}$6 admits an inv-decomposition $b = b_0 b_1 \ldots b_{n-1}$7 if, in the modular decomposition of its inversion graph, every edge class associated to strong modules (parallel/serial/prime types) lies entirely in one subset or the other, with explicit constraints on serial modules matching permutation-induced bipartite graph structures. This criterion ensures each $b = b_0 b_1 \ldots b_{n-1}$8 is indeed an inversion set itself (Katthän, 2011).

Table: Types of Overlap-Free Decomposition

Decomposition Type	Structural Criterion	Always Exists?
s=0 overlap cycle	Cyclic listing of all perms, no overlap required	Yes, for all $b = b_0 b_1 \ldots b_{n-1}$9 (Horan, 2013)
Non-self-overlapping	No matching order isomorph on prefix/suffix intervals	Yes, block decomposition always exists (Kirgizov et al., 2023)
Inv-decomposition	Inversion set splits per modular decomposition constraints	Only when modular decomposition allows (Katthän, 2011)

3. Enumerative and Asymptotic Properties

For non-self-overlapping permutations, generated recursively by block decompositions, the exponential generating function $$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$0 satisfies

$$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$1

Asymptotically, the probability that a random $$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$2 is non-self-overlapping is

$$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$3

where $$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$4 counts non-self-overlapping permutations of size $$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$5 and $$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$6 is the falling factorial. Almost all large $$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$7 permutations are non-self-overlapping (Kirgizov et al., 2023).

For inv-decompositions, the number of distinct decompositions depends only on the modular decomposition tree structure and is given by

$$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$8

subtracting one for the trivial decomposition (Katthän, 2011).

4. Algorithmic and Structural Methods

Construction and Recognition

For overlap-free cycles (s=0), construction is trivial: any cyclic permutation order suffices. For non-self-overlapping block decomposition, recursively extract minimal overlap blocks until irreducibility is reached. For inversion set decompositions:

Algorithm Outline (Katthän, 2011):

1. Construct the inversion graph $$a_{n-s} a_{n-s+1} \ldots a_{n-1} = b_0 b_1 \ldots b_{s-1}$$9 (time $s$0).
2. Compute the modular decomposition tree (linear time).
3. Traverse the module tree, assigning edge classes to one factor or the other per node type (prime, serial, parallel).
4. Reconstruct $s$1 by reversing substitution, respecting the module assignments.

This algorithm ensures all decompositions consistent with the necessary modular/serial/prime structure are found efficiently.

Multiplicative Decompositions

A decomposition $s$2 is multiplicative if $s$3 in the group sense. Every inv-decomposable permutation admits at least one multiplicative inv-decomposition (Katthän, 2011).

5. Connections to Other Permutation Structures

Non-self-overlapping block decomposition parallels those for indecomposable and simple permutations. Generating functions and asymptotics exhibit formal similarities:

• Indecomposable permutations correspond to the block structure with no initial invariant interval; corresponding expansions feature combinatorial coefficients as in the non-self-overlapping case.
• Simple permutations, in contrast, have a limiting $s$4 asymptotic density and non-integral coefficients, and thus only partial analogy with overlap-free decompositions (Kirgizov et al., 2023).

These parallels suggest a deep combinatorial connection via “wreath-product” and “block-decomposition” theories, though non-self-overlapping permutations do not fit into the wreath-product framework as simply as the other classes.

6. Worked Examples

0-Overlap Cycles (Overlap-Free Cycles)

• For $s$5, permutations are $s$6, $s$7; any cyclic order such as $s$8 is a 0-overlap cycle.
• For $s$9, permutations are $a$0, and e.g., $a$1 is a 0-overlap cycle (Horan, 2013).

Inversion Set Decomposition

Given $a$2:

1. Compute inversions: $a$3.
2. $a$4 has two connected components; $a$5 is serial with children $a$6, $a$7, each prime.
3. There are $a$8 inv-decompositions.
4. Choose assignments per the procedure to get, for instance:
   • $a$9, $s$0 with $s$1 and disjoint inversion sets (Katthän, 2011).

7. Asymptotic and Probabilistic Properties

Almost all permutations are non-self-overlapping; the leading asymptotic term for the probability is 1 minus a sum over integer-coefficient falling-factorial corrections, with self-reference through the use of $s$2 coefficients counting smaller non-self-overlapping permutations. This property sharply distinguishes the non-self-overlapping class from the class of simple permutations, whose limiting density is strictly less than 1 (Kirgizov et al., 2023).

For very-tight occurrences of non-self-overlapping patterns, precise expansions for the probability and distribution are expressible in the falling-factorial basis, with explicit coefficients derived via enumerative recurrence (Kirgizov et al., 2023).

---

The study of overlap-free permutation decompositions integrates cyclic, block-structural, and graph-theoretic approaches, providing connections to enumerative combinatorics, modular decomposition theory, and asymptotic analysis. The field remains an active area for investigating deeper algebraic and probabilistic properties of permutation classes.