---
title: 'UniqueOverlap: Rigidity and Consistency'
url: https://www.emergentmind.com/topics/uniqueoverlap
type: topic
---

# UniqueOverlap: Rigidity and Consistency

In the cited literature, **UniqueOverlap** is not a single standardized term but a family of overlap-centered notions that express rigidity, canonicity, or extremal behavior. In substitution tilings it denotes the canonicity of overlap coincidence across admissible control-point choices; in overlapping substitutions it denotes globally consistent realization of allowed overlaps; in communication complexity it is the name of a three-party promise problem; in planted optimization it refers to a near-optimal overlap geometry with a forbidden interval; and in several combinatorial and geometric settings it marks sharp thresholds or minimal-overlap counterexamples [1509.04471] [2407.18666] [2507.11257] [2605.10911].

## 1. UniqueOverlap in substitution tilings

For primitive Pisot substitutions and, more generally, for non-rod self-affine tilings with connected tiles, finite local complexity, primitive substitution, and Meyer return set, the central objects are **strong coincidence** and **overlap coincidence**. Strong coincidence is formulated on the substitution Delone multi-color set arising from a choice of admissible control points, whereas overlap coincidence is formulated on overlaps \((U,y,V)\) in the tiling itself and requires that every overlap eventually inflates to a coincidence under substitution. The main equivalence theorem states that there exists a level \(n=n(T)\ge 1\) such that overlap coincidence is equivalent to **multiple strong coincidence of level \(n\)**: for every admissible control-point choice for \(\Omega^n\) with \(\Lambda-\Lambda\subset G\), the resulting multi-color set admits strong coincidence [1509.04471].

This result makes the “unique” aspect precise. Changing control points by color-dependent shifts changes the digit sets \(D_{ij}\), and strong coincidence can be created or destroyed by that change. The theorem shows that, under the stated Pisot and topological hypotheses, overlap coincidence is nonetheless intrinsic: the tiling has overlap coincidence if and only if strong coincidence holds for all admissible control-point choices in the class constrained by \(\Lambda-\Lambda\subset G\) at a fixed level \(n(T)\). In this sense, overlap coincidence becomes canonical rather than an artifact of a particular normalization of tiles or control points [1509.04471].

The same framework yields spectral consequences. Overlap coincidence is equivalent to pure discreteness of the translation action on the tiling space, and it implies pure point dynamical spectrum, pure point diffraction, and the inter-model-set characterization of the associated Delone multi-color set. The Taylor–Socolar half-hexagonal substitution tiling gives a concrete computed example: the Akiyama–Lee overlap-graph algorithm verifies overlap coincidence for the \(168\)-prototile substitution, and therefore the tiling dynamics has pure point spectrum and quasicrystalline structure [1212.4209].

## 2. UniqueOverlap as consistency of overlapping substitutions

In the theory of **overlapping substitutions**, a pre-overlapping substitution is a triple \((B,\phi,\rho)\) in which inflated proto-tiles are replaced by patches that may overlap along boundaries. Here the essential question is not coincidence but **consistency**: whether iterated substitutions ever create a contradictory overlap, meaning distinct tiles with intersecting interiors inside some \(\rho^n(T)\). In this setting, “UniqueOverlap” is the property that the allowed overlaps are realized in a unique, globally consistent way at every substitution level [2407.18666].

The formalism differs from classical substitution tilings because the associated substitution matrix may have non-integer entries. These entries record fractional coverage, and the right Perron–Frobenius eigenvector gives tile frequencies in the fixed-point tiling. The Perron–Frobenius eigenvalue equals \(\det \phi\), and under finite local complexity and repetitivity the expansion constant is an algebraic integer. Thus overlap is incorporated directly into the substitution matrix rather than treated as a defect to be eliminated [2407.18666].

The uniqueness criteria are dimension-dependent. In one dimension, stabilized adjacency graphs together with matching weighted ends imply that overlaps are uniquely determined and no contradiction appears. In two dimensions, a sufficient condition is given by a graph-directed iterated function system for tile boundaries satisfying the open set condition and the linear GIFS condition; then the boundary pieces are Jordan curves without self-intersection, and the substitution is globally consistent. The resulting “unique overlap” is therefore not spectral canonicity, as in coincidence theory, but uniqueness of admissible geometric gluing [2407.18666].

## 3. UniqueOverlap as a communication-complexity problem

In distributed sketching and communication complexity, **UniqueOverlap** is a formally defined three-party simultaneous-message problem. Alice receives \(X\in\{0,1,\bot\}^m\), Bob receives \(Y\in\{0,1,\bot\}^m\), both supports have size \(s\), and there is a unique index \(\sigma\) at which both entries are defined and satisfy \(X_\sigma\oplus Y_\sigma=1\). The referee Charlie knows the supports and therefore knows the unique overlap index \(\sigma\); the task is to decide whether \((X_\sigma,Y_\sigma)=(0,1)\) or \((1,0)\) [2507.11257].

The central theorem states that for sufficiently large \(m\) there exists \(s=\Theta(m)\) such that every deterministic zero-error simultaneous-message protocol for \(\mathrm{UniqueOverlap}_{m,s}\) has communication complexity \(\Omega(m)\). The proof uses “flippable” coordinates induced by short deterministic messages, then converts correctness into a cross-intersection constraint on families of supports, and finally applies Hilton’s bound for cross-intersecting families. The difficulty is specific to the promise structure: Charlie knows the unique intersection coordinate, so standard reductions from set disjointness or indexing do not directly apply [2507.11257].

The problem was introduced to prove the first deterministic lower bound for a connectivity decision problem in the distributed sketching model. Via a simulation using indistinguishable separated pairs in a bounded-intersection set family, the paper derives that for any super-constant \(k\le \gamma\sqrt{n}\), every deterministic zero-error one-round sketching algorithm deciding \(k\)-edge connectivity requires worst-case message length \(\Omega(k)\) bits per node. Here “UniqueOverlap” no longer concerns geometry of sets or tilings; it is a communication primitive whose unique overlap index drives a lower-bound reduction [2507.11257].

## 4. UniqueOverlap in planted optimization and overlap geometry

For modularity optimization on the \(k\)-block stochastic block model with \(k\ge 3\), the relevant overlap is between a candidate partition \(A\) and the planted partition \(P\), measured by
\[
d(A,P)=1-\max_{\pi\in S_k}\frac{1}{n}\sum_{i=1}^k |A_{\pi(i)}\cap P_i|.
\]
The main theorem establishes an **overlap gap property** for modularity: for any \(\nu\in(1/(2(k-1)),1/k)\), sufficiently near-optimal partitions satisfy either \(d(A,P)\le 1/(2(k-1))\) or \(d(A,P)\ge \nu\), and far near-optimal partitions do exist. A decoy partition obtained by merging two planted blocks lies at distance \(1/k\) and attains modularity near the optimal value, producing the far basin [2605.10911].

Within this landscape, the paper identifies a planted-side “unique overlap” phenomenon. It proves that if \(Q(A)>Q^\star(G)-\delta\), then \(d(A,P)\le \delta'+\epsilon\), where \(\delta'\) is determined by
\[
2\delta'(1-\delta'(k-1))=\frac{a+(k-1)b}{a-b}\,\delta.
\]
Thus every partition with modularity sufficiently near optimum is close to the planted partition. Combined with the OGP, near-optimal solutions are either tightly concentrated near \(P\) or lie in the decoy basin around distance \(1/k\), with no near-optimal partitions in the intermediate interval. Balanced-partition constraints remove this gap by making the analogous objective strictly decreasing in distance [2605.10911].

A useful contrast comes from random Boolean CSP. For random \(2\)-SAT below density \(1\), the set of satisfying assignments forms w.h.p. a single cluster, and for every \(q\in(0,1]\) there exist satisfying assignments with overlap \(\sim q\). For clustered random \(k\)-SAT near threshold, the overlap distribution has discontinuous support and different clusters are separated by linear Hamming distance. In both regimes, there is no single universal overlap value. This contrast clarifies that “UniqueOverlap” in planted modularity is not literal uniqueness of all overlaps, but rigidity of the near-optimal planted side under a specific objective [0703065].

## 5. Thresholds, scarcity, and one-point overlap in other settings

Several other works use overlap-related constructions to express sharp thresholds or minimal-overlap pathologies rather than canonicity. In combinatorics on words, every concatenation of \(10\) or more binary squares contains an overlap, and the bound \(10\) is best possible. Over a ternary alphabet, by contrast, there are infinitely long overlap-free words that are concatenations of squares. Here the salient feature is a sharp forcing threshold: overlap is unavoidable beyond a minimal number of square factors in the binary case [2605.28570].

In descriptive set theory, there is consistently a Borel set \(B\subseteq 2^\omega\) and a sequence \((\eta_\alpha)_{\alpha<\lambda}\) such that \((\eta_\alpha+B)\cap(\eta_\beta+B)\) is uncountable for all \(\alpha,\beta<\lambda\), but there is no perfect set \(P\) such that every distinct \(\eta,\nu\in P\) satisfy \(|(\eta+B)\cap(\nu+B)|\ge 6\). The construction separates “many” uncountable overlaps along a long sequence from “perfectly many” large overlaps, and the threshold \(6\) is a technical finite cap rather than an optimal bound [2302.12964].

In self-similar geometry, an explicit iterated function system on \([0,1]\) yields a totally disconnected attractor \(K\) with exactly one first-level overlap, namely \(K_3\cap K_4=\{h\}\), while the open set condition still fails. Thus even a unique one-point overlap does not imply OSC. The example is post-critically finite, and when \(\log p/\log r\notin\mathbb{Q}\) it fails the weak separation property, hence fails OSC despite the minimal overlap pattern [1809.08595].

A different extremal direction appears in edge unfoldings of convex polyhedra. For the banded geodesic domes \(P_n\) with \(n=160\cdot 4^L\) vertices, the fraction \(f(n)\) of nonoverlapping edge unfoldings tends to \(0\) as \(n\to\infty\), with explicit bounds such as \(f(n)\le (29/32)^{n/8}\). Nonoverlapping unfoldings do exist, but they are asymptotically rare. In this usage, “UniqueOverlap” is best read as prevalence of overlap rather than literal uniqueness [0801.4019].

## 6. Comparative structure of the term

The collected usages show that **UniqueOverlap** functions as a contextual label rather than a single technical invariant.

| Domain | Meaning of overlap rigidity | Representative result |
|---|---|---|
| Substitution tilings | Canonicity of overlap coincidence across admissible control points | Overlap coincidence \(\Leftrightarrow\) multiple strong coincidence at some level \(n(T)\) [1509.04471] |
| Overlapping substitutions | Unique, contradiction-free realization of allowed overlaps | Consistency from adjacency graphs in 1D and OSC + linear GIFS in 2D [2407.18666] |
| Communication complexity | Promise problem with a unique overlap coordinate | Deterministic simultaneous-message complexity \(\Omega(m)\) [2507.11257] |
| SBM modularity | Near-optimal partitions forced near planted or decoy basins | Modularity has OGP; near-optimal implies near-planted on the planted side [2605.10911] |
| Binary words | Sharp forcing threshold for unavoidable overlaps | \(10\) binary squares force an overlap, and \(10\) is best possible [2605.28570] |
| Borel translates | Large overlaps without perfect families of large overlaps | Pairwise uncountable overlaps for \(\lambda\) translates, but no perfect family with pairwise overlap \(\ge 6\) [2302.12964] |

A common pattern nonetheless recurs. In each setting, overlap is not merely an incidental intersection relation; it becomes a structural probe of the ambient object. In tilings it detects pure discreteness and model-set structure; in overlapping substitutions it governs global consistency; in communication lower bounds it isolates the irreducible information carried by a single shared coordinate; in planted optimization it encodes the geometry of near-optimal states; and in extremal combinatorics it appears as a sharp obstruction or rarity phenomenon. The term therefore identifies a research theme—control of overlap under strong structural constraints—rather than a single universally adopted definition.

Source: https://www.emergentmind.com/topics/uniqueoverlap