Coarse Balanced Separators in Fat-Minor-Free Graphs

Abstract: Fat minors are a coarse analogue of graph minors where the subgraphs modeling vertices and edges of the embedded graph are required to be distant from each other, instead of just being disjoint. In this paper, we give a coarse analogue of the classic theorem that an $n$-vertex graph excluding a fixed minor admits a balanced separator of size $O(\sqrt{n})$. Specifically, we prove that for every integer $d$, real $\varepsilon>0$, and graph $H$, there exist constants $c$ and $r$ such that every $n$-vertex graph $G$ excluding $H$ as a $d$-fat minor admits a set $S \subseteq V(G)$ that is a balanced separator of $G$ and can be covered by $c n^{\frac{1}{2}+\varepsilon}$ balls of radius $r$ in $G$. Our proof also works in the weighted setting where the balance of the separator is measured with respect to any weight function on the vertices, and is effective: we obtain a randomized polynomial-time algorithm to compute either such a balanced separator, or a $d$-fat model of $H$ in $G$.

• The paper introduces a coarse analog of balanced separators in fat-minor-free graphs, proving every n-vertex graph admits a separator coverable by O(n^(1/2+ε)) balls.
• It leverages sparse clustering and multicommodity flow duality to construct separators with bounded radii, ensuring algorithmic efficiency via randomized methods.
• The results extend to induced-minor-free graphs with specific bounds and open avenues for improved radius versus size trade-offs in metric graph theory.

Coarse Balanced Separators in Fat-Minor-Free Graphs: An Expert Perspective

Introduction and Context

The paper "Coarse Balanced Separators in Fat-Minor-Free Graphs" (2604.11318) extends classical structural graph theory into the metric/coarse setting, specifically addressing separator theorems for fat-minor-free graphs. Fat minors generalize the minor relation by relaxing disjointness to distance requirements, focusing on separation in the metric sense rather than strict subgraph containment. The work is motivated by foundational conjectures in coarse graph theory, notably Georgakopoulos and Papasoglu's suggestion that $d$-fat-minor-free graphs might be quasi-isometric to minor-free graphs—a conjecture now known to be false (Davies et al., 2024, Albrechtsen et al., 21 Aug 2025, Albrechtsen et al., 9 Jan 2026). Nevertheless, the existence and computation of small separators in this coarser context remains an open and significant structural question.

Main Contributions

The paper's core result is a coarse analog of the classic balanced separator theorem for minor-free graphs:

Theorem 1. Given any fixed graph $H$, integer $d > 0$, and $\epsilon > 0$, there exist constants $c$ and $r$ such that every $n$-vertex graph excluding $H$ as a $d$-fat minor admits a balanced separator that can be covered by $c n^{1/2+\epsilon}$ balls of radius $H$0. The result is extended to the weighted case, and an efficient randomized algorithm is provided that either constructs such a separator or finds a $H$1-fat minor model of $H$2.

Their construction leverages recent advances in clustering methods and the duality between multicommodity flows and balanced separators [Leighton-Rao, 1999; (2604.11318, Chudnovsky et al., 11 Mar 2026)]. Notably, the separator guarantee entails a sublinear number of "local" (bounded-radius) clusters covering the separator, relaxing the requirement of separator sparsity into a metric cover condition.

Separators for Induced-Minor-Free Classes

Two tangential results are established for induced-minor-free graphs, both of interest for fine-grained structural parameterization:

• For any fixed $H$3, every $H$4-vertex $H$5-induced-minor-free weighted graph admits a balanced separator coverable by $H$6 balls of radius $H$7.
• For any fixed $H$8, every $H$9-vertex $d > 0$0-induced-minor-free weighted graph admits a balanced separator coverable by $d > 0$1 balls of radius $d > 0$2.

These results build on new clustering and contraction techniques, connecting combinatorial structure (via domination and quotient graphs) with classic separator theorems. The authors suggest that further improvements in the "radius versus size" trade-off are likely, especially in the induced minor setting.

Technical Approach

Fat Minors and Cluster Partitions

The definition of a $d > 0$3-fat minor introduces a parameterized distance constraint between subgraphs modeling the vertices and edges of $d > 0$4 in $d > 0$5, replacing strict disjointness. The key challenge is that unlike minor-free graphs, $d > 0$6-fat-minor-free graphs lack a general quasi-isometric correspondence with minor-free classes.

The authors circumvent this by developing a sparse clustering (using Filtser's scattering partitions [Filtser, 2024]) in which any ball of radius $d > 0$7 intersects only $d > 0$8 clusters with bounded strong diameter. This clustering is crucial for ensuring that separators in the contracted graph "lift" to small, well-covered sets in the original metric.

Duality: Concurrent Flows and Separators

Building on the classic Leighton–Rao duality, the separator construction proceeds via iterative applications of multicommodity flow versus sparsity separation: either one obtains a low-congestion concurrent flow or a small separator. This is tailored to the coarse setting through a reduction to bounded-diameter quotient graphs.

Crucially, the final separator is described not by its cardinality but by its coverability—i.e., the number of bounded-radius balls needed to cover it. This reframed notion of size (from cardinality to "local" bulk) is fundamental in the metric context.

Algorithmic Aspects

The separator construction is algorithmic and randomized, with efficient (polynomial) running time. The number of clusters (and thus balls covering the separator) is a function of $d > 0$9, and the ball radius is a function of $\epsilon > 0$0 and $\epsilon > 0$1. The probabilistic method underwrites the existence and algorithmic feasibility of the separator, with failure probability easily reduced via repetition.

Key Numerical Bounds and Claims

• Separator Size: For any $\epsilon > 0$2, the covering number is $\epsilon > 0$3, where $\epsilon > 0$4 denotes dependence polynomial in $\epsilon > 0$5.
• Separator Radius: The covering balls have radius $\epsilon > 0$6.
• Induced Minor Results: For $\epsilon > 0$7-induced-minor-free graphs, the current bounds are $\epsilon > 0$8 (radius $\epsilon > 0$9) and for $c$0, $c$1 (radius $c$2).

The paper conjectures that the polylogarithmic or even constant bounds in both radius and covering number might be achievable with improved clustering constructions, aligning with known tight results in the classic minor setting.

Practical and Theoretical Implications

This work shows that, despite the absence of quasi-isometric analogs for fat-minor-free graphs, many of the essential combinatorial consequences of classic minor-exclusion carry over in a coarse sense. The balanced separator property is central to algorithms for divide-and-conquer, parameterized complexity, and structural decomposition (e.g., bounded treewidth and VC-dimension).

• Algorithmic Implications: The existence of such coarse separators implies algorithmic tools such as subexponential parameterized algorithms, efficient approximation schemes, and sparse structure learning for classes characterized by forbidden fat minors.
• Structural Theory: The coarse separator results inform ongoing work on coarse treewidth and related decompositional parameters (Abrishami et al., 27 Feb 2025, Chudnovsky et al., 11 Mar 2026).

In the induced minor context, the paper opens the door to refined separator theorems with lower exponents, particularly relevant for problems where induced minors rather than "fat" minors arise naturally.

Future Developments

The authors identify several open problems:

• Sharpening the covering number from $c$3 to $c$4 or constant, dependent only on $c$5 and $c$6.
• Determining whether coverable balanced separators of radius $c$7 exist in all $c$8-induced-minor-free graphs.
• Developing clustering schemes with improved intersection properties for balls of fixed radius.

A resolution of these would directly inform both the combinatorial decomposition theory and the design of efficient metric algorithms on fat-minor-free classes.

Conclusion

The paper advances the theory of coarse graph structure by establishing robust separator theorems for fat-minor-free and induced-minor-free graphs. The combination of sparse clustering, multicommodity flow duality, and algorithmic partitioning provides a framework for importing key separator properties to settings governed by metric, rather than strictly combinatorial, constraints. These results set the stage for both refined theoretical understanding and algorithmic advances in metric graph theory and discrete geometry.

References:

• "Coarse Balanced Separators in Fat-Minor-Free Graphs" (2604.11318)
• Leighton, F.T. & Rao, S. (1999). "Multicommodity max-flow min-cut theorems and their use in designing approximation algorithms."
• Chudnovsky, M. et al. "Induced Minors and Coarse Tree Decompositions" (Chudnovsky et al., 11 Mar 2026)
• Filtser, A. "Scattering and Sparse Partitions, and Their Applications" (2024)
• Georgakopoulos, A. & Papasoglu, P. (2025)
• Davies, J. et al. "Fat minors cannot be thinned (by quasi-isometries)" (Davies et al., 2024)

See full bibliography in original paper for further details.