Improved bounds for the two-scale treewidth partition parameter

Determine whether the minimum threshold $\ell(p,q)$ has asymptotically better bounds than the currently known $O^*((pq)^9)$ bound, where every graph of treewidth at least $\ell(p,q)$ must contain pairwise disjoint connected subgraphs each of treewidth at least $p$ whose simultaneous contraction produces a minor of treewidth at least $q$.

Background

The paper defines (p,q)\ell(p,q) as the least integer such that every graph of treewidth at least (p,q)\ell(p,q) contains pairwise disjoint connected subgraphs G1,,GnG_1,\ldots,G_n, each with treewidth at least pp, and contracting these subgraphs yields a minor of treewidth at least qq. Using the Grid Minor Theorem, the authors obtain the bound (p,q)O((pq)9)\ell(p,q)\in O^*((pq)^9).

The parameter captures the simultaneous presence of many locally complex subgraphs and a globally complex contracted minor. The authors note that existing partitioning results establish the local treewidth condition but do not establish the contraction condition, motivating the search for sharper bounds.

References

Are there better bounds on $\ell$?

Polynomial Bounds in the Apex Minor Theorem  (2503.04228 - Hendrey et al., 6 Mar 2025) in Section ‘Open Problems’