Track new results across pure and applied mathematics.
Search papers, labs, and topics across Lattice.
This paper establishes a new upper bound for the treewidth of a graph $G$ as $\mathrm{tw}(G)+1\leq\mathrm{tree}\text{-}α(G)\mathrm{tree}\text{-}χ(G)(1+\lfloor\log_2(\mathrm{tree}\text{-}α(G))\rfloor)$, improving upon existing quadratic bounds. The authors demonstrate the tightness of this bound by constructing graphs where the treewidth is $\Omega(nk\log_2 n)$ for given tree-independence and tree-chromatic numbers $n$ and $k$. This work provides a more refined understanding of the relationship between these graph parameters.
A logarithmic improvement on the treewidth bound using tree-independence and tree-chromatic numbers is achieved, with constructions proving its optimality.
For a graph $G$, let $\mathrm{tw}(G)$, $\mathrm{tree}\text{-}α(G)$, and $\mathrm{tree}\text{-}χ(G)$ denote its treewidth, tree-independence number, and tree-chromatic number, respectively. We prove that every nonempty graph $G$ satisfies $\mathrm{tw}(G)+1\leq\mathrm{tree}\text{-}α(G)\mathrm{tree}\text{-}χ(G)(1+\lfloor\log_2(\mathrm{tree}\text{-}α(G))\rfloor)$. This improves the quadratic bound of Koutsoutis, Krause, Liu, Redzic, and Ueckerdt. For every pair of integers $n,k\geq2$, we construct a graph $G$ with $\mathrm{tree}\text{-}α(G)=n$, $\mathrm{tree}\text{-}χ(G)=k$, and $\mathrm{tw}(G)=Ω(nk\log_2 n)$. Thus the upper bound is best possible up to a constant factor.