New Bounds for Induced Turán Problems
Nathan S. Sheffield
TL;DR
The paper investigates the induced extremal number $\mathrm{ex}(n,H\text{-ind})$ in graphs forbidding $K_{s,s}$ and induced copies of a bipartite pattern $H$, aiming to relate it to the standard $\mathrm{ex}(n,H)$. It develops a framework of reductions to connected components and to the $2$-core, and proves a degeneracy-based upper bound $\mathrm{ex}(n,{K_{s,s}, H\text{-ind}}) \le O(n^{2-1/(20 r^4)})$ for $r$-degenerate bipartite $H$ using dependent random choice, with refined bounds when forbidding specific non-edges of $H$. The authors also establish bounds under edge-forbidding scenarios (Cases 1–2) that improve ex-dependence on degeneracy, and discuss potential incidence-geometry counterexamples that could challenge the conjecture relating induced and standard extremal numbers. Overall, the work connects induced extremal behavior to the degeneracy structure of $H$ and motivates further exploration of possible counterexamples from geometric incidence graphs. These results advance understanding of how induced-avoidance interacts with classical Turán-type extremal theory and identify key directions for tightening bounds and assessing conjectures.
Abstract
In a recent paper, Hunter, Milojević, Sudakov and Tomon consider the maximum number of edges in an $n$-vertex graph containing no copy of the complete bipartite graph $K_{s,s}$ and no induced copy of a "pattern" graph $H$. They conjecture that, for $s \geq |V(H)|$, this "induced extremal number" differs by at most a constant factor from the standard extremal number of $H$. Towards this, we give bounds on the induced extremal number in terms of degeneracy, which establish some non-trivial relationship between the induced and standard extremal numbers in general. We also show that (as in the case of standard extremal numbers) the induced extremal number is dominated by that of the 2-core of a single connected component. Finally, we present some graphs arising from incidence geometry which may serve as counterexamples to the conjecture.
