Ramsey-goodness -- and otherwise
Abstract
A celebrated result of Chv\'atal, R\"odl, Szemer\'edi and Trotter states (in slightly weakened form) that, for every natural number , there is a constant r such that, for any connected n-vertex graph G with maximum degree , the Ramsey number R(G,G) is at most r n, provided n is sufficiently large. In 1987, Burr made a strong conjecture implying that one may take r = . However, Graham, R\"odl and Ruci\'nski showed, by taking G to be a suitable expander graph, that necessarily r > 2c for some constant c>0. We show that the use of expanders is essential: if we impose the additional restriction that the bandwidth of G be at most some function β (n) = o(n), then R(G,G) (2(G)+4)n≤ (2+6)n, i.e., r = 2 +6 suffices. On the other hand, we show that Burr's conjecture itself fails even for Pnk, the kth power of a path Pn. Brandt showed that for any c, if is sufficiently large, there are connected n-vertex graphs G with (G)≤ but R(G,K3)>cn. We show that, given and H, there are β>0 and n0 such that, if G is a connected graph on n n0 vertices with maximum degree at most and bandwidth at most β n, then we have R(G,H)=((H)-1)(n-1)+σ(H), where σ(H) is the smallest size of any part in any (H)-partition of H. We also show that the same conclusion holds without any restriction on the maximum degree of G if the bandwidth of G is at most ε(H) n/ n.
Turn this paper into a lesson
ArcXiv compiles a structured reading guide from this paper's metadata: plain-English importance, contributions, prerequisite concepts, which sections to read first, flashcards, and a quiz. Grounded in the abstract, never invented.