Lower bounds for piercing and coloring boxes

From MaRDI portal
(Redirected from Publication:6187716)



Abstract: Given a family mathcalB of axis-parallel boxes in mathbbRd, let au denote its piercing number, and u its independence number. It is an old question whether au/u can be arbitrarily large for given dgeq2. Here, for every u, we construct a family of axis-parallel boxes achieving augeq Omega_d(

u)cdotleft(frac{log u}{loglog

u} ight)^{d-2}. This not only answers the previous question for every dgeq3 positively, but also matches the best known upper bound up to double-logarithmic factors. Our main construction has further implications about the Ramsey and coloring properties of configurations of boxes as well. We show the existence of a family of n boxes in mathbbRd, whose intersection graph has clique and independence number Od(n1/2)cdotleft(fraclognloglognight)−(d−2)/2. This is the first improvement over the trivial upper bound Od(n1/2), and matches the best known lower bound up to double-logarithmic factors. Finally, for every omega satisfying fraclognloglognllomegalln1−varepsilon, we construct an intersection graph of n boxes with clique number at most omega, and chromatic number Omegad,varepsilon(omega)cdotleft(fraclognloglognight)d−2. This matches the best known upper bound up to a factor of Od((logw)(loglogn)d−2).












This page was built for publication: Lower bounds for piercing and coloring boxes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6187716)