On Milgram's construction and the Duke embedding conjectures

Abstract

Milgram constructed a 28-vertex cubic graph of genus 4 that disproved Duke's conjecture relating Betti number to minimum genus. We apply Milgram's method to construct to find graphs of higher genus violating Duke's conjecture, which gives a sharper bound on that relationship. These graphs are also counterexamples to a related conjecture of Nordhaus et al. on the relationship between minimum and maximum genera of graphs. As a side note, we give a simpler proof of correctness for Milgram's method and we show that Duke's conjecture is true for genus at most 3.