English

A new framework for identifying most reliable graphs and a correction to the $K_{3,3}$-theorem

Combinatorics 2024-11-26 v2 Probability

Abstract

Given a multigraph GG, the all-terminal reliability R(G,p)R(G,p) is the probability that GG remains connected under percolation with parameter pp. Fixing the number of vertices nn and edges mm, we investigate which graphs maximize R(G,p)R(G,p) -- such graphs are called optimal -- paying particular attention to uniqueness and to whether the answer depends upon pp. We generalize the concept of a distillation and build a framework with which we identify all optimal graphs for which mn{1,2,3}m-n\in\{1,2,3\}. These graphs are uniformly optimal in pp. Most have been previously identified, but with serious problems, especially when mn=3m-n=3. We obtain partial results for mn{4,5}m-n\in\{4,5\}. For mn=3m-n=3, the optimal graphs were incorrectly identified by Wang in 1994, in the infinite number of cases where m5(mod9)m\equiv5\pmod{9} and m14m\geq14. This erroneous result concerns subdivisions of K3,3K_{3,3} and has been cited extensively, without any mistake being detected. While the optimal graphs were correctly described for other mm, the proof is fundamentally flawed. Our proof of the rectified statement is self-contained. For mn=4m-n=4, the optimal graphs were recently shown to depend upon pp for infinitely many mm. We find a new such set of mm-values, which gives a different perspective on why this phenomenon occurs and leads us to conjecture that uniformly optimal graphs exist only for finitely many mm. However, for mn=5m-n=5, we conjecture that there are again infinitely many uniformly optimal graphs.

Cite

@article{arxiv.2407.20217,
  title  = {A new framework for identifying most reliable graphs and a correction to the $K_{3,3}$-theorem},
  author = {Lorents F. Landgren and Jeffrey E. Steif},
  journal= {arXiv preprint arXiv:2407.20217},
  year   = {2024}
}

Comments

44 pages, 26 figures. Some shortened arguments, clarifications, copyediting, two new figures. Comments welcome