SearcharxivSearch

arXiv subjects

Franca Rinaldi

Publications and source records attributed to Franca Rinaldi.

2 recordsLinked to original sources

Proving hamiltonian properties in connected 4-regular graphs: an ILP-based approach

In this paper we study some open questions related to the smallest order $f({\cal C},\lnot {\cal H})$ of a 4-regular graph which has a connectivity property ${\cal C}$ but does not have a hamiltonian property ${\cal H}$. In particular, ${\cal C}$ is either connectivity, 2-connectivity or 1-toughness and ${\cal H}$ is hamiltonicity, homogeneously traceability or traceability. A standard theoretical approach to these questions had already been used in the literature, but did not succeed in determining the exact value of $f()$. Here we have chosen to use Integer Linear Programming and to encode the graphs that we are looking for as the binary solutions to a suitable set of linear inequalities. This way, there would exist a graph of order $n$ with certain properties if and only if the corresponding ILP had a feasible solution, which we have determined through a branch-and-cut procedure. By using our approach, we have been able to compute $f({\cal C},\lnot {\cal H})$ for all the pairs of considered properties with the exception of ${\cal C}=$1-toughness, ${\cal H}=$traceability. Even in this last case, we have nonetheless significantly reduced the interval $[LB, UB]$ in which $f({\cal C},\lnot {\cal H})$ was known to lie. Finally, we have shown that for each $n \geq f({\cal C},\lnot {\cal H})$ ($n \geq UB$ in the last case) there exists a 4-regular graph on $n$ vertices which has property ${\cal C}$ but not property ${\cal H}$.

math.CO

Vulnerability and power on networks

Inspired by socio-political scenarios, like dictatorships, in which a minority of people exercise control over a majority of weakly interconnected individuals, we propose vulnerability and power measures defined on groups of actors of networks. We establish an unexpected connection between network vulnerability and graph regularizability. We use the Shapley value of coalition games to introduce fresh notions of vulnerability and power at node level defined in terms of the corresponding measures at group level. We investigate the computational complexity of computing the defined measures, both at group and node levels, and provide effective methods to quantify them. Finally we test vulnerability and power on both artificial and real networks.

cs.SI