Seminario Avanzado de Teoría de Grafos
Transcription
Seminario Avanzado de Teoría de Grafos
Seminario Avanzado de Teorı́a de Grafos Flavia Bonomo DC, FCEyN, Universidad de Buenos Aires, Argentina 2016 Flavia Bonomo (DC–FCEN–UBA) SATG 2016 1/9 Outline of the course: algorithmic problems Maximum clique Maximum stable set Vertex coloring Minimum clique cover Minimum dominating set (All NP-complete in general.) Flavia Bonomo (DC–FCEN–UBA) SATG 2016 2/9 Outline of the course: graph classes and concepts modular-decomposition perfect circular-arc circle cographs chordal permutation interval block trees Flavia Bonomo (DC–FCEN–UBA) SATG 2016 3/9 Desirable properties for a graph class C Hereditary: G ∈ C and H induced subgraph of G then H ∈ C ! There exists a characterization by minimal forbidden induced subgraphs, even if the family is infinite. Hereditary on subgraphs: G ∈ C and H subgraph of G then H ∈ C. (Sometimes it is too much to require, every graph is a subgraph of a large enough clique.) Flavia Bonomo (DC–FCEN–UBA) SATG 2016 4/9 Desirable properties for a graph parameter ϕ Monotonicity: H induced subgraph of G then ϕ(H) ≤ ϕ(G ) (or ϕ(H) ≥ ϕ(G ), depending on the problem). Monotonicity on subgraphs: H subgraph of G then ϕ(H) ≤ ϕ(G ) (or ϕ(H) ≥ ϕ(G )). Monotonicity on spanning subgraphs: H spanning subgraph of G (V (H) = V (G )) then ϕ(H) ≤ ϕ(G ) (or ϕ(H) ≥ ϕ(G )). Problem Monotonic M. on subgraphs M. on spanning s. clique ≤ ≤ ≤ stable set ≤ x ≥ coloring ≤ ≤ ≤ clique cover ≤ x ≥ dominating set x x ≥ Flavia Bonomo (DC–FCEN–UBA) SATG 2016 5/9 Connectivity The property of being connected is not hereditary (hereditary version: cliques). Cannot be expressed by minimal forbidden induced subgraphs. Most of the problems (specially monotonic parameters) can be solved on each component. We will often assume connection but we will not require it. Flavia Bonomo (DC–FCEN–UBA) SATG 2016 6/9 Intersection graphs Trees and block graphs The very basics: trees and forests G is a forest iff it has no cycles Cn , n ≥ 3. ! characterization by forbidden induced subgraphs, indeed also by forbidden subgraphs a tree is a connected forest every non-trivial tree has at least two leaves ! dismantling/building sequence every vertex v that is not a leaf is a cutpoint and decomposes the tree into d(v ) parts ! decomposition/composition theorem every edge is a bridge ! decomposition/composition theorem for every vertex v and every k, Nk (v ) is a stable set ! level structure a tree is bipartite ! vertex partition Flavia Bonomo (DC–FCEN–UBA) SATG 2016 7/9 Intersection graphs Trees and block graphs The very basics: trees and forests Algorithmic use: List-coloring ! leaves + recursion Maximum stable set ! cutpoint/level structure + dynamic programming Minimum dominating set ! cutpoint/level structure + dynamic programming Minimum clique-covering ! forced using leaves Flavia Bonomo (DC–FCEN–UBA) SATG 2016 8/9 Intersection graphs Trees and block graphs How to generalize trees? Block graphs The property of cutpoints in trees is equivalent to every biconnected component (block) is a vertex or an edge. Block graph: every biconnected component (block) is a clique. Forbidden induced subgraphs: Cn , n ≥ 4, diamond. leaves ! end blocks Many algorithms for trees extend to block graphs. Diamond-free ⇔ every edge belongs to one maximal clique. diamond Flavia Bonomo (DC–FCEN–UBA) SATG 2016 9/9