Given once, 2019
Deleting to Structured Trees
Pruning a graph into a full binary tree is NP-hard, even if you are only deleting edges
An intermediate research talk, for graduate students and up.
Abstract
Draft, not yet checked by meFeedback Vertex Set asks for the fewest vertex deletions that leave a forest. I ask for more structure in what remains: delete a few vertices (or a few edges) so that the graph becomes a full binary tree, the kind found in phylogenetic trees and in some polymerization processes. Both versions are NP-hard. The edge version is the surprise. Deleting edges of minimum total weight to leave some tree is just a spanning tree computation, yet asking for a full binary tree makes the problem hard. The proofs build on a warm-up reduction for ternary trees. On the algorithmic side, both problems are FPT in the number of deletions.
The slides
As a PDF
The viewer needs JavaScript; the PDF below works without it.