SLIDE 1
Polynomial Kernels for Hitting Forbidden Minors under Structural Parameterizations
Bart M.P. Jansen and Astrid Pieterse
SLIDE 2 Problem
𝐺 is a finite set of connected graphs 𝐺-minor free deletion Given undirected graph 𝐻 and budget 𝑐, can we remove 𝑐 vertices from 𝐻 such that it no longer has 𝐺-minors? 𝐼 is a minor of 𝐻
G
SLIDE 3
Problem
𝐺 is a finite set of connected graphs 𝐺-minor free deletion Given undirected graph 𝐻 and budget 𝑐, can we remove 𝑐 vertices from 𝐻 such that it no longer has 𝐺-minors? 𝐼 is a minor of 𝐻
SLIDE 4
Problem
𝐺 is a finite set of connected graphs 𝐺-minor free deletion Given undirected graph 𝐻 and budget 𝑐, can we remove 𝑐 vertices from 𝐻 such that it no longer has 𝐺-minors? 𝐼 is a minor of 𝐻
SLIDE 5
Problem
𝐺 is a finite set of connected graphs 𝐺-minor free deletion Given undirected graph 𝐻 and budget 𝑐, can we remove 𝑐 vertices from 𝐻 such that it no longer has 𝐺-minors? 𝐼 is a minor of 𝐻
SLIDE 6 Problem
𝐺 is a finite set of connected graphs 𝐺-minor free deletion Given undirected graph 𝐻 and budget 𝑐, can we remove 𝑐 vertices from 𝐻 such that it no longer has 𝐺-minors? 𝐼 is a minor of 𝐻
H
SLIDE 7 𝐺-minor free deletion
Generalizes many known problems Vertex Cover for 𝐺 = {𝐿2} Can we remove 𝑐 vertices, such that 𝐻 becomes edgeless?
Remove 3 vertices
SLIDE 8 𝐺-minor free deletion
Generalizes many known problems Vertex Cover for 𝐺 = {𝐿2} Can we remove 𝑐 vertices, such that 𝐻 becomes edgeless? Feedback Vertex Set for 𝐺 = {𝐿3} Can we remove 𝑐 vertices, such that 𝐻 becomes acyclic?
Remove 1 vertex Remove 3 vertices
SLIDE 9 Kernelization
𝐺-minor free deletion is NP-hard
- Do preprocessing
- Use an additional parameter 𝑙 to measure complexity
For which complexity measure, is good preprocessing possible?
𝑦
𝑙
𝑦′
𝑙′
𝑜 bits 𝑔(𝑙) bits 𝑞𝑝𝑚𝑧 𝑦 , 𝑙 time
SLIDE 10 Previous work
General problem [Fomin, Jansen, Pilipczuk, J. Comput. Syst. Sci.’12] Let 𝑌 be a vertex cover of 𝐻, there is a kernel of size 𝑞𝑝𝑚𝑧 𝑌 for 𝐺-minor free deletion General parameter [Bougeret, Sau, IPEC’17] Let 𝑌 be a modulator to treedepth 𝜃, there is a kernel of size 𝑞𝑝𝑚𝑧 𝑌 for vertex cover
modulator to treedepth 1 = vertex cover vertex cover = {𝐿2}-minor free deletion kernel 𝑐 𝑐′
|X|=3
SLIDE 11
Main result
We generalize both existing results, resolving an open question by Bougeret and Sau on FVS For more information & interesting proof techniques Come see the poster! Theorem 𝐺-minor free deletion parameterized by a modulator to treedepth 𝜃 has a polynomial kernel