Transversals in hypergraphs Anders Yeo AndersYeo@gmail.com - - PowerPoint PPT Presentation

transversals in hypergraphs
SMART_READER_LITE
LIVE PREVIEW

Transversals in hypergraphs Anders Yeo AndersYeo@gmail.com - - PowerPoint PPT Presentation

Transversals in hypergraphs Anders Yeo AndersYeo@gmail.com Engineering Systems and Design Singapore University of Technology and Design 8 Somapah Road 487372, Singapore Based on several papers co-authored by people such as M. A. Henning and


slide-1
SLIDE 1

Transversals in hypergraphs

Anders Yeo

AndersYeo@gmail.com Engineering Systems and Design Singapore University of Technology and Design 8 Somapah Road 487372, Singapore

Based on several papers co-authored by people such as M. A. Henning and C. L¨

  • wenstein

Anders Yeo Transversals in hypergraphs

slide-2
SLIDE 2

This talk

A few definitions and applications. Known results and a number of conjectures. Progress on the Tuza-Vestergaard Conjecture. Progress on a conjecture for linear 4-uniform hypergraphs. Conclusion and further open problems.

Anders Yeo Transversals in hypergraphs

slide-3
SLIDE 3

This talk

A few definitions and applications. Known results and a number of conjectures. Progress on the Tuza-Vestergaard Conjecture. Progress on a conjecture for linear 4-uniform hypergraphs. Conclusion and further open problems.

Anders Yeo Transversals in hypergraphs

slide-4
SLIDE 4

This talk

A few definitions and applications. Known results and a number of conjectures. Progress on the Tuza-Vestergaard Conjecture. Progress on a conjecture for linear 4-uniform hypergraphs. Conclusion and further open problems.

Anders Yeo Transversals in hypergraphs

slide-5
SLIDE 5

This talk

A few definitions and applications. Known results and a number of conjectures. Progress on the Tuza-Vestergaard Conjecture. Progress on a conjecture for linear 4-uniform hypergraphs. Conclusion and further open problems.

Anders Yeo Transversals in hypergraphs

slide-6
SLIDE 6

This talk

A few definitions and applications. Known results and a number of conjectures. Progress on the Tuza-Vestergaard Conjecture. Progress on a conjecture for linear 4-uniform hypergraphs. Conclusion and further open problems.

Anders Yeo Transversals in hypergraphs

slide-7
SLIDE 7

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-8
SLIDE 8

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-9
SLIDE 9

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-10
SLIDE 10

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-11
SLIDE 11

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-12
SLIDE 12

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-13
SLIDE 13

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-14
SLIDE 14

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-15
SLIDE 15

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-16
SLIDE 16

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-17
SLIDE 17

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-18
SLIDE 18

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-19
SLIDE 19

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-20
SLIDE 20

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3

Not linear

x y u v

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-21
SLIDE 21

Definitions

A hypergraph consists of vertices and edges.

x a1 a2 a3 b1 b2 b3 a1 a1 a1

Not linear

x y u v x u

A linear hypergraph has no pair of vertices overlapping in more than one vertex. A transversal in a hypergraph is a vertex-set intersecting every

  • edge. τ(H) denotes the size of a minimum transversal in H.

Anders Yeo Transversals in hypergraphs

slide-22
SLIDE 22

Definitions

x a1 a2 a3 b1 b2 b3 x y u v

The rank of a hypergraph is the size of its largest edge. A hypergraph is k-uniform if all edges have size k. A 2-uniform hypergraph is equivalent to a graph. A hypergraph is r-regular if all verties belong to r edges. That is, they all have degree r.

Anders Yeo Transversals in hypergraphs

slide-23
SLIDE 23

Definitions

x a1 a2 a3 b1 b2 b3

rank = 3

x y u v

rank = 3

The rank of a hypergraph is the size of its largest edge. A hypergraph is k-uniform if all edges have size k. A 2-uniform hypergraph is equivalent to a graph. A hypergraph is r-regular if all verties belong to r edges. That is, they all have degree r.

Anders Yeo Transversals in hypergraphs

slide-24
SLIDE 24

Definitions

x a1 a2 a3 b1 b2 b3

rank = 3

x y u v

rank = 3

The rank of a hypergraph is the size of its largest edge. A hypergraph is k-uniform if all edges have size k. A 2-uniform hypergraph is equivalent to a graph. A hypergraph is r-regular if all verties belong to r edges. That is, they all have degree r.

Anders Yeo Transversals in hypergraphs

slide-25
SLIDE 25

Definitions

x a1 a2 a3 b1 b2 b3

rank = 3 3-uniform

x y u v

rank = 3 Not uniform

The rank of a hypergraph is the size of its largest edge. A hypergraph is k-uniform if all edges have size k. A 2-uniform hypergraph is equivalent to a graph. A hypergraph is r-regular if all verties belong to r edges. That is, they all have degree r.

Anders Yeo Transversals in hypergraphs

slide-26
SLIDE 26

Definitions

x a1 a2 a3 b1 b2 b3

rank = 3 3-uniform

x y u v

rank = 3 Not uniform

The rank of a hypergraph is the size of its largest edge. A hypergraph is k-uniform if all edges have size k. A 2-uniform hypergraph is equivalent to a graph. A hypergraph is r-regular if all verties belong to r edges. That is, they all have degree r.

Anders Yeo Transversals in hypergraphs

slide-27
SLIDE 27

Definitions

x a1 a2 a3 b1 b2 b3

rank = 3 3-uniform Not regular

x y u v

rank = 3 Not uniform 2-regular

The rank of a hypergraph is the size of its largest edge. A hypergraph is k-uniform if all edges have size k. A 2-uniform hypergraph is equivalent to a graph. A hypergraph is r-regular if all verties belong to r edges. That is, they all have degree r.

Anders Yeo Transversals in hypergraphs

slide-28
SLIDE 28

Applications

3-SAT Colinear points in the plane. Total domination Phylogenetic trees in Biology. Feedback vertex sets (in tournaments). Coloring in hypergraphs. 2-coloring is equivalent to vertex-disjoint transversals. And many more..... Why?

Anders Yeo Transversals in hypergraphs

slide-29
SLIDE 29

Applications

3-SAT Colinear points in the plane. Total domination Phylogenetic trees in Biology. Feedback vertex sets (in tournaments). Coloring in hypergraphs. 2-coloring is equivalent to vertex-disjoint transversals. And many more..... Why?

Anders Yeo Transversals in hypergraphs

slide-30
SLIDE 30

Applications

3-SAT Colinear points in the plane. Total domination Phylogenetic trees in Biology. Feedback vertex sets (in tournaments). Coloring in hypergraphs. 2-coloring is equivalent to vertex-disjoint transversals. And many more..... Why?

Anders Yeo Transversals in hypergraphs

slide-31
SLIDE 31

Applications

3-SAT Colinear points in the plane. Total domination Phylogenetic trees in Biology. Feedback vertex sets (in tournaments). Coloring in hypergraphs. 2-coloring is equivalent to vertex-disjoint transversals. And many more..... Why?

Anders Yeo Transversals in hypergraphs

slide-32
SLIDE 32

Applications

3-SAT Colinear points in the plane. Total domination Phylogenetic trees in Biology. Feedback vertex sets (in tournaments). Coloring in hypergraphs. 2-coloring is equivalent to vertex-disjoint transversals. And many more..... Why?

Anders Yeo Transversals in hypergraphs

slide-33
SLIDE 33

Applications

3-SAT Colinear points in the plane. Total domination Phylogenetic trees in Biology. Feedback vertex sets (in tournaments). Coloring in hypergraphs. 2-coloring is equivalent to vertex-disjoint transversals. And many more..... Why?

Anders Yeo Transversals in hypergraphs

slide-34
SLIDE 34

Applications

3-SAT Colinear points in the plane. Total domination Phylogenetic trees in Biology. Feedback vertex sets (in tournaments). Coloring in hypergraphs. 2-coloring is equivalent to vertex-disjoint transversals. And many more..... Why?

Anders Yeo Transversals in hypergraphs

slide-35
SLIDE 35

Applications

3-SAT Colinear points in the plane. Total domination Phylogenetic trees in Biology. Feedback vertex sets (in tournaments). Coloring in hypergraphs. 2-coloring is equivalent to vertex-disjoint transversals. And many more..... Why?

Anders Yeo Transversals in hypergraphs

slide-36
SLIDE 36

3-SAT

Consider an instance, I, of 3-SAT with n vriables and m clauses. For example, I = (v1 ∨ v2 ∨ v3) ∧ (v2 ∨ v3 ∨ v4) ∧ · · · . · · ·

v1 v1 v2 v2 v3 v3 v4 v4 v5 v5 vn vn

We build a hypergraph, H, with 2n vertices as above. Each variable gives rise to a 2-edge. Each clause gives rise to a 3-edge. H has a transversal of size n if and only if I is satisfiable. We believe this reduction may be used to improve currently best algorithms for 3-SAT.

Anders Yeo Transversals in hypergraphs

slide-37
SLIDE 37

3-SAT

Consider an instance, I, of 3-SAT with n vriables and m clauses. For example, I = (v1 ∨ v2 ∨ v3) ∧ (v2 ∨ v3 ∨ v4) ∧ · · · . · · ·

v1 v1 v2 v2 v3 v3 v4 v4 v5 v5 vn vn

We build a hypergraph, H, with 2n vertices as above. Each variable gives rise to a 2-edge. Each clause gives rise to a 3-edge. H has a transversal of size n if and only if I is satisfiable. We believe this reduction may be used to improve currently best algorithms for 3-SAT.

Anders Yeo Transversals in hypergraphs

slide-38
SLIDE 38

3-SAT

Consider an instance, I, of 3-SAT with n vriables and m clauses. For example, I = (v1 ∨ v2 ∨ v3) ∧ (v2 ∨ v3 ∨ v4) ∧ · · · . · · ·

v1 v1 v2 v2 v3 v3 v4 v4 v5 v5 vn vn

We build a hypergraph, H, with 2n vertices as above. Each variable gives rise to a 2-edge. Each clause gives rise to a 3-edge. H has a transversal of size n if and only if I is satisfiable. We believe this reduction may be used to improve currently best algorithms for 3-SAT.

Anders Yeo Transversals in hypergraphs

slide-39
SLIDE 39

3-SAT

Consider an instance, I, of 3-SAT with n vriables and m clauses. For example, I = (v1 ∨ v2 ∨ v3) ∧ (v2 ∨ v3 ∨ v4) ∧ · · · . · · ·

v1 v1 v2 v2 v3 v3 v4 v4 v5 v5 vn vn

We build a hypergraph, H, with 2n vertices as above. Each variable gives rise to a 2-edge. Each clause gives rise to a 3-edge. H has a transversal of size n if and only if I is satisfiable. We believe this reduction may be used to improve currently best algorithms for 3-SAT.

Anders Yeo Transversals in hypergraphs

slide-40
SLIDE 40

3-SAT

Consider an instance, I, of 3-SAT with n vriables and m clauses. For example, I = (v1 ∨ v2 ∨ v3) ∧ (v2 ∨ v3 ∨ v4) ∧ · · · . · · ·

v1 v1 v2 v2 v3 v3 v4 v4 v5 v5 vn vn

We build a hypergraph, H, with 2n vertices as above. Each variable gives rise to a 2-edge. Each clause gives rise to a 3-edge. H has a transversal of size n if and only if I is satisfiable. We believe this reduction may be used to improve currently best algorithms for 3-SAT.

Anders Yeo Transversals in hypergraphs

slide-41
SLIDE 41

3-SAT

Consider an instance, I, of 3-SAT with n vriables and m clauses. For example, I = (v1 ∨ v2 ∨ v3) ∧ (v2 ∨ v3 ∨ v4) ∧ · · · . · · ·

v1 v1 v2 v2 v3 v3 v4 v4 v5 v5 vn vn

We build a hypergraph, H, with 2n vertices as above. Each variable gives rise to a 2-edge. Each clause gives rise to a 3-edge. H has a transversal of size n if and only if I is satisfiable. We believe this reduction may be used to improve currently best algorithms for 3-SAT.

Anders Yeo Transversals in hypergraphs

slide-42
SLIDE 42

3-SAT

Consider an instance, I, of 3-SAT with n vriables and m clauses. For example, I = (v1 ∨ v2 ∨ v3) ∧ (v2 ∨ v3 ∨ v4) ∧ · · · . · · ·

v1 v1 v2 v2 v3 v3 v4 v4 v5 v5 vn vn

We build a hypergraph, H, with 2n vertices as above. Each variable gives rise to a 2-edge. Each clause gives rise to a 3-edge. H has a transversal of size n if and only if I is satisfiable. We believe this reduction may be used to improve currently best algorithms for 3-SAT.

Anders Yeo Transversals in hypergraphs

slide-43
SLIDE 43

Colinear points in the plane

We are given a number of points in the plane. Can we find the largest set of points where no 3 are co-linear (lie

  • n a line)?

Add 3-edges between all sets of 3 points that are colinear. Remove the points of a minimum transversal. If no four points are co-linear then the resulting hypergraph is linear.

Anders Yeo Transversals in hypergraphs

slide-44
SLIDE 44

Colinear points in the plane

We are given a number of points in the plane. Can we find the largest set of points where no 3 are co-linear (lie

  • n a line)?

Add 3-edges between all sets of 3 points that are colinear. Remove the points of a minimum transversal. If no four points are co-linear then the resulting hypergraph is linear.

Anders Yeo Transversals in hypergraphs

slide-45
SLIDE 45

Colinear points in the plane

We are given a number of points in the plane. Can we find the largest set of points where no 3 are co-linear (lie

  • n a line)?

Add 3-edges between all sets of 3 points that are colinear. Remove the points of a minimum transversal. If no four points are co-linear then the resulting hypergraph is linear.

Anders Yeo Transversals in hypergraphs

slide-46
SLIDE 46

Colinear points in the plane

We are given a number of points in the plane. Can we find the largest set of points where no 3 are co-linear (lie

  • n a line)?

Add 3-edges between all sets of 3 points that are colinear. Remove the points of a minimum transversal. If no four points are co-linear then the resulting hypergraph is linear.

Anders Yeo Transversals in hypergraphs

slide-47
SLIDE 47

Colinear points in the plane

We are given a number of points in the plane. Can we find the largest set of points where no 3 are co-linear (lie

  • n a line)?

Add 3-edges between all sets of 3 points that are colinear. Remove the points of a minimum transversal. If no four points are co-linear then the resulting hypergraph is linear.

Anders Yeo Transversals in hypergraphs

slide-48
SLIDE 48

Colinear points in the plane

We are given a number of points in the plane. Can we find the largest set of points where no 3 are co-linear (lie

  • n a line)?

Add 3-edges between all sets of 3 points that are colinear. Remove the points of a minimum transversal. If no four points are co-linear then the resulting hypergraph is linear.

Anders Yeo Transversals in hypergraphs

slide-49
SLIDE 49

Colinear points in the plane

We are given a number of points in the plane. Can we find the largest set of points where no 3 are co-linear (lie

  • n a line)?

Add 3-edges between all sets of 3 points that are colinear. Remove the points of a minimum transversal. If no four points are co-linear then the resulting hypergraph is linear.

Anders Yeo Transversals in hypergraphs

slide-50
SLIDE 50

Total Domination

S is a total dominating set in G, if and only if every vertex in G has a neighbour in S. G

u1 u2 u3 u4 u5

G ′

v1 v2 v3 v4 v5 v6

γt(G) denotes the size of a smallest total dominating set in G.

Anders Yeo Transversals in hypergraphs

slide-51
SLIDE 51

Total Domination

S is a total dominating set in G, if and only if every vertex in G has a neighbour in S. G

u1 u2 u3 u4 u5

G ′

v1 v2 v3 v4 v5 v6

γt(G) denotes the size of a smallest total dominating set in G.

Anders Yeo Transversals in hypergraphs

slide-52
SLIDE 52

Total Domination

S is a total dominating set in G, if and only if every vertex in G has a neighbour in S. G

u1 u2 u3 u4 u5

G ′

v1 v2 v3 v4 v5 v6

γt(G) denotes the size of a smallest total dominating set in G.

Anders Yeo Transversals in hypergraphs

slide-53
SLIDE 53

Total Domination

S is a total dominating set in G, if and only if every vertex in G has a neighbour in S. G

u1 u2 u3 u4 u5

G ′

v1 v2 v3 v4 v5 v6

γt(G) denotes the size of a smallest total dominating set in G.

Anders Yeo Transversals in hypergraphs

slide-54
SLIDE 54

Total Domination

S is a total dominating set in G, if and only if every vertex in G has a neighbour in S. G

u1 u2 u3 u4 u5

G ′

v1 v2 v3 v4 v5 v6

γt(G) denotes the size of a smallest total dominating set in G.

Anders Yeo Transversals in hypergraphs

slide-55
SLIDE 55

Total Domination

S is a total dominating set in G, if and only if every vertex in G has a neighbour in S. G

u1 u2 u3 u4 u5

G ′

v1 v2 v3 v4 v5 v6

γt(G) denotes the size of a smallest total dominating set in G.

Anders Yeo Transversals in hypergraphs

slide-56
SLIDE 56

Total Domination

S is a total dominating set in G, if and only if every vertex in G has a neighbour in S. G

u1 u2 u3 u4 u5

G ′

v1 v2 v3 v4 v5 v6

γt(G) denotes the size of a smallest total dominating set in G.

Anders Yeo Transversals in hypergraphs

slide-57
SLIDE 57

Total Domination

S is a total dominating set in G, if and only if every vertex in G has a neighbour in S. G

u1 u2 u3 u4 u5

G ′

v1 v2 v3 v4 v5 v6

γt(G) denotes the size of a smallest total dominating set in G.

Anders Yeo Transversals in hypergraphs

slide-58
SLIDE 58

Total Domination, Open Neighbourhood hypergraph

Given G, the Open Neighbourhood Hypergraph, ONH(G), of G has vertex-set V (G) and edge-set {N(v) | v ∈ V (G)}. G

u1 u2 u3 u4 u5

ONH(G)

u1 u2 u3 u4 u5

A transversal in ONH(G) is equivalent to a total dominating set in G. Most recent results on Total Domination are found using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-59
SLIDE 59

Total Domination, Open Neighbourhood hypergraph

Given G, the Open Neighbourhood Hypergraph, ONH(G), of G has vertex-set V (G) and edge-set {N(v) | v ∈ V (G)}. G

u1 u2 u3 u4 u5

ONH(G)

u1 u2 u3 u4 u5

A transversal in ONH(G) is equivalent to a total dominating set in G. Most recent results on Total Domination are found using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-60
SLIDE 60

Total Domination, Open Neighbourhood hypergraph

Given G, the Open Neighbourhood Hypergraph, ONH(G), of G has vertex-set V (G) and edge-set {N(v) | v ∈ V (G)}. G

u1 u2 u3 u4 u5

ONH(G)

u1 u2 u3 u4 u5

A transversal in ONH(G) is equivalent to a total dominating set in G. Most recent results on Total Domination are found using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-61
SLIDE 61

Total Domination, Open Neighbourhood hypergraph

Given G, the Open Neighbourhood Hypergraph, ONH(G), of G has vertex-set V (G) and edge-set {N(v) | v ∈ V (G)}. G

u1 u2 u3 u4 u5

ONH(G)

u1 u2 u3 u4 u5

A transversal in ONH(G) is equivalent to a total dominating set in G. Most recent results on Total Domination are found using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-62
SLIDE 62

Total Domination, Open Neighbourhood hypergraph

Given G, the Open Neighbourhood Hypergraph, ONH(G), of G has vertex-set V (G) and edge-set {N(v) | v ∈ V (G)}. G

u1 u2 u3 u4 u5

ONH(G)

u1 u2 u3 u4 u5

A transversal in ONH(G) is equivalent to a total dominating set in G. Most recent results on Total Domination are found using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-63
SLIDE 63

Total Domination, Open Neighbourhood hypergraph

Given G, the Open Neighbourhood Hypergraph, ONH(G), of G has vertex-set V (G) and edge-set {N(v) | v ∈ V (G)}. G

u1 u2 u3 u4 u5

ONH(G)

u1 u2 u3 u4 u5

A transversal in ONH(G) is equivalent to a total dominating set in G. Most recent results on Total Domination are found using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-64
SLIDE 64

Total Domination, Open Neighbourhood hypergraph

Given G, the Open Neighbourhood Hypergraph, ONH(G), of G has vertex-set V (G) and edge-set {N(v) | v ∈ V (G)}. G

u1 u2 u3 u4 u5

ONH(G)

u1 u2 u3 u4 u5

A transversal in ONH(G) is equivalent to a total dominating set in G. Most recent results on Total Domination are found using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-65
SLIDE 65

Total Domination, Open Neighbourhood hypergraph

Given G, the Open Neighbourhood Hypergraph, ONH(G), of G has vertex-set V (G) and edge-set {N(v) | v ∈ V (G)}. G

u1 u2 u3 u4 u5

ONH(G)

u1 u2 u3 u4 u5

A transversal in ONH(G) is equivalent to a total dominating set in G. Most recent results on Total Domination are found using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-66
SLIDE 66

More....

In Phylogenetic trees we want to remove as few species as possible such that the known Phylogenetic trees dont contradict each other. This can be modelled as a transversal problem in 3-uniform hypergraphs. Feedback vertex sets (in tournaments). Given a graph or digraph put a hyperdege around the vertex set of all chordless

  • cycles. A transversal in the resulting hypergraph is a minimum

feedback vertex set. For tournaments all chordless cycles are 3-cycles so our hypergraph is 3-uniform. We have best known algorithms for both the above using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-67
SLIDE 67

More....

In Phylogenetic trees we want to remove as few species as possible such that the known Phylogenetic trees dont contradict each other. This can be modelled as a transversal problem in 3-uniform hypergraphs. Feedback vertex sets (in tournaments). Given a graph or digraph put a hyperdege around the vertex set of all chordless

  • cycles. A transversal in the resulting hypergraph is a minimum

feedback vertex set. For tournaments all chordless cycles are 3-cycles so our hypergraph is 3-uniform. We have best known algorithms for both the above using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-68
SLIDE 68

More....

In Phylogenetic trees we want to remove as few species as possible such that the known Phylogenetic trees dont contradict each other. This can be modelled as a transversal problem in 3-uniform hypergraphs. Feedback vertex sets (in tournaments). Given a graph or digraph put a hyperdege around the vertex set of all chordless

  • cycles. A transversal in the resulting hypergraph is a minimum

feedback vertex set. For tournaments all chordless cycles are 3-cycles so our hypergraph is 3-uniform. We have best known algorithms for both the above using transversals in hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-69
SLIDE 69

Known results and conjectures, order and size

For connected k-uniform hypergraphs there are bounds just based

  • n the order, n, and size, m.

(Erd¨

  • s & Tuza) If k = 2 then τ(H) ≤ 2(n+m+1)

7

. (Chv´ atal & McDiarmid) If k ≥ 2 then τ(H) ≤ n+⌊k/2⌋m

⌊3k/2⌋ .

The above bounds are tight for hy- pergraphs of average degree 2. For larger average degrees not much is known.

k 2 k 2 k 2

n = 3k/2 m = 3 τ = 2

The following is tight for average degrees 3-4. (Thomass´ e & Yeo) If k = 4 then τ(H) ≤ 5n+4m

21

.

Anders Yeo Transversals in hypergraphs

slide-70
SLIDE 70

Known results and conjectures, order and size

For connected k-uniform hypergraphs there are bounds just based

  • n the order, n, and size, m.

(Erd¨

  • s & Tuza) If k = 2 then τ(H) ≤ 2(n+m+1)

7

. (Chv´ atal & McDiarmid) If k ≥ 2 then τ(H) ≤ n+⌊k/2⌋m

⌊3k/2⌋ .

The above bounds are tight for hy- pergraphs of average degree 2. For larger average degrees not much is known.

k 2 k 2 k 2

n = 3k/2 m = 3 τ = 2

The following is tight for average degrees 3-4. (Thomass´ e & Yeo) If k = 4 then τ(H) ≤ 5n+4m

21

.

Anders Yeo Transversals in hypergraphs

slide-71
SLIDE 71

Known results and conjectures, order and size

For connected k-uniform hypergraphs there are bounds just based

  • n the order, n, and size, m.

(Erd¨

  • s & Tuza) If k = 2 then τ(H) ≤ 2(n+m+1)

7

. (Chv´ atal & McDiarmid) If k ≥ 2 then τ(H) ≤ n+⌊k/2⌋m

⌊3k/2⌋ .

The above bounds are tight for hy- pergraphs of average degree 2. For larger average degrees not much is known.

k 2 k 2 k 2

n = 3k/2 m = 3 τ = 2

The following is tight for average degrees 3-4. (Thomass´ e & Yeo) If k = 4 then τ(H) ≤ 5n+4m

21

.

Anders Yeo Transversals in hypergraphs

slide-72
SLIDE 72

Known results and conjectures, order and size

For connected k-uniform hypergraphs there are bounds just based

  • n the order, n, and size, m.

(Erd¨

  • s & Tuza) If k = 2 then τ(H) ≤ 2(n+m+1)

7

. (Chv´ atal & McDiarmid) If k ≥ 2 then τ(H) ≤ n+⌊k/2⌋m

⌊3k/2⌋ .

The above bounds are tight for hy- pergraphs of average degree 2. For larger average degrees not much is known.

k 2 k 2 k 2

n = 3k/2 m = 3 τ = 2

The following is tight for average degrees 3-4. (Thomass´ e & Yeo) If k = 4 then τ(H) ≤ 5n+4m

21

.

Anders Yeo Transversals in hypergraphs

slide-73
SLIDE 73

Known results and conjectures, corollaries

Given the results on the previous slide, we automatically get results for total domination in graphs with given minimum degree. If k = 3 and n = m then τ(H) ≤ n/2. So, γt(G) ≤ n/4 when δ(G) ≥ 3. If k = 4 and n = m then τ(H) ≤ 3n/7. So, γt(G) ≤ 3n/7 when δ(G) ≥ 4. Conjecture: If k = 5 and n = m then τ(H) ≤ 4n/11. The above conjecture would be best possible. Open Problem: For k ≥ 4 prove better bounds for linear hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-74
SLIDE 74

Known results and conjectures, corollaries

Given the results on the previous slide, we automatically get results for total domination in graphs with given minimum degree. If k = 3 and n = m then τ(H) ≤ n/2. So, γt(G) ≤ n/4 when δ(G) ≥ 3. If k = 4 and n = m then τ(H) ≤ 3n/7. So, γt(G) ≤ 3n/7 when δ(G) ≥ 4. Conjecture: If k = 5 and n = m then τ(H) ≤ 4n/11. The above conjecture would be best possible. Open Problem: For k ≥ 4 prove better bounds for linear hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-75
SLIDE 75

Known results and conjectures, corollaries

Given the results on the previous slide, we automatically get results for total domination in graphs with given minimum degree. If k = 3 and n = m then τ(H) ≤ n/2. So, γt(G) ≤ n/4 when δ(G) ≥ 3. If k = 4 and n = m then τ(H) ≤ 3n/7. So, γt(G) ≤ 3n/7 when δ(G) ≥ 4. Conjecture: If k = 5 and n = m then τ(H) ≤ 4n/11. The above conjecture would be best possible. Open Problem: For k ≥ 4 prove better bounds for linear hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-76
SLIDE 76

Known results and conjectures, corollaries

Given the results on the previous slide, we automatically get results for total domination in graphs with given minimum degree. If k = 3 and n = m then τ(H) ≤ n/2. So, γt(G) ≤ n/4 when δ(G) ≥ 3. If k = 4 and n = m then τ(H) ≤ 3n/7. So, γt(G) ≤ 3n/7 when δ(G) ≥ 4. Conjecture: If k = 5 and n = m then τ(H) ≤ 4n/11. The above conjecture would be best possible. Open Problem: For k ≥ 4 prove better bounds for linear hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-77
SLIDE 77

Known results and conjectures, corollaries

Given the results on the previous slide, we automatically get results for total domination in graphs with given minimum degree. If k = 3 and n = m then τ(H) ≤ n/2. So, γt(G) ≤ n/4 when δ(G) ≥ 3. If k = 4 and n = m then τ(H) ≤ 3n/7. So, γt(G) ≤ 3n/7 when δ(G) ≥ 4. Conjecture: If k = 5 and n = m then τ(H) ≤ 4n/11. The above conjecture would be best possible. Open Problem: For k ≥ 4 prove better bounds for linear hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-78
SLIDE 78

Known results and conjectures, corollaries

Given the results on the previous slide, we automatically get results for total domination in graphs with given minimum degree. If k = 3 and n = m then τ(H) ≤ n/2. So, γt(G) ≤ n/4 when δ(G) ≥ 3. If k = 4 and n = m then τ(H) ≤ 3n/7. So, γt(G) ≤ 3n/7 when δ(G) ≥ 4. Conjecture: If k = 5 and n = m then τ(H) ≤ 4n/11. The above conjecture would be best possible. Open Problem: For k ≥ 4 prove better bounds for linear hypergraphs.

Anders Yeo Transversals in hypergraphs

slide-79
SLIDE 79

Known results and conjectures, regularity

Very little is known regarding the regularity of a hypergraphs and the following has been conjectured. Conjecture For all k ≥ 2, if H is a k-uniform hypergraph, then the following holds. (a) If H is 3-regular then τ(H) ≤ 3n

2k (holds for k ≤ 4).

(b) If H is 4-regular then τ(H) ≤ 12n

7k (holds for k ≤ 4).

(c) If H is 5-regular then τ(H) ≤ 20n

11k (holds for k ≤ 3).

(d) If H is 6-regular then τ(H) ≤ 24n

13k (holds for k ≤ 2).

We now believe we have a proof of (a) for when k = 6. That is we can prove, Tuza-Vestergaard Conjecture (Now proved?): Every 3-regular 6-uniform hypergraph, H, satisfies τ(H) ≤ |V (H)|/4.

Anders Yeo Transversals in hypergraphs

slide-80
SLIDE 80

Known results and conjectures, regularity

Very little is known regarding the regularity of a hypergraphs and the following has been conjectured. Conjecture For all k ≥ 2, if H is a k-uniform hypergraph, then the following holds. (a) If H is 3-regular then τ(H) ≤ 3n

2k (holds for k ≤ 4).

(b) If H is 4-regular then τ(H) ≤ 12n

7k (holds for k ≤ 4).

(c) If H is 5-regular then τ(H) ≤ 20n

11k (holds for k ≤ 3).

(d) If H is 6-regular then τ(H) ≤ 24n

13k (holds for k ≤ 2).

We now believe we have a proof of (a) for when k = 6. That is we can prove, Tuza-Vestergaard Conjecture (Now proved?): Every 3-regular 6-uniform hypergraph, H, satisfies τ(H) ≤ |V (H)|/4.

Anders Yeo Transversals in hypergraphs

slide-81
SLIDE 81

Known results and conjectures, regularity

Very little is known regarding the regularity of a hypergraphs and the following has been conjectured. Conjecture For all k ≥ 2, if H is a k-uniform hypergraph, then the following holds. (a) If H is 3-regular then τ(H) ≤ 3n

2k (holds for k ≤ 4).

(b) If H is 4-regular then τ(H) ≤ 12n

7k (holds for k ≤ 4).

(c) If H is 5-regular then τ(H) ≤ 20n

11k (holds for k ≤ 3).

(d) If H is 6-regular then τ(H) ≤ 24n

13k (holds for k ≤ 2).

We now believe we have a proof of (a) for when k = 6. That is we can prove, Tuza-Vestergaard Conjecture (Now proved?): Every 3-regular 6-uniform hypergraph, H, satisfies τ(H) ≤ |V (H)|/4.

Anders Yeo Transversals in hypergraphs

slide-82
SLIDE 82

Known results and conjectures, regularity

Very little is known regarding the regularity of a hypergraphs and the following has been conjectured. Conjecture For all k ≥ 2, if H is a k-uniform hypergraph, then the following holds. (a) If H is 3-regular then τ(H) ≤ 3n

2k (holds for k ≤ 4).

(b) If H is 4-regular then τ(H) ≤ 12n

7k (holds for k ≤ 4).

(c) If H is 5-regular then τ(H) ≤ 20n

11k (holds for k ≤ 3).

(d) If H is 6-regular then τ(H) ≤ 24n

13k (holds for k ≤ 2).

We now believe we have a proof of (a) for when k = 6. That is we can prove, Tuza-Vestergaard Conjecture (Now proved?): Every 3-regular 6-uniform hypergraph, H, satisfies τ(H) ≤ |V (H)|/4.

Anders Yeo Transversals in hypergraphs

slide-83
SLIDE 83

Progress on the Tuza-Vestergaard Conjecture

The well-known Tuza-Vestergaard conjecture was published in 2002. Bounds of 0.27778n, 0.27037n and 0.268116n have been showed. There are infinitely many hypergraphs achieving equality in the bound n/4. (Take any 3-regular 3-uniform hyper- graph, H, with τ(H) = n/2 and du- plicate every vertex). Our proof of the Tuza-Vestergaard conjecture also implies a number of other results. For example, the following for k = 6. Conjecture: For all k ≥ 2, if H is a k-uniform hypergraph with maximum degree ∆(H) ≤ 3 then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

.

Anders Yeo Transversals in hypergraphs

slide-84
SLIDE 84

Progress on the Tuza-Vestergaard Conjecture

The well-known Tuza-Vestergaard conjecture was published in 2002. Bounds of 0.27778n, 0.27037n and 0.268116n have been showed. There are infinitely many hypergraphs achieving equality in the bound n/4. (Take any 3-regular 3-uniform hyper- graph, H, with τ(H) = n/2 and du- plicate every vertex). Our proof of the Tuza-Vestergaard conjecture also implies a number of other results. For example, the following for k = 6. Conjecture: For all k ≥ 2, if H is a k-uniform hypergraph with maximum degree ∆(H) ≤ 3 then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

.

Anders Yeo Transversals in hypergraphs

slide-85
SLIDE 85

Progress on the Tuza-Vestergaard Conjecture

The well-known Tuza-Vestergaard conjecture was published in 2002. Bounds of 0.27778n, 0.27037n and 0.268116n have been showed. There are infinitely many hypergraphs achieving equality in the bound n/4. (Take any 3-regular 3-uniform hyper- graph, H, with τ(H) = n/2 and du- plicate every vertex). Our proof of the Tuza-Vestergaard conjecture also implies a number of other results. For example, the following for k = 6. Conjecture: For all k ≥ 2, if H is a k-uniform hypergraph with maximum degree ∆(H) ≤ 3 then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

.

Anders Yeo Transversals in hypergraphs

slide-86
SLIDE 86

Progress on the Tuza-Vestergaard Conjecture

The well-known Tuza-Vestergaard conjecture was published in 2002. Bounds of 0.27778n, 0.27037n and 0.268116n have been showed. There are infinitely many hypergraphs achieving equality in the bound n/4. (Take any 3-regular 3-uniform hyper- graph, H, with τ(H) = n/2 and du- plicate every vertex). Our proof of the Tuza-Vestergaard conjecture also implies a number of other results. For example, the following for k = 6. Conjecture: For all k ≥ 2, if H is a k-uniform hypergraph with maximum degree ∆(H) ≤ 3 then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

.

Anders Yeo Transversals in hypergraphs

slide-87
SLIDE 87

Progress on the Tuza-Vestergaard Conjecture

The well-known Tuza-Vestergaard conjecture was published in 2002. Bounds of 0.27778n, 0.27037n and 0.268116n have been showed. There are infinitely many hypergraphs achieving equality in the bound n/4. (Take any 3-regular 3-uniform hyper- graph, H, with τ(H) = n/2 and du- plicate every vertex). Our proof of the Tuza-Vestergaard conjecture also implies a number of other results. For example, the following for k = 6. Conjecture: For all k ≥ 2, if H is a k-uniform hypergraph with maximum degree ∆(H) ≤ 3 then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

.

Anders Yeo Transversals in hypergraphs

slide-88
SLIDE 88

Progress on the Tuza-Vestergaard Conjecture

In order to prove the Tuza-Vestergaard Conjecture we prove a much stronger result. If H is a 6-uniform hypergraph on n vertices and m edges with ∆(H) ≤ 3, then 20τ(H) ≤ 2n + 6m + ρ, where ρ is a number that depends on the existence of certain hypergraphs, which will be described in more detail below. SB1 B1 B2 B3 B4

Anders Yeo Transversals in hypergraphs

slide-89
SLIDE 89

Progress on the Tuza-Vestergaard Conjecture

In order to prove the Tuza-Vestergaard Conjecture we prove a much stronger result. If H is a 6-uniform hypergraph on n vertices and m edges with ∆(H) ≤ 3, then 20τ(H) ≤ 2n + 6m + ρ, where ρ is a number that depends on the existence of certain hypergraphs, which will be described in more detail below. SB1 B1 B2 B3 B4

Anders Yeo Transversals in hypergraphs

slide-90
SLIDE 90

Progress on the Tuza-Vestergaard Conjecture

We recusively define an infinite family B of bad hypergraphs as follows. Let B = {B1, B2, B3, B4} and recursively do the following....

In B →

No minimum transversal contains a red and green vertex. Any hypergraph in B is 2 short in the formula 20τ ≤ 2n + 6m. What about SB1?

Anders Yeo Transversals in hypergraphs

slide-91
SLIDE 91

Progress on the Tuza-Vestergaard Conjecture

We recusively define an infinite family B of bad hypergraphs as follows. Let B = {B1, B2, B3, B4} and recursively do the following....

In B →

No minimum transversal contains a red and green vertex. Any hypergraph in B is 2 short in the formula 20τ ≤ 2n + 6m. What about SB1?

Anders Yeo Transversals in hypergraphs

slide-92
SLIDE 92

Progress on the Tuza-Vestergaard Conjecture

We recusively define an infinite family B of bad hypergraphs as follows. Let B = {B1, B2, B3, B4} and recursively do the following....

In B →

No minimum transversal contains a red and green vertex. Any hypergraph in B is 2 short in the formula 20τ ≤ 2n + 6m. What about SB1?

Anders Yeo Transversals in hypergraphs

slide-93
SLIDE 93

Progress on the Tuza-Vestergaard Conjecture

We recusively define an infinite family B of bad hypergraphs as follows. Let B = {B1, B2, B3, B4} and recursively do the following....

In B →

No minimum transversal contains a red and green vertex. Any hypergraph in B is 2 short in the formula 20τ ≤ 2n + 6m. What about SB1?

Anders Yeo Transversals in hypergraphs

slide-94
SLIDE 94

Progress on the Tuza-Vestergaard Conjecture

We recusively define an infinite family B of bad hypergraphs as follows. Let B = {B1, B2, B3, B4} and recursively do the following....

In B →

No minimum transversal contains a red and green vertex. Any hypergraph in B is 2 short in the formula 20τ ≤ 2n + 6m. What about SB1?

Anders Yeo Transversals in hypergraphs

slide-95
SLIDE 95

Progress on the Tuza-Vestergaard Conjecture

We recusively define an infinite family B of bad hypergraphs as follows. Let B = {B1, B2, B3, B4} and recursively do the following....

In B →

No minimum transversal contains a red and green vertex. Any hypergraph in B is 2 short in the formula 20τ ≤ 2n + 6m. What about SB1?

Anders Yeo Transversals in hypergraphs

slide-96
SLIDE 96

Progress on the Tuza-Vestergaard Conjecture

We recusively define an infinite family B of bad hypergraphs as follows. Let B = {B1, B2, B3, B4} and recursively do the following....

In B →

No minimum transversal contains a red and green vertex. Any hypergraph in B is 2 short in the formula 20τ ≤ 2n + 6m. What about SB1?

Anders Yeo Transversals in hypergraphs

slide-97
SLIDE 97

Progress on the Tuza-Vestergaard Conjecture

We recusively define an infinite family B of bad hypergraphs as follows. Let B = {B1, B2, B3, B4} and recursively do the following....

In B →

No minimum transversal contains a red and green vertex. Any hypergraph in B is 2 short in the formula 20τ ≤ 2n + 6m. What about SB1?

Anders Yeo Transversals in hypergraphs

slide-98
SLIDE 98

Progress on the Tuza-Vestergaard Conjecture

SB1 is 4 short in the formula 20τ ≤ 2n + 6m. If C is a SB1 subgraph in H, then define f (C) as follows: # degree 2’s in C 1 2 3 4 5 6 7 8 9 f (C) 1 1 1 2 2 3 4 If R ∈ B then f (R) =    2 R is a component in H 1 R is intersected by 1 edge R is intersected by ≥ 2 edges . Let ρ = max

Q∈Q f (Q) over all families of vertex-disjoint

subgraphs Q in H (F(Q) = 0 if Q ∈ B ∪ SB1).

Anders Yeo Transversals in hypergraphs

slide-99
SLIDE 99

Progress on the Tuza-Vestergaard Conjecture

SB1 is 4 short in the formula 20τ ≤ 2n + 6m. If C is a SB1 subgraph in H, then define f (C) as follows: # degree 2’s in C 1 2 3 4 5 6 7 8 9 f (C) 1 1 1 2 2 3 4 If R ∈ B then f (R) =    2 R is a component in H 1 R is intersected by 1 edge R is intersected by ≥ 2 edges . Let ρ = max

Q∈Q f (Q) over all families of vertex-disjoint

subgraphs Q in H (F(Q) = 0 if Q ∈ B ∪ SB1).

Anders Yeo Transversals in hypergraphs

slide-100
SLIDE 100

Progress on the Tuza-Vestergaard Conjecture

SB1 is 4 short in the formula 20τ ≤ 2n + 6m. If C is a SB1 subgraph in H, then define f (C) as follows: # degree 2’s in C 1 2 3 4 5 6 7 8 9 f (C) 1 1 1 2 2 3 4 If R ∈ B then f (R) =    2 R is a component in H 1 R is intersected by 1 edge R is intersected by ≥ 2 edges . Let ρ = max

Q∈Q f (Q) over all families of vertex-disjoint

subgraphs Q in H (F(Q) = 0 if Q ∈ B ∪ SB1).

Anders Yeo Transversals in hypergraphs

slide-101
SLIDE 101

Progress on the Tuza-Vestergaard Conjecture

SB1 is 4 short in the formula 20τ ≤ 2n + 6m. If C is a SB1 subgraph in H, then define f (C) as follows: # degree 2’s in C 1 2 3 4 5 6 7 8 9 f (C) 1 1 1 2 2 3 4 If R ∈ B then f (R) =    2 R is a component in H 1 R is intersected by 1 edge R is intersected by ≥ 2 edges . Let ρ = max

Q∈Q f (Q) over all families of vertex-disjoint

subgraphs Q in H (F(Q) = 0 if Q ∈ B ∪ SB1).

Anders Yeo Transversals in hypergraphs

slide-102
SLIDE 102

Progress on the Tuza-Vestergaard Conjecture

Using the above complicated definition of ρ we can prove that 20τ(H) ≤ 2n + 6m + ρ. This implies the Tuza-Vestergaard Conjecture, as follows. When H is 3-regular then 3n =

v∈V d(v) = 6m.

Also, ρ = 0 as f (Q) = 0 if all vertices of Q have degree three. 20τ(H) ≤ 2n + 6m + ρ = 2n + 6m = 2n + 3n = 5n So, τ(H) ≤ n/4. QED.

Anders Yeo Transversals in hypergraphs

slide-103
SLIDE 103

Progress on the Tuza-Vestergaard Conjecture

Using the above complicated definition of ρ we can prove that 20τ(H) ≤ 2n + 6m + ρ. This implies the Tuza-Vestergaard Conjecture, as follows. When H is 3-regular then 3n =

v∈V d(v) = 6m.

Also, ρ = 0 as f (Q) = 0 if all vertices of Q have degree three. 20τ(H) ≤ 2n + 6m + ρ = 2n + 6m = 2n + 3n = 5n So, τ(H) ≤ n/4. QED.

Anders Yeo Transversals in hypergraphs

slide-104
SLIDE 104

Progress on the Tuza-Vestergaard Conjecture

Using the above complicated definition of ρ we can prove that 20τ(H) ≤ 2n + 6m + ρ. This implies the Tuza-Vestergaard Conjecture, as follows. When H is 3-regular then 3n =

v∈V d(v) = 6m.

Also, ρ = 0 as f (Q) = 0 if all vertices of Q have degree three. 20τ(H) ≤ 2n + 6m + ρ = 2n + 6m = 2n + 3n = 5n So, τ(H) ≤ n/4. QED.

Anders Yeo Transversals in hypergraphs

slide-105
SLIDE 105

Progress on the Tuza-Vestergaard Conjecture

Using the above complicated definition of ρ we can prove that 20τ(H) ≤ 2n + 6m + ρ. This implies the Tuza-Vestergaard Conjecture, as follows. When H is 3-regular then 3n =

v∈V d(v) = 6m.

Also, ρ = 0 as f (Q) = 0 if all vertices of Q have degree three. 20τ(H) ≤ 2n + 6m + ρ = 2n + 6m = 2n + 3n = 5n So, τ(H) ≤ n/4. QED.

Anders Yeo Transversals in hypergraphs

slide-106
SLIDE 106

Progress on the Tuza-Vestergaard Conjecture

Using the above complicated definition of ρ we can prove that 20τ(H) ≤ 2n + 6m + ρ. This implies the Tuza-Vestergaard Conjecture, as follows. When H is 3-regular then 3n =

v∈V d(v) = 6m.

Also, ρ = 0 as f (Q) = 0 if all vertices of Q have degree three. 20τ(H) ≤ 2n + 6m + ρ = 2n + 6m = 2n + 3n = 5n So, τ(H) ≤ n/4. QED.

Anders Yeo Transversals in hypergraphs

slide-107
SLIDE 107

Progress on a conjecture for linear 4-uniform hypergraphs

Theorem: If H is 2-uniform then τ(H) ≤ n+m

3 .

Theorem: If H is 3-uniform then τ(H) ≤ n+m

4 .

The unique simple k-uniform hypergraph on k + 1 vertices with k + 1 edges has transversal number 2. So the above theorems are tight (also due to the single edge). It is not true that τ(H) ≤ n+m

5

for 4-uniform hypergraphs H. What happens if we look at linear hypergraphs?

Anders Yeo Transversals in hypergraphs

slide-108
SLIDE 108

Progress on a conjecture for linear 4-uniform hypergraphs

Theorem: If H is 2-uniform then τ(H) ≤ n+m

3 .

Theorem: If H is 3-uniform then τ(H) ≤ n+m

4 .

The unique simple k-uniform hypergraph on k + 1 vertices with k + 1 edges has transversal number 2. So the above theorems are tight (also due to the single edge). It is not true that τ(H) ≤ n+m

5

for 4-uniform hypergraphs H. What happens if we look at linear hypergraphs?

Anders Yeo Transversals in hypergraphs

slide-109
SLIDE 109

Progress on a conjecture for linear 4-uniform hypergraphs

Theorem: If H is 2-uniform then τ(H) ≤ n+m

3 .

Theorem: If H is 3-uniform then τ(H) ≤ n+m

4 .

The unique simple k-uniform hypergraph on k + 1 vertices with k + 1 edges has transversal number 2. So the above theorems are tight (also due to the single edge). It is not true that τ(H) ≤ n+m

5

for 4-uniform hypergraphs H. What happens if we look at linear hypergraphs?

Anders Yeo Transversals in hypergraphs

slide-110
SLIDE 110

Progress on a conjecture for linear 4-uniform hypergraphs

Theorem: If H is 2-uniform then τ(H) ≤ n+m

3 .

Theorem: If H is 3-uniform then τ(H) ≤ n+m

4 .

The unique simple k-uniform hypergraph on k + 1 vertices with k + 1 edges has transversal number 2. So the above theorems are tight (also due to the single edge). It is not true that τ(H) ≤ n+m

5

for 4-uniform hypergraphs H. What happens if we look at linear hypergraphs?

Anders Yeo Transversals in hypergraphs

slide-111
SLIDE 111

Progress on a conjecture for linear 4-uniform hypergraphs

Theorem: If H is 2-uniform then τ(H) ≤ n+m

3 .

Theorem: If H is 3-uniform then τ(H) ≤ n+m

4 .

The unique simple k-uniform hypergraph on k + 1 vertices with k + 1 edges has transversal number 2. So the above theorems are tight (also due to the single edge). It is not true that τ(H) ≤ n+m

5

for 4-uniform hypergraphs H. What happens if we look at linear hypergraphs?

Anders Yeo Transversals in hypergraphs

slide-112
SLIDE 112

Progress on a conjecture for linear 4-uniform hypergraphs

Theorem: If H is 2-uniform then τ(H) ≤ n+m

3 .

Theorem: If H is 3-uniform then τ(H) ≤ n+m

4 .

The unique simple k-uniform hypergraph on k + 1 vertices with k + 1 edges has transversal number 2. So the above theorems are tight (also due to the single edge). It is not true that τ(H) ≤ n+m

5

for 4-uniform hypergraphs H. What happens if we look at linear hypergraphs?

Anders Yeo Transversals in hypergraphs

slide-113
SLIDE 113

Progress on a conjecture for linear 4-uniform hypergraphs

Surprisingly little is know about best possible bounds for linear k-uniform hypergraphs. Question: Is it true that τ(H) ≤ n+m

k+1 for all linear k-uniform

hypergraphs? The answer is yes for k = 2 and k = 3 (even if we do not require linearity). What about k = 4?

Anders Yeo Transversals in hypergraphs

slide-114
SLIDE 114

Progress on a conjecture for linear 4-uniform hypergraphs

Surprisingly little is know about best possible bounds for linear k-uniform hypergraphs. Question: Is it true that τ(H) ≤ n+m

k+1 for all linear k-uniform

hypergraphs? The answer is yes for k = 2 and k = 3 (even if we do not require linearity). What about k = 4?

Anders Yeo Transversals in hypergraphs

slide-115
SLIDE 115

Progress on a conjecture for linear 4-uniform hypergraphs

Surprisingly little is know about best possible bounds for linear k-uniform hypergraphs. Question: Is it true that τ(H) ≤ n+m

k+1 for all linear k-uniform

hypergraphs? The answer is yes for k = 2 and k = 3 (even if we do not require linearity). What about k = 4?

Anders Yeo Transversals in hypergraphs

slide-116
SLIDE 116

Progress on a conjecture for linear 4-uniform hypergraphs

Surprisingly little is know about best possible bounds for linear k-uniform hypergraphs. Question: Is it true that τ(H) ≤ n+m

k+1 for all linear k-uniform

hypergraphs? The answer is yes for k = 2 and k = 3 (even if we do not require linearity). What about k = 4?

Anders Yeo Transversals in hypergraphs

slide-117
SLIDE 117

Progress on a conjecture for linear 4-uniform hypergraphs

Conjecture: τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs.

We believe we can prove the above conjecture (we will write up the result next week!). This result would also imply that the following conjecture holds for k = 4 (it is known to hold for k = 2, 3). Conjecture: For all k ≥ 2, if H is a k-uniform linear hypergraph then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

. Does the above conjecture look familiar? Substitute ”linear” with ”∆ ≤ 3”...

Anders Yeo Transversals in hypergraphs

slide-118
SLIDE 118

Progress on a conjecture for linear 4-uniform hypergraphs

Conjecture: τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs.

We believe we can prove the above conjecture (we will write up the result next week!). This result would also imply that the following conjecture holds for k = 4 (it is known to hold for k = 2, 3). Conjecture: For all k ≥ 2, if H is a k-uniform linear hypergraph then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

. Does the above conjecture look familiar? Substitute ”linear” with ”∆ ≤ 3”...

Anders Yeo Transversals in hypergraphs

slide-119
SLIDE 119

Progress on a conjecture for linear 4-uniform hypergraphs

Conjecture: τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs.

We believe we can prove the above conjecture (we will write up the result next week!). This result would also imply that the following conjecture holds for k = 4 (it is known to hold for k = 2, 3). Conjecture: For all k ≥ 2, if H is a k-uniform linear hypergraph then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

. Does the above conjecture look familiar? Substitute ”linear” with ”∆ ≤ 3”...

Anders Yeo Transversals in hypergraphs

slide-120
SLIDE 120

Progress on a conjecture for linear 4-uniform hypergraphs

Conjecture: τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs.

We believe we can prove the above conjecture (we will write up the result next week!). This result would also imply that the following conjecture holds for k = 4 (it is known to hold for k = 2, 3). Conjecture: For all k ≥ 2, if H is a k-uniform linear hypergraph then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

. Does the above conjecture look familiar? Substitute ”linear” with ”∆ ≤ 3”...

Anders Yeo Transversals in hypergraphs

slide-121
SLIDE 121

Progress on a conjecture for linear 4-uniform hypergraphs

Conjecture: τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs.

We believe we can prove the above conjecture (we will write up the result next week!). This result would also imply that the following conjecture holds for k = 4 (it is known to hold for k = 2, 3). Conjecture: For all k ≥ 2, if H is a k-uniform linear hypergraph then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

. Does the above conjecture look familiar? Substitute ”linear” with ”∆ ≤ 3”...

Anders Yeo Transversals in hypergraphs

slide-122
SLIDE 122

Progress on a conjecture for linear 4-uniform hypergraphs

Conjecture: τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs.

We believe we can prove the above conjecture (we will write up the result next week!). This result would also imply that the following conjecture holds for k = 4 (it is known to hold for k = 2, 3). Conjecture: For all k ≥ 2, if H is a k-uniform linear hypergraph then τ(H) ≤ |V (H)|

k

+ |E(H)|

6

. Does the above conjecture look familiar? Substitute ”linear” with ”∆ ≤ 3”...

Anders Yeo Transversals in hypergraphs

slide-123
SLIDE 123

Progress on a conjecture for linear 4-uniform hypergraphs

How do we prove that τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs?

Again we need to prove a much stronger result. And we need to introduce a new concept called deficiency, def (H),

  • f a hypergraph H.

Theorem (stronger result): If H is a 4-uniform linear hypergraph, with ∆(H) ≤ 3, then 45τ(H) ≤ 6|V (H)| + 13|E(H)| + def (H).

Anders Yeo Transversals in hypergraphs

slide-124
SLIDE 124

Progress on a conjecture for linear 4-uniform hypergraphs

How do we prove that τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs?

Again we need to prove a much stronger result. And we need to introduce a new concept called deficiency, def (H),

  • f a hypergraph H.

Theorem (stronger result): If H is a 4-uniform linear hypergraph, with ∆(H) ≤ 3, then 45τ(H) ≤ 6|V (H)| + 13|E(H)| + def (H).

Anders Yeo Transversals in hypergraphs

slide-125
SLIDE 125

Progress on a conjecture for linear 4-uniform hypergraphs

How do we prove that τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs?

Again we need to prove a much stronger result. And we need to introduce a new concept called deficiency, def (H),

  • f a hypergraph H.

Theorem (stronger result): If H is a 4-uniform linear hypergraph, with ∆(H) ≤ 3, then 45τ(H) ≤ 6|V (H)| + 13|E(H)| + def (H).

Anders Yeo Transversals in hypergraphs

slide-126
SLIDE 126

Progress on a conjecture for linear 4-uniform hypergraphs

How do we prove that τ(H) ≤ n+m

k+1 for all 4-uniform hypergraphs?

Again we need to prove a much stronger result. And we need to introduce a new concept called deficiency, def (H),

  • f a hypergraph H.

Theorem (stronger result): If H is a 4-uniform linear hypergraph, with ∆(H) ≤ 3, then 45τ(H) ≤ 6|V (H)| + 13|E(H)| + def (H).

Anders Yeo Transversals in hypergraphs

slide-127
SLIDE 127

Progress on a conjecture for linear 4-uniform hypergraphs

The following 16 ”bad” hypergraphs add to the deficiency.

Anders Yeo Transversals in hypergraphs

slide-128
SLIDE 128

Progress on a conjecture for linear 4-uniform hypergraphs

Let g(R) be the amount that the graph G is short in 45τ(R) ≤ 6|V (R)| + 13|E(R)|. Let R be a vertex disjoint collection of bad subhypergraphs and let e(R) be the number of edges intersecting these in H. The deficiency of H is the maximum possible value of (

R∈R g(R)) − 13e(R) over all R.

def (H) = 4 × 8 − 2 × 13 = 6.

Anders Yeo Transversals in hypergraphs

slide-129
SLIDE 129

Progress on a conjecture for linear 4-uniform hypergraphs

Let g(R) be the amount that the graph G is short in 45τ(R) ≤ 6|V (R)| + 13|E(R)|. Let R be a vertex disjoint collection of bad subhypergraphs and let e(R) be the number of edges intersecting these in H. The deficiency of H is the maximum possible value of (

R∈R g(R)) − 13e(R) over all R.

def (H) = 4 × 8 − 2 × 13 = 6.

Anders Yeo Transversals in hypergraphs

slide-130
SLIDE 130

Progress on a conjecture for linear 4-uniform hypergraphs

Let g(R) be the amount that the graph G is short in 45τ(R) ≤ 6|V (R)| + 13|E(R)|. Let R be a vertex disjoint collection of bad subhypergraphs and let e(R) be the number of edges intersecting these in H. The deficiency of H is the maximum possible value of (

R∈R g(R)) − 13e(R) over all R.

def (H) = 4 × 8 − 2 × 13 = 6.

Anders Yeo Transversals in hypergraphs

slide-131
SLIDE 131

Progress on a conjecture for linear 4-uniform hypergraphs

Let g(R) be the amount that the graph G is short in 45τ(R) ≤ 6|V (R)| + 13|E(R)|. Let R be a vertex disjoint collection of bad subhypergraphs and let e(R) be the number of edges intersecting these in H. The deficiency of H is the maximum possible value of (

R∈R g(R)) − 13e(R) over all R.

def (H) = 4 × 8 − 2 × 13 = 6.

Anders Yeo Transversals in hypergraphs

slide-132
SLIDE 132

Progress on a conjecture for linear 4-uniform hypergraphs

Let g(R) be the amount that the graph G is short in 45τ(R) ≤ 6|V (R)| + 13|E(R)|. Let R be a vertex disjoint collection of bad subhypergraphs and let e(R) be the number of edges intersecting these in H. The deficiency of H is the maximum possible value of (

R∈R g(R)) − 13e(R) over all R.

def (H) = 4 × 8 − 2 × 13 = 6.

Anders Yeo Transversals in hypergraphs

slide-133
SLIDE 133

Progress on a conjecture for linear 4-uniform hypergraphs

Let g(R) be the amount that the graph G is short in 45τ(R) ≤ 6|V (R)| + 13|E(R)|. Let R be a vertex disjoint collection of bad subhypergraphs and let e(R) be the number of edges intersecting these in H. The deficiency of H is the maximum possible value of (

R∈R g(R)) − 13e(R) over all R.

def (H) = 4 × 8 − 2 × 13 = 6.

Anders Yeo Transversals in hypergraphs

slide-134
SLIDE 134

Consequences

Apart from proving the two conjectures stated, the following also follows. Theorem? Let H be a 3-regular 4-uniform hypergraph on n

  • vertices. If H is linear, then τ(H) ≤ 7n/20.

Theorem? If G is a graph of order n, with no 4-cycles and with δ(G) ≥ 4, then γt(G) ≤ 2n/5.

Anders Yeo Transversals in hypergraphs

slide-135
SLIDE 135

Consequences

Apart from proving the two conjectures stated, the following also follows. Theorem? Let H be a 3-regular 4-uniform hypergraph on n

  • vertices. If H is linear, then τ(H) ≤ 7n/20.

Theorem? If G is a graph of order n, with no 4-cycles and with δ(G) ≥ 4, then γt(G) ≤ 2n/5.

Anders Yeo Transversals in hypergraphs

slide-136
SLIDE 136

Consequences

Apart from proving the two conjectures stated, the following also follows. Theorem? Let H be a 3-regular 4-uniform hypergraph on n

  • vertices. If H is linear, then τ(H) ≤ 7n/20.

Theorem? If G is a graph of order n, with no 4-cycles and with δ(G) ≥ 4, then γt(G) ≤ 2n/5.

Anders Yeo Transversals in hypergraphs

slide-137
SLIDE 137

The End.

The End! Thank you. Any Questions?

Anders Yeo Transversals in hypergraphs

slide-138
SLIDE 138

The End.

The End! Thank you. Any Questions?

Anders Yeo Transversals in hypergraphs