department of mathematics nebraska wesleyan university · 2016. 9. 8. · nebraska wesleyan...
TRANSCRIPT
![Page 1: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/1.jpg)
Spanning Subgraph Isomorphism
Drew Meier, Dr. Austin Mohr, Thomas Schuler
Department of MathematicsNebraska Wesleyan University
August 4, 2016
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 2: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/2.jpg)
Outline
1 What is Subgraph Isomorphism
2 Cut Vertex Refinement
3 Perfect Matching Refinements
4 Skeleton
5 Data
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 3: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/3.jpg)
Outline
1 What is Subgraph Isomorphism
2 Cut Vertex Refinement
3 Perfect Matching Refinements
4 Skeleton
5 Data
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 4: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/4.jpg)
Outline
1 What is Subgraph Isomorphism
2 Cut Vertex Refinement
3 Perfect Matching Refinements
4 Skeleton
5 Data
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 5: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/5.jpg)
Outline
1 What is Subgraph Isomorphism
2 Cut Vertex Refinement
3 Perfect Matching Refinements
4 Skeleton
5 Data
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 6: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/6.jpg)
Outline
1 What is Subgraph Isomorphism
2 Cut Vertex Refinement
3 Perfect Matching Refinements
4 Skeleton
5 Data
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 7: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/7.jpg)
Subgraph Isomorphism
Given a pattern graph and a host graph, SubgraphIsomorphism is the problem of determining whether the hostgraph contains a subgraph that is isomorphic to the patterngraph
We restrict ourselves to patterns that span the host
Ullman (1976)
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 8: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/8.jpg)
Subgraph Isomorphism
Given a pattern graph and a host graph, SubgraphIsomorphism is the problem of determining whether the hostgraph contains a subgraph that is isomorphic to the patterngraph
We restrict ourselves to patterns that span the host
Ullman (1976)
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 9: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/9.jpg)
Subgraph Isomorphism
Given a pattern graph and a host graph, SubgraphIsomorphism is the problem of determining whether the hostgraph contains a subgraph that is isomorphic to the patterngraph
We restrict ourselves to patterns that span the host
Ullman (1976)
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 10: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/10.jpg)
Candidate Matrix
We use a matrix, C , to hold information about candidate vertices,whose rows correspond to vertices of the pattern, P, and columns
correspond the vertices of the host, H.
C =
0 1 1 01 0 1 01 1 1 11 1 1 0
Ci ,j = 1 indicates vertex i of P is a candidate for vertex j of H.
Reductions eliminate portions of the search space by eliminatingcandidates
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 11: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/11.jpg)
Candidate Matrix
We use a matrix, C , to hold information about candidate vertices,whose rows correspond to vertices of the pattern, P, and columns
correspond the vertices of the host, H.
C =
0 1 1 01 0 1 01 1 1 11 1 1 0
Ci ,j = 1 indicates vertex i of P is a candidate for vertex j of H.Reductions eliminate portions of the search space by eliminatingcandidates
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 12: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/12.jpg)
Candidate Matrix
We use a matrix, C , to hold information about candidate vertices,whose rows correspond to vertices of the pattern, P, and columns
correspond the vertices of the host, H.
C =
0 1 1 01 0 1 01 1 1 11 1 1 0
Ci ,j = 1 indicates vertex i of P is a candidate for vertex j of H.
Reductions eliminate portions of the search space by eliminatingcandidates
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 13: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/13.jpg)
Candidate Matrix
We use a matrix, C , to hold information about candidate vertices,whose rows correspond to vertices of the pattern, P, and columns
correspond the vertices of the host, H.
C =
0 1 1 01 0 1 01 1 1 11 1 1 0
Ci ,j = 1 indicates vertex i of P is a candidate for vertex j of H.
Reductions eliminate portions of the search space by eliminatingcandidates
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 14: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/14.jpg)
Trivial Reductions
Reduction by Degree
Reduction by Adjacency
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 15: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/15.jpg)
Trivial Reductions
Reduction by Degree
Reduction by Adjacency
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 16: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/16.jpg)
Cut Vertex Refinement
A cut vertex, u, is a cut vertex on a graph, G , if the graph G − uhas more components than G .
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 17: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/17.jpg)
Cut Vertex Refinement
A cut vertex, u, is a cut vertex on a graph, G , if the graph G − uhas more components than G .
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 18: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/18.jpg)
Cut Vertex Refinement
Theorem
A cut vertex, u, of the host, H, must be mapped to a cut vertex,v , of the pattern P
Proof.
Suppose u is a cut vertex of the host graph, H. Now suppose weassign a non-cut vertex v of the pattern, P, to u. Consider onlythe connected components of H and P containing u and v , Hu andPv . Since u is a cut vertex, Hu − u has at least 2 components.Since v is not a cut vertex, Pv − v is connected. Clearly, nomapping of the connected pattern, Pv − v , into the disconnectedhost Hu − u can exist. Thus, a cut vertex of the host graph mustmap to a cut vertex of the pattern graph.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 19: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/19.jpg)
Cut Vertex Refinement
Theorem
A cut vertex, u, of the host, H, must be mapped to a cut vertex,v , of the pattern P
Proof.
Suppose u is a cut vertex of the host graph, H. Now suppose weassign a non-cut vertex v of the pattern, P, to u. Consider onlythe connected components of H and P containing u and v , Hu andPv . Since u is a cut vertex, Hu − u has at least 2 components.Since v is not a cut vertex, Pv − v is connected. Clearly, nomapping of the connected pattern, Pv − v , into the disconnectedhost Hu − u can exist. Thus, a cut vertex of the host graph mustmap to a cut vertex of the pattern graph.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 20: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/20.jpg)
Cut Vertex Refinement
Theorem
A cut vertex, u, of the host, H, must be mapped to a cut vertex,v , of the pattern P
Proof.
Suppose u is a cut vertex of the host graph, H. Now suppose weassign a non-cut vertex v of the pattern, P, to u. Consider onlythe connected components of H and P containing u and v , Hu andPv . Since u is a cut vertex, Hu − u has at least 2 components.Since v is not a cut vertex, Pv − v is connected. Clearly, nomapping of the connected pattern, Pv − v , into the disconnectedhost Hu − u can exist. Thus, a cut vertex of the host graph mustmap to a cut vertex of the pattern graph.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 21: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/21.jpg)
Cut Vertex Refinement
Theorem
A cut vertex, u, of the host, H, must be mapped to a cut vertex,v , of the pattern P
Proof.
Suppose u is a cut vertex of the host graph, H. Now suppose weassign a non-cut vertex v of the pattern, P, to u. Consider onlythe connected components of H and P containing u and v , Hu andPv . Since u is a cut vertex, Hu − u has at least 2 components.Since v is not a cut vertex, Pv − v is connected. Clearly, nomapping of the connected pattern, Pv − v , into the disconnectedhost Hu − u can exist. Thus, a cut vertex of the host graph mustmap to a cut vertex of the pattern graph.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 22: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/22.jpg)
Cut Vertex Refinement
Theorem
A cut vertex, u, of the host, H, must be mapped to a cut vertex,v , of the pattern P
Proof.
Suppose u is a cut vertex of the host graph, H. Now suppose weassign a non-cut vertex v of the pattern, P, to u. Consider onlythe connected components of H and P containing u and v , Hu andPv . Since u is a cut vertex, Hu − u has at least 2 components.Since v is not a cut vertex, Pv − v is connected. Clearly, nomapping of the connected pattern, Pv − v , into the disconnectedhost Hu − u can exist. Thus, a cut vertex of the host graph mustmap to a cut vertex of the pattern graph.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 23: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/23.jpg)
Cut Vertex Refinement
Theorem
A cut vertex, u, of the host, H, must be mapped to a cut vertex,v , of the pattern P
Proof.
Suppose u is a cut vertex of the host graph, H. Now suppose weassign a non-cut vertex v of the pattern, P, to u. Consider onlythe connected components of H and P containing u and v , Hu andPv . Since u is a cut vertex, Hu − u has at least 2 components.Since v is not a cut vertex, Pv − v is connected. Clearly, nomapping of the connected pattern, Pv − v , into the disconnectedhost Hu − u can exist. Thus, a cut vertex of the host graph mustmap to a cut vertex of the pattern graph.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 24: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/24.jpg)
Cut Vertex Refinement
Theorem
A cut vertex, u, of the host, H, must be mapped to a cut vertex,v , of the pattern P
Proof.
Suppose u is a cut vertex of the host graph, H. Now suppose weassign a non-cut vertex v of the pattern, P, to u. Consider onlythe connected components of H and P containing u and v , Hu andPv . Since u is a cut vertex, Hu − u has at least 2 components.Since v is not a cut vertex, Pv − v is connected. Clearly, nomapping of the connected pattern, Pv − v , into the disconnectedhost Hu − u can exist. Thus, a cut vertex of the host graph mustmap to a cut vertex of the pattern graph.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 25: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/25.jpg)
Component Partition Refinement
Definition
Let a = (a1, a2, ..., an) and b = (b1, b2, ..., bk) be two partitions ofa common integer. We say the partition a refines the partition bprovided there is a partition of the indexing set {1, ..., n} intoblocks J1, J2, ..., J` such that bi =
∑j∈Ji aj for all 1 ≤ i ≤ k
Example:a = [1, 1, 3, 4] b = [2, 7]
Let J1 = {1, 2} and J2 = {3, 4}. Then,
a1 + a2 = 2 = b1 and a3 + a4 = 7 = b2
Thus, a refines b.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 26: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/26.jpg)
Component Partition Refinement
Definition
Let a = (a1, a2, ..., an) and b = (b1, b2, ..., bk) be two partitions ofa common integer. We say the partition a refines the partition bprovided there is a partition of the indexing set {1, ..., n} intoblocks J1, J2, ..., J` such that bi =
∑j∈Ji aj for all 1 ≤ i ≤ k
Example:
a = [1, 1, 3, 4] b = [2, 7]
Let J1 = {1, 2} and J2 = {3, 4}. Then,
a1 + a2 = 2 = b1 and a3 + a4 = 7 = b2
Thus, a refines b.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 27: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/27.jpg)
Component Partition Refinement
Definition
Let a = (a1, a2, ..., an) and b = (b1, b2, ..., bk) be two partitions ofa common integer. We say the partition a refines the partition bprovided there is a partition of the indexing set {1, ..., n} intoblocks J1, J2, ..., J` such that bi =
∑j∈Ji aj for all 1 ≤ i ≤ k
Example:a = [1, 1, 3, 4] b = [2, 7]
Let J1 = {1, 2} and J2 = {3, 4}. Then,
a1 + a2 = 2 = b1 and a3 + a4 = 7 = b2
Thus, a refines b.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 28: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/28.jpg)
Component Partition Refinement
Definition
Let a = (a1, a2, ..., an) and b = (b1, b2, ..., bk) be two partitions ofa common integer. We say the partition a refines the partition bprovided there is a partition of the indexing set {1, ..., n} intoblocks J1, J2, ..., J` such that bi =
∑j∈Ji aj for all 1 ≤ i ≤ k
Example:a = [1, 1, 3, 4] b = [2, 7]
Let J1 = {1, 2} and J2 = {3, 4}. Then,
a1 + a2 = 2 = b1 and a3 + a4 = 7 = b2
Thus, a refines b.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 29: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/29.jpg)
Component Partition Refinement
Definition
Let a = (a1, a2, ..., an) and b = (b1, b2, ..., bk) be two partitions ofa common integer. We say the partition a refines the partition bprovided there is a partition of the indexing set {1, ..., n} intoblocks J1, J2, ..., J` such that bi =
∑j∈Ji aj for all 1 ≤ i ≤ k
Example:a = [1, 1, 3, 4] b = [2, 7]
Let J1 = {1, 2} and J2 = {3, 4}. Then,
a1 + a2 = 2 = b1 and a3 + a4 = 7 = b2
Thus, a refines b.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 30: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/30.jpg)
Component Partition Refinement
Definition
Given a graph G of order n and a vertex u, let G1, ...,Gk denotethe connected components of G − u labeled so that|G1| ≥ |G2| ≥ · · · ≥ |Gk |. The component partition C (v) is thepartition of n − 1 given by (|G1|, |G2|, ..., |Gk |).
Theorem
Let P and H be graphs of equal order and f : V (P)→ V (H) be asubgraph isomorphism. For any vertex u ∈ V (H), the componentpartition C (f −1(v)) refines the component partition C (u).
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 31: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/31.jpg)
Component Partition Refinement
Definition
Given a graph G of order n and a vertex u, let G1, ...,Gk denotethe connected components of G − u labeled so that|G1| ≥ |G2| ≥ · · · ≥ |Gk |. The component partition C (v) is thepartition of n − 1 given by (|G1|, |G2|, ..., |Gk |).
Theorem
Let P and H be graphs of equal order and f : V (P)→ V (H) be asubgraph isomorphism. For any vertex u ∈ V (H), the componentpartition C (f −1(v)) refines the component partition C (u).
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 32: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/32.jpg)
Component Partition Refinement (Example)
Degree Reduction
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 33: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/33.jpg)
Component Partition Refinement (Example)
Degree Reduction
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 34: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/34.jpg)
Component Partition Refinement (Example)
Degree Reduction Cut Vertex Reduction
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 35: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/35.jpg)
Matching Reductions
Definition
A matching on a graph, G , is a subset of edges in G such that noedges share common vertices.
Definition
A perfect matching on a graph, G , is a matching on G such thatall vertices of G are visited by an edge in the set.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 36: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/36.jpg)
Matching Reductions
Definition
A matching on a graph, G , is a subset of edges in G such that noedges share common vertices.
Definition
A perfect matching on a graph, G , is a matching on G such thatall vertices of G are visited by an edge in the set.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 37: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/37.jpg)
Matching Reductions
Upon assignment of vertex v of the pattern to vertex u of thehost, we set up a bipartite graph as follows:
One partition is the set of neighbors of u
The other neighbors of v
Vertices i , j in the bipartition have an edge if vertex i is acandidate for vertex j .
We call this the neighborhood bipartition, Ku,v .
Theorem
For vertex u to be a candidate for vertex v upon assignment of vto u, a matching on Ku,v the size of the partition v must exist.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 38: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/38.jpg)
Matching Reductions
Upon assignment of vertex v of the pattern to vertex u of thehost, we set up a bipartite graph as follows:
One partition is the set of neighbors of u
The other neighbors of v
Vertices i , j in the bipartition have an edge if vertex i is acandidate for vertex j .
We call this the neighborhood bipartition, Ku,v .
Theorem
For vertex u to be a candidate for vertex v upon assignment of vto u, a matching on Ku,v the size of the partition v must exist.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 39: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/39.jpg)
Matching Reductions
Upon assignment of vertex v of the pattern to vertex u of thehost, we set up a bipartite graph as follows:
One partition is the set of neighbors of u
The other neighbors of v
Vertices i , j in the bipartition have an edge if vertex i is acandidate for vertex j .
We call this the neighborhood bipartition, Ku,v .
Theorem
For vertex u to be a candidate for vertex v upon assignment of vto u, a matching on Ku,v the size of the partition v must exist.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 40: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/40.jpg)
Matching Reductions
Upon assignment of vertex v of the pattern to vertex u of thehost, we set up a bipartite graph as follows:
One partition is the set of neighbors of u
The other neighbors of v
Vertices i , j in the bipartition have an edge if vertex i is acandidate for vertex j .
We call this the neighborhood bipartition, Ku,v .
Theorem
For vertex u to be a candidate for vertex v upon assignment of vto u, a matching on Ku,v the size of the partition v must exist.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 41: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/41.jpg)
Matching Reductions
Upon assignment of vertex v of the pattern to vertex u of thehost, we set up a bipartite graph as follows:
One partition is the set of neighbors of u
The other neighbors of v
Vertices i , j in the bipartition have an edge if vertex i is acandidate for vertex j .
We call this the neighborhood bipartition, Ku,v .
Theorem
For vertex u to be a candidate for vertex v upon assignment of vto u, a matching on Ku,v the size of the partition v must exist.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 42: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/42.jpg)
Matching Reductions
Upon assignment of vertex v of the pattern to vertex u of thehost, we set up a bipartite graph as follows:
One partition is the set of neighbors of u
The other neighbors of v
Vertices i , j in the bipartition have an edge if vertex i is acandidate for vertex j .
We call this the neighborhood bipartition, Ku,v .
Theorem
For vertex u to be a candidate for vertex v upon assignment of vto u, a matching on Ku,v the size of the partition v must exist.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 43: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/43.jpg)
Perfect Matching Requirement
We set up a bipartite graph with the same procedure as with thematching reduction; however, the partitions are the entire set ofvertices for H and P. We call this graph, KH,P
Theorem
Given KH,P , for there to exist a subgraph isomorphism from P toH, a perfect matching must exist on KH,P .
Proof.
By way of contradiction suppose there is no perfect matching onKH,P . Since the partitions of KH,P are of equal order, this impliesat least one vertex of each partition has no edge connecting them.This means at least one vertex from P cannot be mapped to anyvertex from H. Since at least one vertex from the pattern is leftout regardless of which edge configuration we choose for thematching, at least one vertex cannot be mapped.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 44: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/44.jpg)
Perfect Matching Requirement
We set up a bipartite graph with the same procedure as with thematching reduction; however, the partitions are the entire set ofvertices for H and P. We call this graph, KH,P
Theorem
Given KH,P , for there to exist a subgraph isomorphism from P toH, a perfect matching must exist on KH,P .
Proof.
By way of contradiction suppose there is no perfect matching onKH,P . Since the partitions of KH,P are of equal order, this impliesat least one vertex of each partition has no edge connecting them.This means at least one vertex from P cannot be mapped to anyvertex from H. Since at least one vertex from the pattern is leftout regardless of which edge configuration we choose for thematching, at least one vertex cannot be mapped.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 45: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/45.jpg)
Perfect Matching Requirement
We set up a bipartite graph with the same procedure as with thematching reduction; however, the partitions are the entire set ofvertices for H and P. We call this graph, KH,P
Theorem
Given KH,P , for there to exist a subgraph isomorphism from P toH, a perfect matching must exist on KH,P .
Proof.
By way of contradiction suppose there is no perfect matching onKH,P . Since the partitions of KH,P are of equal order, this impliesat least one vertex of each partition has no edge connecting them.This means at least one vertex from P cannot be mapped to anyvertex from H. Since at least one vertex from the pattern is leftout regardless of which edge configuration we choose for thematching, at least one vertex cannot be mapped.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 46: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/46.jpg)
Perfect Matching Requirement (Example)
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 47: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/47.jpg)
Perfect Matching Requirement (Example)
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 48: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/48.jpg)
Skeleton
Reduce the candidate matrix as much as possible
Once all reductions are made, assign a vertex
Reduce the new candidate matrix as much as possible, checkfor fail conditions
Repeat until search tree is exhausted
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 49: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/49.jpg)
Skeleton
Reduce the candidate matrix as much as possible
Once all reductions are made, assign a vertex
Reduce the new candidate matrix as much as possible, checkfor fail conditions
Repeat until search tree is exhausted
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 50: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/50.jpg)
Skeleton
Reduce the candidate matrix as much as possible
Once all reductions are made, assign a vertex
Reduce the new candidate matrix as much as possible, checkfor fail conditions
Repeat until search tree is exhausted
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 51: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/51.jpg)
Skeleton
Reduce the candidate matrix as much as possible
Once all reductions are made, assign a vertex
Reduce the new candidate matrix as much as possible, checkfor fail conditions
Repeat until search tree is exhausted
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 52: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/52.jpg)
Data
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 53: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/53.jpg)
Data
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 54: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/54.jpg)
References
Douglas West, Introduction to Graph Theory, Prentice-Hall,2nd edition, 2001
Julian R. Ullman, An algorithm for subgraph isomorphism, J.of the ACM 23 (1976), 31-42.
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism
![Page 55: Department of Mathematics Nebraska Wesleyan University · 2016. 9. 8. · Nebraska Wesleyan University August 4, 2016 Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph](https://reader035.vdocuments.site/reader035/viewer/2022071216/6046a60244b5e93fd06bbadc/html5/thumbnails/55.jpg)
Thanks
Thanks for listening!
Drew Meier, Dr. Austin Mohr, Thomas Schuler Spanning Subgraph Isomorphism