Trees and the Matrix Tree Theorem Flashcards

1
Q

acyclic:

A

a graph is acyclic if it has no cycles

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

tree:

A

a connected, acyclic graph

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

forest:

A

a graph whose connected components are all trees
note - a forest is not a tree, a tree must be connected and a forest cannot be

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

leaf:

A

a vertex in a tree is a leaf (node) if deg(v)=1
it’s an internal node if deg(v)>=1

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

binary tree:

A

a tree in which every internal node has deg(v)=3 (2 out, one to lead to it)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

rooted tree:

A

a tree with a leaf node that is designated the root node

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

isolated:

A

a vertex is isolated if it has no neighbours, isn’t connected to anything, deg(v)=0

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

G\v:

A

the subgraph created from G by deleting the vertex v and all associated edges (incident edges to call them properly)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

G\e:

A

the subgraph created from G by deleting the edge e (not any attached vertices, they can stay)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

a connected graph on n vertices has at least (n-1) edges proof:

A

base case - |V|=1, |E|=0, all by definition, which fits
hypothesis - suppose it’s true for all graphs G(V,E) with 1<=|V|<=n0, for some fixed n0
induction - consider the connected graph G(V,E) with |V|=n0+1, so we want to reach |E|>=n0. choose any vertex v and create G\v, with vertex set V’=V\v and edge set E’.
if G\v is connected still, then |V’|=|V|-1=n0, hypothesis says then |E’|>=n0-1, but we deleted a vertex that must have been connected to at least one other by an edge cause the G is connected, so |E|>=|E’|+1>=n0-1+1, so |E|>=n0
if G\v is now k>=2 connected components, call them G1(V1,E1),…,Gk(Vk,Ek) with nj=|Vj| (arbitrary number 1<=j<=k)
(k)Σ(j=1)nj=(k)Σ(j=1)|Vj|=|V’|=|V|-1=n0
hypothesis applies to each component so
|E’|=(k)Σ(j=1)|Ej|>=(k)Σ(j=1)(nj-1)>=((k)Σ(j=1)nj)-k>=n0-k
and cause we know G is connected, deleted v must have been connected to k other vertices, so |E|>=|E’|+k>=n0-k+k, so |E|>=n0 as desired

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

an acyclic graph on n vertices has at most (n-1) edges proof:

A

base case - K1, acyclic with max n-1 edges
hypothesis - suppose it’s true for all acyclic graphs with |V|<=n0 for some fixed n0
induction - consider an acyclic graph G(V,E) with |V|=n0+1, so we want to reach |E|=n0. choose any edge e and delete it to give G\e, which has the same vertex set V but E’=E\e. G\e must have one more connected component than G, as G is acyclic so nothing else can be connecting those 2 vertices. so G\e has k>=2 connected components, G1(V1,E1),…,Gk(Vk,Ek). nj=|Vj|(arbitrary number 1<=j<=k), nj<=n0 for all j so the hypothesis applies to each component, |Ej|<=nj-1
|E’|=(k)Σ(j=1)|Ej|<=(k)Σ(j=1)(nj-1)<=((k)Σ(j=1)nj)-k<=n0+1-k
|E|=|E’|+1<=n0+1-k+1<=n0+2-k<=n0 (cause G’ has k>=2 connected components

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

if a graph G has n>=2 vertices, none of which are isolated, and (n-1) edges, then G has at least 2 vertices for which deg(v)=1 proof:

A

say the vertices are numbered and arranged in order of increasing degree (so deg(v1)<=deg(v2)<=deg(v3)…)
by the handshaking lemma, (n)Σ(j=1)deg(vj)=2|E|=2(n-1)=2n-2
no vertices are isolated so deg(vj)>=1 for all j
assume, aiming for a contradiction, that there is at most 1 vertex with degree 1
then (n)Σ(j=1)deg(vj)=deg(v1)+(n)Σ(j=2)deg(vj)>=1+(n)Σ(j=2)2>=1+2(n-1)>=2n-1 which is a contradiction

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

prove that any 2 of the following imply the third:
(a) G is connected
(b) G is acyclic
(c) G has n-1 edges

A

(a) and (b): a connected graph has |E|>=(n-1), an acyclic graph has |E|<=(n-1), so |E| must equal n-1
(a) and (c): assume for a contradiction that you can have a connected graph with n-1 edges and a cycle. choose some edge e and delete it for G\e. G is then still connected, and has n-2 edges, which is a contradiction.
(b) and (c): base case - |V|=1, it’s acyclic, |E|=n-1, and it’s connected by definition
hypothesis - suppose all acyclic graphs with 1<=|V|<=n0 vertices and |E|=|V|-1 edges are connected
induction - consider an acyclic graph G(V,E) with |V|=n0+1 and |E|=n0. such a graph cannot have any isolated vertices cause then we could delete it to have a graph with |V|=n0 and |E|=n0 which is a contradiction. therefore, it has at least 2 vertices of degree 1. take one of these and delete it for G\v=G’(V’,E’). G’ is still acyclic cause deleting stuff can’t create a cycle, and |V’|=|V|-1=n0 vertices and |E’|=|E|-1=n0-1 edges. this means the hypothesis applies so it’s connected, and cause G’ is connected, G must also be

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

spanning tree:

A

a subgraph of G(V,E) that is a tree that contains every vertex in V (but not necessarily all edges)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

graph laplacian:

A

a matrix where Lij=deg(vj) if vi=vj, -1 if vi!=vj and (vi,vj) is an edge, and 0 otherwise
alternatively, L=the matrix where the diagonal is the degree of that vertex - the adjacency matrix

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

kirchoff’s matrix-tree theorem:

A

if G(V,E) is an undirected graph with graph laplacian L, then the number NT of spanning trees contained in G is given by taking L, removing some vertex (one row and one column) and computing the determinant of that matrix

17
Q

root:

A

a vertex v in a digraph is a root if every other vertex is accessible from v

18
Q

arborescence:

A

a digraph is an arborescence (or directed tree) if it contains a root and the graph obtained by ignoring direction of edges is a tree

19
Q

spanning arborescence:

A

a subgraph that’s an arborescence that contains all vertices of G

20
Q

tutte’s matrix-tree theorem:

A

kirchoffs but for digraphs
if G(V,E) is an undirected graph and L is a matrix with Lij=deg(in)(vj) if vi=vj, -1 if vi!=vj and (vi,vj) is an edge, and 0 otherwise, then the number NT of spanning arborescences with root vj is given by taking L, removing vj’s row and column and computing the determinant of that matrix

21
Q

fix(a) of a permutation:

A

{j|a(j)=j} - where it’s the same before and after permutation

22
Q

symmetric group:

A

Sn
the identity element is the permutation for which a(j)=j for all j in the set
Sn has n! elements

23
Q

cycle (permutation):

A

a permutation a specified by a sequence of distinct integers such that a(j)=j if j isn’t in the sequence, a(i(j))=i(j+1) for 1<=j<l (l being the length of the sequence), and a(i(l))=i(1)
brackets after i are subscript

24
Q

transposition:

A

a cycle (permutation) with l=2

25
Q

sign of a permutation:

A

sgn: Sn -> ±1
if a is the identity permutation, sgn(a)=1
if a is a cycle of length l, sgn(a)=(-1)^(l-1)
if a has a decomposition into k>=2 disjoint cycles with lengths l1,..,lk, sgn(a)=-1^(L-k) where L=(k)Σ(j=1)lj
all that’s the same as saying sgn(a)=1 if a is the product of an even number of transpositions, -1 otherwise

26
Q

Ga:

A

a is a permutation
V={1,…,n}
E={j,a(j)|j in V and a(j)!=j}

27
Q

the cycle (i1,…,il) appears in the cycle decomposition of a iff:

A

the directed cycle defined by the vertex sequence (i1,…,il,i1) is a subgraph of Ga

28
Q

inclusion/exclusion:

A

|X1∪X2|=|X1|+|X2|-|X1∩X2|

29
Q

spreg:

A

a single predecessor graph is a subgraph T(V,E’) of digraph G(V,E) with distinguished vertex v in which each vertex has exactly one predeccesor except v, which has 0
note that the vertex set is the same and that the graph need not be connected, as long as there’s only one v and everything else has 1 predeccesor

30
Q

spanning arborescences with root v are spregs with distinguished vertex v proof:

A

the vertex set is the same by definition
we are checking all the vertices that aren’t the distinguished vertex have exactly 1 predecessor and that there is a distinguished one, v, that has 0
assume for a contradiction that deg(in)(v)>0
therefore T must include a directed cycle - consider the predecessor of v, u0 - it’s accessible from v, so there must be a directed path from v to u0, which means a path from v to v, which contradicts T’s definition as an arborescence, so deg(in)(v)=0
assume for a contradiction that there exists some vertex u such that deg(in)(u)>=2. take two predecessors v1 and v2, noting that one could be v. consider the paths from v to v1 and v2. this shows that there must be a cycle, which contradicts T being an arborescence

31
Q

characterising spregs:

A

a spreg with distinguished vertex v consists of an arborescence rooted at v plus 0 or more disjoint weakly connected components, each of which contains a single directed cycle. proof is similar to the proof that spanning arborescences with root v are spregs with distinguished v