• 沒有找到結果。

Slides credited from Hsueh-I Lu, Hsu-Chun Hsiao, & Michael Tsai

N/A
N/A
Protected

Academic year: 2022

Share "Slides credited from Hsueh-I Lu, Hsu-Chun Hsiao, & Michael Tsai"

Copied!
82
0
0

加載中.... (立即查看全文)

全文

(1)

Slides credited from Hsueh-I Lu, Hsu-Chun Hsiao, & Michael Tsai

(2)

Mini-HW 7 released

Due on 11/30 (Thur) 17:20

Homework 3 released tonight

Due on 12/14 (Thur) 17:20 (three weeks)

Class change!!

Start from 12/07 (Thur) to forever (NOT next week!!)

Location: R103

Frequently check the website for the updated information!

(3)

3

(4)

Graph Basics

Graph Theory

Graph Representations

Graph Traversal

Breadth-First Search (BFS)

Depth-First Search (DFS)

DFS Applications

Connected Components

Strongly Connected Components

Topological Sorting

(5)

A graph G is defined as

V: a finite, nonempty set of vertices

E: a set of edges / pairs of vertices

5

3 5

1

4

2

(6)

Graph type

Undirected: edge 𝑢, 𝑣 = 𝑣, 𝑢

Directed: edge 𝑢, 𝑣 goes from vertex 𝑢 to vertex 𝑣; 𝑢, 𝑣 ≠ 𝑣, 𝑢

Weighted: edges associate with weights

3 5

1 4

2

3 5

1 4

2

(7)

Adjacent (相鄰)

If there is an edge 𝑢, 𝑣 , then 𝑢 and 𝑣 are adjacent.

Incident (作用)

If there is an edge 𝑢, 𝑣 , the edge

𝑢, 𝑣

is incident from 𝑢 and is incident to 𝑣.

Subgraph (子圖)

If a graph 𝐺

= 𝑉

, 𝐸′ is a subgraph of 𝐺 = 𝑉, 𝐸 , then 𝑉

⊆ 𝑉 and 𝐸

⊆ 𝐸

7

(8)

Degree

The degree of a vertex

𝑢

is the number of edges incident on

𝑢

In-degree of 𝑢: #edges 𝑥, 𝑢 in a directed graph

Out-degree of 𝑢: #edges 𝑢, 𝑥 in a directed graph

Degree = in-degree + out-degree

Isolated vertex: degree = 0

𝐸 = σ

𝑖

𝑑

𝑖

2

(9)

Path

a sequence of edges that connect a sequence of vertices

If there is a path from 𝑢 (source) to 𝑣 (target), there are a sequence of edges 𝑢, 𝑖

1

, 𝑖

1

, 𝑖

2

, … , 𝑖

𝑘−1

, 𝑖

𝑘

, (𝑖

𝑘

, 𝑣)

Reachable: 𝑣 is reachable from 𝑢 if there exists a path from 𝑢 to 𝑣

Simple Path

All vertices except for 𝑢 and 𝑣 are all distinct

Cycle

A simple path where 𝑢 and 𝑣 are the same

Subpath

A subsequence of the path

9

(10)

Connected

Two vertices are connected if there is a path between them

A connected graph has a path from every vertex to every other

Tree

a connected, acyclic, undirected graph

Forest

an acyclic, undirected but possibly disconnected graph

3 5

1 4

2

3 5

1 4

2

3 5

1 4

2

(11)

Theorem. Let 𝐺 be an undirected graph. The following statements are equivalent:

𝐺 is a tree

Any two vertices in 𝐺 are connected by a unique simple path

𝐺 is connected, but if any edge is removed from 𝐸, the resulting graph is disconnected.

𝐺 is connected and 𝐸 = 𝑉 − 1

𝐺 is acyclic, and 𝐸 = 𝑉 − 1

𝐺 is acyclic, but if any edge is added to 𝐸, the resulting graph contains a cycle

11

Proofs in Textbook Appendix B.5

(12)

12

(13)

How to traverse all bridges where each one can only be passed through once

13

A

B

D

C C

A

B

D

(14)

Euler path

Can you traverse each edge in a connected graph exactly once without lifting the pen from the paper?

Euler tour

Can you finish where you started?

14

C A

B D

C A

B D

C A

B D

Euler path Euler path Euler path

(15)

Solved by Leonhard Euler in 1736

𝐺 has an Euler path iff 𝐺 has exactly 0 or 2 odd vertices

𝐺 has an Euler tour iff all vertices must be even vertices

15

Is it possible to determine whether a graph has an Euler path or an Euler tour, without necessarily having to find one explicitly?

Even vertices = vertices with even degrees Odd vertices = vertices with odd degrees

(16)

Hamiltonian Path

A path that visits each vertex exactly once

Hamiltonian Cycle

A Hamiltonian path where the start and destination are the same

Both are NP-complete

(17)

Modeling applications using graph theory

What do the vertices represent?

What do the edges represent?

Undirected or directed?

Social Network Knowledge Graph 17

(18)

18

(19)

How to represent a graph in computer programs?

Two standard ways to represent a graph 𝐺 = 𝑉, 𝐸

Adjacency matrix

Adjacency list

19

(20)

Matrix

Adjacency matrix = 𝑉 × 𝑉 matrix 𝐴 with 𝐴[𝑢][𝑣] = 1 if (𝑢, 𝑣) is an edge

1 2 3 4 5 6

1 1 1

2 1 1 1

3 1 1 1

4 1 1 1

5 1

6 1 1

1

2

3

5 4

6

• For undirected graphs, 𝐴 is symmetric; i.e., 𝐴 = 𝐴𝑇

• If weighted, store weights instead of bits in 𝐴

(21)

Matrix

Space:

Time for querying an edge:

Time for inserting an edge:

Time for deleting an edge:

Time for listing all neighbors of a vertex:

Time for identifying all edges:

Time for finding in-degree and out-degree of a vertex?

21

(22)

List

Adjacency lists = vertex indexed array of lists

One list per vertex, where for 𝑢 ∈ 𝑉, 𝐴[𝑢] consists of all vertices adjacent to 𝑢

1 2 3 4 5 6

1 4

3

2 3

2

3 2

5

1 4 6

4

6 1

2

3

5 4

6

If weighted, store weights also in adjacency lists

(23)

List

Space:

Time for querying an edge:

Time for inserting an edge:

Time for deleting an edge:

Time for listing all neighbors of a vertex:

Time for identifying all edges:

Time for finding in-degree and out-degree of a vertex?

23

(24)

Matrix representation is suitable for dense graphs

List representation is suitable for sparse graphs

Besides graph density, you may also choose a data structure based on the performance of other operations

Space Query an edge

Insert an edge

Delete an edge

List a vertex’s neighbors

Identify all edges

Adjacency Matrix Adjacency List

(25)

Textbook Chapter 22 – Elementary Graph Algorithms

25

(26)

From a source vertex, systematically follow the edges of a graph to visit all reachable vertices of the graph

Useful to discover the structure of a graph

Standard graph-searching algorithms

Breadth-First Search (BFS, 廣度優先搜尋)

Depth-First Search (DFS, 深度優先搜尋)

(27)

Textbook Chapter 22.2 – Breadth-first search

27

(28)

Source 𝒔

Layer 1

Layer 2

(29)

Input: directed/undirected graph 𝐺 = (𝑉, 𝐸) and source 𝑠

Output: a breadth-first tree with root 𝑠 (𝑇

BFS

) that contains all reachable vertices

𝑣. 𝑑: distance from 𝑠 to 𝑣, for all 𝑣 ∈ 𝑉

Distance is the length of a shortest path in G

𝑣. 𝑑 = ∞ if 𝑣 is not reachable from 𝑠

𝑣. 𝑑 is also the depth of 𝑣 in 𝑇BFS

𝑣. 𝜋 = 𝑢 if (𝑢, 𝑣) is the last edge on shortest path to 𝑣

𝑢 is 𝑣’s predecessor in 𝑇BFS

29

(30)

Initially 𝑇

BFS

contains only 𝑠

As 𝑣 is discovered from 𝑢, 𝑣 and (𝑢, 𝑣) are added to 𝑇

BFS

𝑇BFS is not explicitly stored; can be reconstructed from 𝑣. 𝜋

Implemented via a FIFO queue

Color the vertices to keep track of progress:

GRAY: discovered (first time encountered)

BLACK: finished (all adjacent vertices discovered)

WHITE: undiscovered

BFS(G, s)

for each vertex u in G.V-{s}

u.color = WHITE u.d = ∞

u.pi = NIL s.color = GRAY s.d = 0

s.pi = NIL Q = {}

ENQUEUE(Q, s) while Q! = {}

u = DEQUEUE(Q)

for each v in G.Adj[u]

if v.color == WHITE v.color = GRAY v.d = u.d + 1 v.pi = u

ENQUEUE(Q,v) u.color = BLACK

(31)

𝑠 0

31

𝑤 𝑟

1 1

𝑟 𝑡 𝑥

1 2 2

𝑡 𝑥 𝑣

2 2 2

𝑥 𝑣 𝑢

2 2 3

𝑣 𝑢 𝑦

2 3 3

(32)

𝑢 𝑦

3 3

𝑦 3

(33)

Definition of 𝛿(𝑠, 𝑣): the shortest-path distance from 𝑠 to 𝑣 = the minimum number of edges in any path from 𝑠 to 𝑣

If there is no path from 𝑠 to 𝑣, then 𝛿 𝑠, 𝑣 = ∞

The BFS algorithm finds the shortest-path distance to each reachable vertex in a graph 𝐺 from a given source vertex 𝑠 ∈ 𝑉.

33

(34)

Proof

Case 1: 𝑢 is reachable from 𝑠

𝑠- 𝑢- 𝑣 is a path from 𝑠 to 𝑣 with length 𝛿 𝑠, 𝑢 + 1

Hence, 𝛿 𝑠, 𝑣 ≤ 𝛿 𝑠, 𝑢 + 1

Case 2: 𝑢 is unreachable from 𝑠

Then 𝑣 must be unreachable too.

Hence, the inequality still holds.

Lemma 22.1

Let 𝐺 = 𝑉, 𝐸 be a directed or undirected graph, and let 𝑠 ∈ 𝑉 be an arbitrary vertex. Then, for any edge 𝑢, 𝑣 ∈ 𝐸, 𝛿 𝑠, 𝑣 ≤ 𝛿 𝑠, 𝑢 + 1.

𝑠-𝑣的最短路徑一定會小於等於𝑠-𝑢的最短路徑距離+1

s

v

𝛿 𝑠, 𝑢 u

(35)

Proof by induction

Holds when 𝑛 = 1: 𝑠 is in the queue and 𝑣. 𝑑 = ∞ for all 𝑣 ∈ 𝑉 𝑠

After 𝑛 + 1 ENQUEUE ops, consider a white vertex 𝑣 that is discovered during the search from a vertex 𝑢

Vertex 𝑣 is never enqueued again, so 𝑣. 𝑑 never changes again

35

Lemma 22.2

Let 𝐺 = 𝑉, 𝐸 be a directed or undirected graph, and suppose BFS is run on 𝐺 from a given source vertex 𝑠 ∈ 𝑉. Then upon termination, for each vertex 𝑣 ∈ 𝑉, the value 𝑣. 𝑑 computed by BFS satisfies 𝑣. 𝑑 ≥ 𝛿 𝑠, 𝑣 .

BFS算出的d值必定大於等於真正距離

Inductive hypothesis: 𝑣. 𝑑 ≥ 𝛿 𝑠, 𝑣 after 𝑛 ENQUEUE ops

(by induction hypothesis) (by Lemma 22.1)

(36)

Proof by induction

Holds when 𝑄 = 𝑠 .

Consider two operations for inductive step:

Dequeue op: when 𝑄 = 𝑣1, 𝑣2, … , 𝑣𝑟 and dequeue 𝑣1

Enqueue op: when 𝑄 = 𝑣1, 𝑣2, … , 𝑣𝑟 and enqueue 𝑣𝑟+1

Lemma 22.3

Suppose that during the execution of BFS on a graph 𝐺 = 𝑉, 𝐸 , the queue 𝑄 contains the vertices 𝑣1, 𝑣2, … , 𝑣𝑟 , where 𝑣1 is the head of 𝑄 and 𝑣𝑟 is the tail. Then, 𝑣𝑟. 𝑑 ≤ 𝑣1. 𝑑 + 1 and 𝑣𝑖. 𝑑 ≤ 𝑣𝑖+1. 𝑑 for 1 ≤ 𝑖 < 𝑟.

• Q中最後一個點的d值 ≤ Q中第一個點的d值+1

• Q中第i個點的d值 ≤ Q中第i+1點的d值

Inductive hypothesis:𝑣𝑟. 𝑑 ≤ 𝑣1. 𝑑 + 1 and 𝑣𝑖. 𝑑 ≤ 𝑣𝑖+1. 𝑑 after 𝑛 queue ops

(37)

Dequeue op

Enqueue op

37

Inductive hypothesis:

𝑣1 𝑣2 … 𝑣𝑟−1 𝑣𝑟

𝑣2 … 𝑣𝑟−1 𝑣𝑟 (induction hypothesis H2)

𝑣1 𝑣2 … 𝑣𝑟−1 𝑣𝑟

(induction hypothesis H2) 𝑣1 𝑣2 … 𝑣𝑟−1 𝑣𝑟 𝑣𝑟+1

𝑢

Let 𝑢 be 𝑣𝑟+1’s predecessor,

Since 𝑢 has been removed from 𝑄, the new head 𝑣1 satisfies

(induction hypothesis H1) H1

H2

 H1 holds

 H2 holds

 H1 holds 𝑢

(induction hypothesis H1)

 H2 holds

(Q中最後一個點的d值 ≤ Q中第一個點的d值+1)

(Q中第i個點的d值 ≤ Q中第i+1點的d值)

(38)

Proof

Lemma 22.3 proves that 𝑣𝑖. 𝑑 ≤ 𝑣𝑖+1. 𝑑 for 1 ≤ 𝑖 < 𝑟

Each vertex receives a finite 𝑑 value at most once during the course of BFS

Hence, this is proved.

Corollary 22.4

Suppose that vertices 𝑣𝑖 and 𝑣𝑗 are enqueued during the execution of BFS, and that 𝑣𝑖 is enqueued before 𝑣𝑗. Then 𝑣𝑖. 𝑑 ≤ 𝑣𝑗. 𝑑 at the time that 𝑣𝑗 is enqueued.

若𝑣𝑖比𝑣𝑗早加入queue  𝑣𝑖. 𝑑 ≤ 𝑣𝑗. 𝑑

(39)

Proof of (1)

All vertices 𝑣 reachable from 𝑠 must be discovered; otherwise they would have 𝑣. 𝑑 = ∞ > 𝛿 𝑠, 𝑣 . (contradicting with Lemma 22.2)

39

Theorem 22.5 – BFS Correctness

Let 𝐺 = 𝑉, 𝐸 be a directed or undirected graph, and and suppose that BFS is run on 𝐺 from a given source vertex 𝑠 ∈ 𝑉.

1) BFS discovers every vertex 𝑣 ∈ 𝑉 that is reachable from the source 𝑠 2) Upon termination, 𝑣. 𝑑 = 𝛿 𝑠, 𝑣 for all 𝑣 ∈ 𝑉

3) For any vertex 𝑣 ≠ 𝑠 that is reachable from 𝑠, one of the shortest paths from 𝑠 to 𝑣 is a shortest path from 𝑠 to 𝑣. 𝜋 followed by the edge 𝑣. 𝜋, 𝑣

(40)

(2)

Proof of (2) by contradiction

Assume some vertices receive 𝑑 values not equal to its shortest-path distance

Let 𝑣 be the vertex with minimum 𝛿 𝑠, 𝑣 that receives such an incorrect 𝑑 value; clearly 𝑣 ≠ 𝑠

By Lemma 22.2, 𝑣. 𝑑 ≥ 𝛿 𝑠, 𝑣 , thus 𝑣. 𝑑 > 𝛿 𝑠, 𝑣 (𝑣 must be reachable)

Let 𝑢 be the vertex immediately preceding 𝑣 on a shortest path from 𝑠 to 𝑣, so 𝛿 𝑠, 𝑣 = 𝛿 𝑠, 𝑢 + 1

Because 𝛿 𝑠, 𝑢 < 𝛿 𝑠, 𝑣 and 𝑣 is the minimum 𝛿 𝑠, 𝑣 , we have 𝑢. 𝑑 = 𝛿 𝑠, 𝑢

𝑣. 𝑑 > 𝛿 𝑠, 𝑣 = 𝛿 𝑠, 𝑢 + 1 = 𝑢. 𝑑 + 1

(41)

(2)

Proof of (2) by contradiction (cont.)

𝑣. 𝑑 > 𝛿 𝑠, 𝑣 = 𝛿 𝑠, 𝑢 + 1 = 𝑢. 𝑑 + 1

When dequeuing 𝑢 from 𝑄, vertex 𝑣 is either WHITE, GRAY, or BLACK

WHITE: 𝑣. 𝑑 = 𝑢. 𝑑 + 1, contradiction

BLACK: it was already removed from the queue

By Corollary 22.4, we have 𝑣. 𝑑 ≤ 𝑢. 𝑑, contradiction

GRAY: it was painted GRAY upon dequeuing some vertex 𝑤

Thus 𝑣. 𝑑 = 𝑤. 𝑑 + 1 (by construction)

𝑤 was removed from 𝑄 earlier than 𝑢, so 𝑤. 𝑑 ≤ 𝑢. 𝑑 (by Corollary 22.4)

𝑣. 𝑑 = 𝑤. 𝑑 + 1 ≤ 𝑢. 𝑑 + 1, contradiction

Thus, (2) is proved.

41

(42)

(3) For any vertex 𝑣 ≠ 𝑠 that is reachable from 𝑠, one of the shortest paths from 𝑠 to 𝑣 is a shortest path from 𝑠 to 𝑣. 𝜋 followed by the edge 𝑣. 𝜋, 𝑣

Proof of (3)

If 𝑣. 𝜋 = 𝑢, then 𝑣. 𝑑 = 𝑢. 𝑑 + 1. Thus, we can obtain a shortest path from 𝑠 to 𝑣 by taking a shortest path from 𝑠 to 𝑣. 𝜋 and then traversing the edge 𝑣. 𝜋, 𝑣 .

(43)

BFS(G, s) forms a BFS tree with all reachable 𝑣 from 𝑠

We can extend the algorithm to find a BFS forest that contains every vertex in 𝐺

43

BFS-Visit(G, s) s.color = GRAY s.d = 0

s.π = NIL Q = empty ENQUEUE(Q, s) while Q ≠ empty

u = DEQUEUE(Q) for v in G.adj[u]

if v.color == WHITE v.color = GRAY v.d = u.d + 1 v.π = u

ENQUEUE(Q, v) u.color = BLACK

//explore full graph and builds up a collection of BFS trees

BFS(G)

for u in G.V

u.color = WHITE u.d = ∞

u.π = NIL for s in G.V

if(s.color == WHITE) // build a BFS tree BFS-Visit(G, s)

(44)

Textbook Chapter 22.3 – Depth-first search

44

(45)

Search as deep as possible and then backtrack until finding a new path

Timestamps: discovery time / finishing time 45

1

2

3 4

8

9 12 13

14

5 6

7

10 11

(46)

Implemented via recursion (stack)

Color the vertices to keep track of progress:

GRAY: discovered (first time encountered)

BLACK: finished (all adjacent vertices discovered)

// Explore full graph and builds up a collection of DFS trees

DFS(G)

for each vertex u in G.V u.color = WHITE

u.pi = NIL

time = 0 // global timestamp for each vertex u in G.V

if u.color == WHITE DFS-VISIT(G, u)

DFS-Visit(G, u) time = time + 1

u.d = time // discover time u.color = GRAY

for each v in G.Adj[u]

if v.color == WHITE v.pi = u

DFS-VISIT(G, v) u.color = BLACK

time = time + 1

u.f = time // finish time

(47)

Parenthesis Theorem

Parenthesis structure: represent the discovery of vertex 𝑢 with a left

parenthesis “(𝑢” and represent its finishing by a right parenthesis “𝑢)”. In DFS, the parentheses are properly nested.

White Path Theorem

In a DFS forest of a directed or undirected graph 𝐺 = 𝑉, 𝐸 ,

vertex 𝑣 is a descendant of vertex 𝑢 in the forest  at the time 𝑢. 𝑑 that the search discovers 𝑢, there is a path from 𝑢 to 𝑣 in 𝐺 consisting entirely of WHITE vertices

Classification of Edges in 𝐺

Tree Edge

Back Edge

Forward Edge

Cross Edge

47

(48)

Parenthesis Theorem

Parenthesis structure: represent the discovery of vertex 𝑢 with a left parenthesis “(𝑢” and represent its finishing by a right parenthesis

“𝑢)”. In DFS, the parentheses are properly nested.

Properly nested: (x (y y) x) Not properly nested: (x (y x) y)

Proof in textbook p. 608

(49)

White Path Theorem

In a DFS forest of a directed or undirected graph 𝐺 = 𝑉, 𝐸 ,

vertex 𝑣 is a descendant of vertex 𝑢 in the forest  at the time 𝑢. 𝑑 that the search discovers 𝑢, there is a path from 𝑢 to 𝑣 in 𝐺 consisting entirely of WHITE vertices

Proof.

Since 𝑣 is a descendant of 𝑢, 𝑢. 𝑑 < 𝑣. 𝑑

Hence, 𝑣 is WHITE at time 𝑢. 𝑑

In fact, since 𝑣 can be any descendant of 𝑢, any vertex on the path from 𝑢 to 𝑣 are WHITE at time 𝑢. 𝑑

 (textbook p. 608)

49

(50)

Classification of Edges in 𝐺

Tree Edge (GRAY to WHITE)

Edges in the DFS forest

Found when encountering a new vertex 𝑣 by exploring 𝑢, 𝑣

Back Edge (GRAY to GRAY)

𝑢, 𝑣 , from descendant 𝑢 to ancestor 𝑣 in a DFS tree

Forward Edge (GRAY to BLACK)

𝑢, 𝑣 , from ancestor 𝑢 to descendant 𝑣. Not a tree edge.

Cross Edge (GRAY to BLACK)

Any other edge between trees or subtrees. Can go between vertices in same DFS tree or in different DFS trees

In an undirected graph, back edge = forward edge.

To avoid ambiguity, classify edge as the first type in the list that applies.

(51)

Edge classification by the color of 𝑣 when visiting 𝑢, 𝑣

WHITE: tree edge

GRAY: back edge

BLACK: forward edge or cross edge

𝑢. 𝑑 < 𝑣. 𝑑  forward edge

𝑢. 𝑑 > 𝑣. 𝑑  cross edge

51

Why?

Theorem 22.10

In DFS of an undirected graph, there are only tree edges and back edges without forward and cross edge.

(52)

Connected Components

Strongly Connected Components

Topological Sort

(53)

53

(54)

Input: a graph 𝐺 = 𝑉, 𝐸

Output: a connected component of 𝐺

a maximal subset 𝑈 of 𝑉 s.t. any two nodes in 𝑈 are connected in 𝐺

(55)

55

10 1

2

5

3 4

6

7

8 9

Time Complexity:

BFS and DSF both find the connected components with the same complexity

(56)
(57)

Textbook Chapter 22.5 – Strongly connected components

57

(58)

Input: a directed graph 𝐺 = 𝑉, 𝐸

Output: a connected component of 𝐺

a maximal subset 𝑈 of 𝑉 s.t. any two nodes in 𝑈 are reachable in 𝐺

1

2

4

6 3

5

7

8

Why must the strongly connected components of a graph be disjoint?

(59)

59

Step 1: Run DFS on 𝐺 to obtain the finish time 𝑣. 𝑓 for 𝑣 ∈ 𝑉.

Step 2: Run DFS on the transpose of 𝐺 where the vertices 𝑉 are processed in the decreasing order of their finish time.

Step 3: output the vertex partition by the second DFS

(60)

1

2

4

6 3

5

1

2

4

6 3

5

(61)

61

1

3

2

6 5

4 1

2

4

5 3

6

(62)

Proof by contradiction

Assume that 𝑣, 𝑤 is an incoming edge to 𝐶.

Since 𝐶 is a strongly connected component of 𝐺, there cannot be any path from any node of 𝐶 to 𝑣 in 𝐺.

Therefore, the finish time of 𝑣 has to be larger than any node in 𝐶, including 𝑢.  𝑣. 𝑓 > 𝑢. 𝑓, contradiction

𝑢

G

C

𝑤

𝑣

Lemma

Let 𝐶 be the strongly connected component of 𝐺 (and 𝐺𝑇) that contains the node 𝑢 with the largest finish time 𝑢. 𝑓. Then 𝐶 cannot have any incoming edge from any node of 𝐺 not in 𝐶.

(63)

Practice to prove using induction

63

𝑢

G

C

Theorem

By continuing the process from the vertex 𝑢 whose finish time 𝑢. 𝑓 is the largest excluding those in 𝐶, the algorithm returns the strongly connected components.

𝑢

G

T

C

(64)

1

3

2

6 5

4

(65)

65

1

3

2

6 5

4

(66)

Step 1: Run DFS on 𝐺 to obtain the finish time 𝑣. 𝑓 for 𝑣 ∈ 𝑉.

Step 2: Run DFS on the transpose of 𝐺 where the vertices 𝑉 are processed in the decreasing order of their finish time.

Step 3: output the vertex partition by the second DFS

Time Complexity:

(67)

67

(68)

Textbook Chapter 22.4 – Topological sort

68

(69)

69

1

2

3

5 4

6 1

2

3

5 4

6

(70)

Definition

a directed graph without any directed cycle

1

2

3

5 4

6

(71)

Taking courses should follow the specific order

How to find a course taking order?

71

計程 資料結構 演算法

計概 作業系統

計算機網路 微積分上 微積分下 機率

計組

(72)

Input: a directed acyclic graph 𝐺 = (𝑉, 𝐸)

Output: a linear order of 𝑉 s.t. all edges of 𝐺 going from lower- indexed nodes to higher-indexed nodes

a b d

f c e

a

b

d

f c

f b d e

a c e

(73)

73

Run DFS on the input DAG G.

Output the nodes in decreasing order of their finish time.

DFS(G)

for each vertex u in G.V u.color = WHITE

u.pi = NIL time = 0

for each vertex u in G.V if u.color == WHITE

DFS-VISIT(G, u)

DFS-Visit(G, u) time = time + 1 u.d = time

u.color = GRAY

for each v in G.Adj[u] (outgoing) if v.color == WHITE

v.pi = u

DFS-VISIT(G, v) u.color = BLACK

time = time + 1

u.f = time // finish time

(74)

Example Illustration

a

b

d

f c

e

a b d

f c e

1

2

3 4

5

6

(75)

Example Illustration

75

a

b

d

f c

e

f b d

a c e

1

2

3 4

6

5

(76)

Run DFS on the input DAG G.

Output the nodes in decreasing order of their finish time.

As each vertex is finished, insert it onto the front of a linked list

Return the linked list of vertices

DFS(G)

for each vertex u in G.V u.color = WHITE

u.pi = NIL time = 0

for each vertex u in G.V if u.color == WHITE

DFS-VISIT(G, u)

DFS-Visit(G, u) time = time + 1 u.d = time

u.color = GRAY

for each v in G.Adj[u]

if v.color == WHITE v.pi = u

DFS-VISIT(G, v) u.color = BLACK

time = time + 1

u.f = time // finish time

Time Complexity:

(77)

Proof

: suppose there is a back edge 𝑢, 𝑣

𝑣 is an ancestor of 𝑢 in DFS forest

There is a path from 𝑣 to 𝑢 in 𝐺 and 𝑢, 𝑣 completes the cycle

 : suppose there is a cycle 𝑐

Let 𝑣 be the first vertex in 𝑐 to be discovered and 𝑢 is a predecessor of 𝑣 in 𝑐

Upon discovering 𝑣 the whole cycle from 𝑣 to 𝑢 is WHITE

At time 𝑣. 𝑑, the vertices of 𝑐 form a path of white vertices from 𝑣 to 𝑢

By the white-path theorem, vertex 𝑢 becomes a descendant of 𝑣 in the DFS forest

Therefore, 𝑢, 𝑣 is a back edge

77

Lemma 22.11

A directed graph is acyclic  a DFS yields no back edges.

(78)

Proof

When 𝑢, 𝑣 is being explored, 𝑢 is GRAY and there are three cases for 𝑣:

Case 1 – GRAY

𝑢, 𝑣 is a back edge (contradicting Lemma 22.11), so 𝑣 cannot be GRAY

Case 2 – WHITE

𝑣 becomes descendant of 𝑢

𝑣 will be finished before 𝑢

Case 3 – BLACK

𝑣 is already finished

Theorem 22.12

The algorithm produces a topological sort of the input DAG. That is, if 𝑢, 𝑣 is a directed edge (from 𝑢 to 𝑣) of 𝐺, then 𝑢. 𝑓 > 𝑣. 𝑓.

(79)

79

(80)

Since cycle detection becomes back edge detection (Lemma 22.11), DFS can be used to test whether a graph is a DAG

Is there a topological order for cyclic graphs?

Given a topological order, is there always a DFS traversal

that produces such an order?

(81)

81

(82)

Course Website: http://ada17.csie.org Email: ada-ta@csie.ntu.edu.tw

82

Important announcement will be sent to @ntu.edu.tw mailbox

& post to the course website

參考文獻

相關文件

 Combine: find closet pair with one point in each region, and return the best of three

jobs

Greedy-Choice Property : making locally optimal (greedy) choices leads to a globally optimal

 From a source vertex, systematically follow the edges of a graph to visit all reachable vertices of the graph.  Useful to discover the structure of

 “Greedy”: always makes the choice that looks best at the moment in the hope that this choice will lead to a globally optimal solution.  When to

✓ Express the solution of the original problem in terms of optimal solutions for subproblems. Construct an optimal solution from

✓ Express the solution of the original problem in terms of optimal solutions for subproblems.. Construct an optimal solution from

◦ Action, State, and Reward Markov Decision Process Reinforcement Learning.