Scribd
Upload
24 views

Week11-Chap5-Graph-Algorithms Sum Path On Tree

Uploaded by

Ha Minh Duc
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
24 views

Week11-Chap5-Graph-Algorithms Sum Path On Tree

Uploaded by

Ha Minh Duc
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 14

APPLIED ALGORITHMS

APPLIED ALGORITHMS
DEPTH FIRST SEARCH (DFS) AND APPLICATIONS

3
CONTENTS

• The longest path on a tree (Đường đi dài nhất trên cây)


• Total path length on the tree (Tổng đường đi trên cây)

4
THE LONGEST PATH ON THE TREE

• Given a tree T = (V, E), each edge (u,v) has weight w(u,v). Find the path with the
longest length on T (the length of the path is the sum of weight on all edges of the
path).
• Denote A[v] is the set of vertices adjacent to vertex v on T
• The algorithm is based on depth-first-search (DFS)
• Choose an arbitrary vertex s on T
• Perform DFS(s) to find vertex x farthest from s
• Perform DFS(x) to find the vertex y that is farthest from x
• The path from x to y found will be the longest path on T

5
THE LONGEST PATH ON THE TREE

Init(V, A) { LongestPathOnTree(V, A){


for v in V do d[v] = -1; Init(V, A);
} s = select a node in V;
DFS(u) { DFS(s);
for x in A[u] do { x = select u in V such that d[u] is maximal;
if d[x] < 0 then { Init(V, A);
d[x] = d[u] + w(u,x); DFS(x);
DFS(x); y = select u in V such that d[u] is maximal;
} P = unique path between x and y in T;
} return P;
} }

6
THE LONGEST PATH ON THE TREE

• The complexity: O(|V| + |E|)

7
TOTAL PATH LENGTH ON THE TREE

• Given a tree T = (V, E), each edge (u,v) has weight w(u,v). Vertex set V includes n vertices.
• Denote:
• A[v]: is the set of vertices adjacent to vertex v on T
• c(u,v) is the length of the unique path between two vertices u and v on T
• f(u): total path length from other vertices to u on T: f(u) = σ𝑣𝑉 𝑐(𝑣, 𝑢)
• Find f(u) for every u  V

8
TOTAL PATH LENGTH ON THE TREE

• Choose an arbitrary vertex s on T as the root, perform DFS on T starting from s:


• p(u): parent vertex of u (the vertex from which the algorithm visits u)
• d(u): total path length from descendant vertices of u to u
• N(u): number of descendant vertices of u (including vertex u)

9
TOTAL PATH LENGTH ON THE TREE

• DFS1(u): depth-first search in the first phase


• Purpose: calculate d(x) and N(x) for all vertices x that are descendants of u
• When DFS1(u) is completed, d(u) is calculated and it will be used to calculate d(p(u))
• Do: for each vertex v A[u]:
• Call DFS1(v)
• Update: d(u) = d(u) + N(v)*d(v)
• N(u) = N(u) + N(v)

10
TOTAL PATH LENGTH ON THE TREE

• DFS1(u): depth-first search in the first phase


• Purpose: calculate d(x) and N(x) for all vertices x that are descendants of u
• When DFS1(u) is completed, d(u) is calculated and it will be used to calculate d(p(u))
• Do: for each vertex v A[u]:
• Call DFS1(v)
• Update: d(u) = d(u) + d(v) + N(v)*w(u,v)
• N(u) = N(u) + N(v)
• DFS2(u): depth-first search in the second phase
• Purpose: When DFS2(u) is called, f(u) has been already calculated and we will calculate f(v) for
each vertex v being a child of u
• Do: for each vertex v  A[u] not has been visited
• F = f(u) – (d(v) + w(u,v)*N(v))
• f(v) = F + d(v) + w(u,v)*(n – N(v))
• Call DFS2(v)

11
TOTAL PATH LENGTH ON THE TREE

DFS1(u){ DFS2(u){
for v in A[u] do { for v in A[u] do {
if p(v) = 0 then { if p(v) = 0 then {
p(v) = u; F = f(u) – (d(v) + N(v)*w(u,v));
DFS1(v); f(v) = F + d(v) + w(u,v)*(n – N(v));
d(u) = d(u) + d(v) + N(v)*w(u,v); p(v) = u; DFS2(v);
N(u) = N(u) + N(v); }
} }
} }
} Phase2(){
Phase1(){ for v in V do { p(v) = 0; }
for v in V do { f(1) = d(1); p(1) = 1; DFS2(1);
p(v) = 0; d(v) = 0; N(v) = 1; f(v) = 0; }
} Main(){
p(1) = 1; DFS1(1); Phase1(); Phase2();
} }

12
TOTAL PATH LENGTH ON THE TREE

• The complexity: O(|V| + |E|)

13
THANK YOU !

14

You might also like