SHORTEST PATH

7.6 SHORTEST PATHS

7.6.1 Find the shortest paths from a source vertex to all other vertices in G
（ Single-Source Shortest-Paths Problem ）
1. Questions Raised
A weighted graph G=(V,E) as Transportation network,
• V= vertices: cities
• E= edges: the transport links between cities
• W(U,V)= weight:the length of the line or the time or cost of
transportation along the line
Problem:THE PATH that MINIMIZES the sum of the weights of EACH edge
from a vertex to another vertex——The SHORTEST PATH

2. Dijkstra Algorithm
Generate shortest paths in increasing path length order
 7.6 SHORTEST PATHS
7.6.1 Find the shortest paths from a source vertex to all other vertices in G
（ Single-Source Shortest-Paths Problem ）

3. Assistant Array dist

• dist[i], i=1,...,n stores the current length of the shortest path from
source V0 to vertex i,
• The initial value of dist[i] is the weight of edge (v0, i); If edge (v0, i)
doesn't exist, it's set as infinity 
4.implementation
Let a set to store the vertices of the generated shortest path
initially, there is just the source vertex v0 in U
15
2 4
5
7
1 9
5 13
4 8
9
3 3

20 6

i 1 2 3 4 5 6 7 8
dist[ 5 4 ∞ 9 20 ∞ ∞
i]
4.implementation
Let a set to store the vertices of the generated shortest path
initially, there is just the source vertex v0 in U
15 1st Path ： One of
2 4
5 the paths from 1 to
7
1 9 2，5，3，6
5 13
4 8
9
3 3

20 6

i 1 2 3 4 5 6 7 8
dist[ 5 4 ∞ 9 20 ∞ ∞
i]
4.implementation
Let a set to store the vertices of the generated shortest path
initially, there is just the source vertex v0 in U
15 1st Path ： One of
2 4
5 the paths from 1 to
7
1 9 2,5,3,6
5 13 Add i that dist[i]is
4 8 minimum to U
9
3 i=3
3
20 6

i 1 2 3 4 5 6 7 8
dist[ 5 4 ∞ 9 20 ∞ ∞
i]
4.implementation
Let a set to store the vertices of the generated shortest path
initially, there is just the source vertex v0 in U
15 1st Path ： One of
2 4
5 the paths from 1 to
7
1 9 2,5,3,6
5 13
4
Add i that dist[i]is
9 8
minimum to U i=3
3 3
7 13 After adding 3 to U,
20 6 update some dist[i]
2nd Path ： One of
the paths from 1 to
i 1 2 3 4 5 6 7 8 2,5,8,6.
dist[ 5 4 ∞ 9 20 ∞ ∞
i]
4.implementation
Let a set to store the vertices of the generated shortest path
initially, there is just the source vertex v0 in U
15 1st Path ： One of
2 4
5 the paths from 1 to
7
1 9 2,5,3,6
5 13
4
Add i that dist[i]is
9 8
minimum to U i=3
3 3
After adding 3 to U,
20 6 update some dist[i]
2nd Path ： One of
the paths from 1 to
i 1 2 3 4 5 6 7 8 2,5,8,6.
dist[ 5 4 ∞ 9 7 ∞ 13
i]
Theorem ： If the next shortest path goes to vertex x, then all
the intermediate points/vertices from the source point to x
must be in U

Proof (contradiction): If not, then at least one intermediate

point u does not belong to U

vo
U x
u
Theorem ： If the next shortest path goes to vertex x, then all
the intermediate points/vertices from the source point to x
must be in U

Proof (contradiction): If not, then at least one intermediate

point u does not belong to U

vo
U x
u

Then,there must be:

dist[u]+ w(u,x) > dist[x] ， According to Algo., the shorter
path has generated, that is u belongs to U
4.implementation
Let a set to store the vertices of the generated shortest path
initially, there is just the source vertex v0 in U
15
2 4
5
7
1 9
5 13
4 8
9 Then,there must be:
3 dist[u]+ w(u,x) > dist[x] ，
3
According to Algo., the
20 6 shorter path has
generated, that is u
belongs to U
i 1 2 3 4 5 6 7 8
dist[ 5 4 ∞ 9 7 ∞ 13
i]
If the next shortest path goes to vertex x, then all the intermediate
points from the source point to x must be in U
vo
Proof (contradiction): If not, then at least one U x
intermediate point u does not belong to U
u
6.
G:Dijkstra Algorithm
adjacency matrix,
Initialize set U and assistant array dist[] ，   13 8  30  32
 
Source vertex: vo ; |V|= n 。      9 7
g[i,j]: weight of edge(i, j) ；    5   
 
if (i, j) does not exist, g[i,j] is set as
g      6  
infinity ∞;  
     2 
1. dist[i]= g[v0,i] i=1,2,….,n  
      17 
2. U=[v0];  
      
3. Repeat following steps (n-1) times
① Select v that: dist[v]=min{ dist[u]|
uU}
② Add v into U: U=U{v}
③ Update dist[w] that wU:
dist[w]=min{dist[w],dist[v]+g[v,w]|
1 wU} 32
8 13 2
3
30 9 7
5 7
4
6 17
6 2
5
Gn*n
6. : adjacency
Dijkstra matrix,
Algorithm
  13 8  30  32
Initialize set U and assistant array dist[] ，  
Source vertex: vo ; |V|= n 。      9 7
g[i,j]: weight of edge(i, j) ；    5   
 
if (i, j) does not exist, g[i,j] is set as g      6  
infinity ∞;  
     2 
1. dist[i]= g[v0,i] i=1,2,….,n  
      17 
2. U=[v0];  
      
3. Repeat following steps (n-1) times
① Select v that: dist[v]=min{ dist[u]| i 1 2 3 4 5 6 7
uU} dist ∞ 1 8 ∞ 3 ∞ 32
② Add v into U: U=U{v} 3 0
③ Update dist[w] that wU: i 1 2 3 4 5 6 7
dist[w]=min{dist[w],dist[v]+g[v,w]| dist ∞ 1 8 13 3 ∞ 32
1 wU} 32 3 0
8 13 i 1 2 3 4 5 6 7
2
3 dist ∞ 1 8 13 3 22 20
30 9 7 3 0
5 7 i 1 2 3 4 5 6 7
4 dist ∞ 1 8 13 1 22 20
6 17
6 2 3 9
i 1 2 3 4 5 6 7 i 1 2 3 4 5 6 7
5
dist ∞ 1 8 13 1 21 20 dist ∞ 1 8 13 1 21 20
 7. Dijkstra Algorithm in C
since the index (subscript) of C/C++ starts from 0, we make
some changes and using NUM denoting the number of
vertices of G
#define NUM
void shortpath_dij(g[][NUM],v0) //v0: source ， w:1~NUM
{
for(i=0;i<NUM;i++){
set[i]=0;
dist[i]=g[v0-1][i];}
set[v0-1]=1;
for(i=1;i<NUM;i++)
{ min=MAXINT;
for(w=0;w<NUM;w++)
if(set[w]= =0 && dist[w]<min)
{ v=w; min=dist[w]; } Time
set[v]=1; Complexity:
for(w=0;w<NUM;w++) O(n2)
if(set[w]= =0 && dist[v]+g[v][w]<dist[w])
dist[w]=dist[v]+g[v][w];
}//for i
} //shortpath_dij
#include<iostream.h> int search(char e);// 查找顶点在顺序表中的下表
#define max 20 int indegree(char e);// 计算 e 的入度
class Arcbox{ int outdegree(char e);// 计算 e 的出度
public: };
int tailvex,headvex; int OLgraph::search(char e){
Arcbox *hlink,*tlink;}; for(int i=0;i<vexnum;i++)
class VexNode{ if(xlist[i].data==e) return i;
public: return -1;}
char data; int OLgraph::indegree(char e){
Arcbox int count=0;
*firstin,*firstout;};
int i=search(e);
class OLgraph{
Arcbox *p=xlist[i].firstout;
public:
while(p){ count++;
VexNode xlist[max];
p=p->tlink;}
int vexnum,arcnum;
return count;
void CreateDG();
}
// 创建图 ( 十字链表 )
void OLgraph::CreateDG(){ p=new Arcbox;
char v,w; p->tailvex=i;p->headvex=j;
int i,j,k; p->hlink=xlist[j].firstin;
Arcbox *p; p->tlink=xlist[i].firstout;
cin>>vexnum>>arcnum; xlist[j].firstin=p;
for( i=0;i<vexnum;i++){ xlist[i].firstout=p;}
cin>>xlist[i].data; }
xlist[i].firstin=NULL; int OLgraph::outdegree(char e){
xlist[i].firstout=NULL;} int count=0;
for( k=0;k<arcnum;++k){ int i=search(e);
lab:cin>>v>>w; Arcbox *p=xlist[i].firstin;
i=search(v); while(p){
j=search(w); count++;
if(i==-1||j==-1){ p=p->hlink;}
cout<<" 查无此点 "; return count;
goto lab;} }
void main(void){
char ch;
OLgraph G;
G.CreateDG();
cin>>ch;
cout<<G.indegree(ch)<<endl;
cout<<G.outdegree(ch)<<endl;
}