动态规划算法学习十例之十

凸多边形最优三角剖分

   一凸8边形P的顶点顺时针为{v1,v2,… ,v8},任意两顶点间的线段的权重由矩阵D给出。
若vi与vj是P上不相邻的两个顶点,则线段vivj 称为P的一条弦。求P的一个弦的集合T,使得T中所有的弦恰好将P 分割成互不重迭的三角形,且各三角形的权重之和为最小(一个三角形的权重是其各边的权重之和)。

/*
    int graph[][]={
      {0, 14, 25, 27, 10, 11, 24, 16},
      {14, 0, 18, 15, 27, 28, 16, 14},
      {25, 18, 0, 19, 14, 19, 16, 10},
      {27, 15, 19, 0, 22, 23, 15, 14},
      {10, 27, 14, 22, 0, 14, 13, 20},
      {11, 28, 19, 23, 14, 0, 15, 18},
      {24, 16, 16, 15, 13, 15, 0, 27},
      {16, 14, 10, 14, 20, 18, 27, 0}
    };*/

解答:题目中顶点坐标编号从1开始,为了方便编程,将顶点从0开始,顶点的编号变为0到7。定义t[i][j],0=<i<j<n为凸多边形{Vi,Vi,…Vj}的最优三角剖分所对应的权函数值,退化的两点多边形{Vi,Vi+1}的权值为0。要计算凸n边形的最优权值为t[0][n-1]。

由于退化的两点多边形{Vi,Vi+1}的权值为0,t[i][i]=0。最优子结构的性质,t[i][j]的值是t[i][k]的值加上t[k][j]的值,再加上三角形ViVkVj的权值,其中,i<k<=j-1。K的位置有j-i-1个。因此可以在这j-i-1个未知中选出使t[i][j]值达到最小的位置,递归式为: t[i][j]=0 (j-i<=1)
t[i][j]=t[i][k]+t[k][j]+w(i,k,j) (j-i>=2)

动态规划算法学习十例之十
public class TestTriangulation{
  
    int t[][];
    int s[][];
    int graph[][];
    int n;

    public TestTriangulation(int n,int[][] graph){
      this.n=n;
      this.graph=graph;
      t=new int[n][n];
      s=new int[n][n];
    }

    private int w(int m, int n, int p){
      return graph[m][n] + graph[m][p] + graph[n][p];
    }

    private void sp(int m, int n)//输出最优剖分需要添加的线段
    {
     if (n-m<=2)
       return;
     if(s[m][n]-m>=2){
       System.out.printf("V%d-V%d\n", m, s[m][n]);
       sp(m, s[m][n]);
     }
     if (n-s[m][n]>=2){
       System.out.printf("V%d-V%d\n", s[m][n], n);
        sp(s[m][n], n);
      }
   }

   public int  minWeight(){
    int j, u;
    for (int r=2; r<=n-1; r++)
     for (int i=0; i<n-r; i++){
      j = i + r;
      t[i][j] =t[i][i+1] + t[i+1][j] +w(i, i+1, j);
      s[i][j] = i+1;
      for (int k=i+2; k<i+r; k++){
        u = t[i][k] + t[k][j] + w(i, k, j);
        if (u<t[i][j]){
          t[i][j] = u;
          s[i][j] = k;
        }
     }
   }
   return t[0][n-1];
  }


    public static void main(String args[]){
    /*
    int graph[][]={
      {0, 14, 25, 27, 10, 11, 24, 16},
      {14, 0, 18, 15, 27, 28, 16, 14},
      {25, 18, 0, 19, 14, 19, 16, 10},
      {27, 15, 19, 0, 22, 23, 15, 14},
      {10, 27, 14, 22, 0, 14, 13, 20},
      {11, 28, 19, 23, 14, 0, 15, 18},
      {24, 16, 16, 15, 13, 15, 0, 27},
      {16, 14, 10, 14, 20, 18, 27, 0}
    };*/

     int[][] graph = { { 0, 2, 2, 3, 1, 4 }, { 2, 0, 1, 5, 2, 3 }, { 2, 1, 0, 2, 1, 4 }, 
        { 3, 5, 2, 0, 6, 2 }, { 1, 2, 1, 6, 0, 1 }, { 4, 3, 4, 2, 1, 0 } };

    TestTriangulation tri=new TestTriangulation(6,graph);
    System.out.printf("最优剖分总权值\n");
    System.out.printf("%d\n",tri.minWeight());
    System.out.printf("剖分时添加的线段\n");
    tri.sp(0,5);
   }
}


动态规划算法学习十例之十

运行:
C:\java>java   TestTriangulation
最优剖分总权值
24
剖分时添加的线段
V0-V4
V1-V4
V2-V4

你可能感兴趣的:(java,数据结构,编程,算法)