FM算法组合估计

#include <stdio.h>
#include <math.h>
#include <stdlib.h>
#include <time.h>

#define NUM (int)pow(2.0,25)+1

int length=0;//计算长度
int data[350000];//存放数据流

struct node
{
    int data;
    struct node* next;
};
struct node* creat()//创建链表
{
    int x;
    struct node *tail,*p,*head;
    head=(struct node *)malloc(sizeof(struct node));
    head->next=NULL;
    tail=head;
    //FILE *fp=fopen("./stream_for_fm.txt","r");
    FILE *fp=fopen("./stream_sample_fm.txt","r");//样本
    if(fp==NULL)
    {
        printf("file open failed!\n");
        exit(0);
    }
    while(!feof(fp))
    {
        fscanf(fp,"%d",&x);
        p=(struct node *)malloc(sizeof(struct node));
        p->data=x;
        tail->next=p;
        p->next=NULL;
        tail=p;
        data[length++]=x;
    }
    printf("链表创建成功!\n");
    return head;
}
/*排序链表*/
struct node *sort(struct node *head)
{
    struct node *p,*q;
    for(p=head->next;p!=NULL;p=p->next)
    {
        for(q=p->next;q!=NULL;q=q->next)
        {
            if(p->data > q->data)
            {
                int tmp=p->data;
                p->data=q->data;
                q->data=tmp;
            }
        }
    }
    printf("链表排序完成！\n");
    return head;
}

/*计算有多少不同元素*/
int different_node_count(struct node *head)
{
    struct node *p,*q,*tmp;
    int count_equal=0,count_total=0;
    tmp=NULL;

    for(p=head->next;p!=NULL;p=p->next)
    {
        for(q=p->next;q!=NULL;q=q->next)
        {
            if(p->data == q->data)
            {
                count_equal++;//相同元素加
                tmp=q;//记录最后一个相同的元素位置
            }
            else
            {
                break;
            }
        }
        if(tmp!=NULL)
        {
            p=tmp;
            tmp=NULL;
        }
        //count_total++;//统计全部节点的个数
    }
    printf("不同节点的个数%d\n节点总数：%d\n",length-count_equal,length);
    return length-count_equal;
}
//计算尾数
int get_tailnum(long long a)
{
    int i=0,count =0;
    while(a!=0)
    {
        if(0 == a%2)
        {
            count++;
        }
        else
        {
            break;
        }
        a=a/2;
        i++;
    }
    return count;
}    

int main()
{
    struct node *head;
    int only_value;
    int M=4,L=4;
    int i,j,k,x;
    int circle=20;
    long long h,a,b;
    int R[20][20];//尾数
    int max_tailnum=0,tmp;//最大尾长
    double ave_tail[20]={0};//最大尾数的平均数
    int mid_tail;//平均尾数的中位数
    int estimate[30];//存入估计值
    double error;//误差
    long long sum=0;

    head=creat();
    head=sort(head);
    only_value=different_node_count(head);
    srand(time(0));
    for(x=0;x<circle;x++)
    {
        for(i=0;i<M;i++)//分组M
        {
            for(j=0;j<L;j++)//每组L个hash
            {
                a=rand()%NUM;
                b=rand()%NUM;
                max_tailnum=0;
                for(k=0;k<length;k++)
                {
                    tmp=get_tailnum(a*data[k]+b);
                    if(max_tailnum<tmp)
                    {
                        max_tailnum=tmp;
                    }
                }
                R[i][j]=max_tailnum;
            }
            for(j=0;j<L;j++)
            {
                ave_tail[i]+=R[i][j];
            }
            ave_tail[i]/=L;//每组的平均值
        }

        if(M%2==0)
        {
            mid_tail=(ave_tail[M/2]+ave_tail[M/2+1])/2;
        }
        else
        {
            mid_tail=ave_tail[M/2];
        }
        
        estimate[x]=(int)pow(2.0,mid_tail);

        //ave_tail尾长的平均值清零
        for(int i=0 ;i < 20 ;i ++)
        {
            ave_tail[i]=0;
        }
    }
    int temp=0;
    printf("估计元素个数");
    for(i=0;i<circle;i++)
    {
        temp+=estimate[i];
        //printf("%8d",estimate[i]);
        sum+=(estimate[i]-only_value)*(estimate[i]-only_value);
    }
    printf("%lf",temp/(double)circle);
    putchar('\n');
    error=sqrt(sum/(double)only_value);
    printf("误差（方差）:%lf\n",error);
    system("pause");
    return 0;
}