我2018最早的ac程序就没有通过我自己做的数据，就是因为浮点误差导致的

> It was Accepted in uva,but why WA here?
> 
> #include<stdio.h>
> #include<math.h>
> int main()
> {
>     int a,b,i,j,m=0;
>     double r,t,t1,t2,near;
>     while(scanf("%d %d",&a,&b)==2)
>     {
>         if(m)printf("\n");
>         r=(double)a/b;
>         j=1;near=r;
>           i=(int)(r);
>           t1=(double)i;t2=(double)(i+1);
>           if(fabs(t1-near)<fabs(t2-near))
>             {printf("%d/%d\n",i,j);near=fabs(t1-r);}
>           else 
>             {printf("%d/%d\n",i+1,j);near=fabs(t2-r);}
>         j++;
>         while(j<=b)
>         {
>           i=(int)(r*j);
>           t1=(double)i/j;t2=(double)(i+1)/j;
>           if(fabs(t1-near)<fabs(t2-near)&&fabs(t1-r)<near)
>             {printf("%d/%d\n",i,j);near=fabs(t1-r);}
>           else if(fabs(t2-r)<near)
>             {printf("%d/%d\n",i+1,j);near=fabs(t2-r);}
>           j++;
>         }    
>         m++;
>     }    
>     return 0;
> }    

• 数据没问题 (44) hawk 2005-04-21 12:54:08
• 看程序写法了，完全可以做到两个都acc的 (44) hawk 2005-04-21 12:54:32
  • 那么就对分子分别取ceil和floor各算一个，取一个好的，至少我就是这么做的 (0) frkstyc 2005-04-21 13:01:00