因为精度问题而一直错，所以干脆都乘一个c，最后再除，这样大大减少了误差。
#include <stdio.h>
#include <string.h>
#include <algorithm>
int a[20];
inline int abs(int x)
{
if(x > 0) return x;
return -x;
}
int main()
{
// freopen("in", "r", stdin);
int c, s, n, cas = 1;
while(~scanf("%d %d", &c, &s))
{
memset(a, 0, sizeof(a));
n = (c << 1) - s;
int ave = 0;
for(int i = n; i < (c << 1); i++)
{
scanf("%d", &a[i]);
ave += a[i];
}
// printf("ave = %lf\n", ave);
std::sort(a, a + (c << 1));
printf("Set #%d\n", cas++);
double sum = 0;
for(int i = 0; i < c; i++)
{
printf(" %d:", i);
sum += abs(c * a[i] + c * a[(c << 1) - i - 1] - ave);
if(a[i]) printf(" %d", a[i]);
if(a[(c << 1) - i - 1]) printf(" %d", a[(c << 1) - i - 1]);
printf("\n");
}
printf("IMBALANCE = %.5lf\n\n", sum / c);
}
return 0;
}