// #pragma GCC optimize("O3,unroll-loops")
// #pragma GCC target("avx2,bmi,bmi2,popcnt,lzcnt")
//#pragma GCC target("arch=skylake")
//#pragma GCC optimize("03,unroll-loops")
//#pragma GCC target("arch=skylake")
#include<bits/stdc++.h>
#define int long long
#define pb push_back
#define fi first
#define se second
#define el cout << '\n';
//#define float long double
#define MASK(i) (1LL << (i))
//#define c_bit(i) __builtin_popcountll(i) // dem so bit dang bat
#define BIT(x, i) ((x) & MASK(i)) // trang thai bit i trong x
//#define SET_ON(x, i) ((x) | MASK(i)) // b?t bit i trong x
//#define SET_OFF(x, i) ((x) & ~MASK(i)) // tat bit i trong x
using namespace std;
const int N = 1e5+5;
const int inf = 1e18;
const int mod = 1e9+7;
const int limit = 3e6;
const int S = 350;
int n;
void solve(){
cin>>n;
long double a[n+5],b[n+5];
long double pr1[n+5],pr2[n+5];
for(int i=1; i<=n;i++){
cin>>a[i];
cin>>b[i];
}
// for(int i=1; i<=n; i++){
// cout << a[i] << ' ' << b[i] << '\n';
// }
sort(a+1,a+1+n);sort(b+1,b+1+n);
// for(int i=1; i<=n; i++){
// cout << a[i] << ' ' << b[i] << '\n';
// }
// el;
pr1[0]=0;pr2[0]=0;
for(int i=1; i<=n; i++){
pr1[i] = pr1[i-1] + a[n-i+1];
pr2[i] = pr2[i-1] + b[n-i+1];
//cout << pr1[i] << ' ' << pr2[i] << '\n';
}
long double ans = 0;
// int l=1,r=n;
for(int i=0; i<=n; i++){
int l=0,r=n;
while(l<=r){
int m1 = l+(r-l)/3;
int m2 = r-(r-l)/3;
long double t1 = min((long double)pr1[i] - i -m1, (long double)pr2[m1]-i-m1);
long double t2 = min((long double)pr1[i] - i-m2 , (long double)pr2[m2]-i-m2);
// cout << t1 << ' ' << t2 << '\n';
if(t1<t2){
ans = max(ans,t2);
l=m1+1;
}
else {
ans = max(ans , t1);
r=m2-1;
}
}
}
cout << fixed << setprecision(4) << ans;
}
signed main(){
// freopen("MAXSQR.INP", "r", stdin);
//freopen("MAXSQR.OUT", "w", stdout);
ios_base::sync_with_stdio(false);cin.tie(0);cout.tie(0);
int t=1;
//cin>>t;
while(t--){
solve();el;
//tuab
}
}
// 23
// 10111
// 10110
// 10101
// 10011
// 00111
