////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
//
// Solution
// Webbly, 27.01.2023
//
//
// Arsen ne katai
//
//
///////////////////////////////////////////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////////////////////////////////////////
#include<bits/stdc++.h>
/**
#include <algorithm>
#include <iostream>
#include <vector>
#include <queue>
#include <cmath>
#include <map>
#include <set>
#include <time.h>
*/
//#pragma GCC optimize("O3")
//#pragma GCC optimize("fast-loops")
//#pragma comment(linker, "/stack:200000000")
//#pragma GCC optimize("Ofast")
//#pragma GCC target("sse,sse2,sse3,ssse3,sse4,popcnt,abm,mmx,avx,tune=native")
/**
#pragma GCC optimize("-funsafe-loop-optimizations")
#pragma GCC optimize("-funroll-loops")
#pragma GCC optimize("-fwhole-program")
#pragma GCC optimize("Ofast,no-stack-protector")
#pragma GCC optimize("-fthread-jumps")
#pragma GCC optimize("-falign-functions")
#pragma GCC optimize("-falign-jumps")
#pragma GCC optimize("-falign-loops")
#pragma GCC optimize("-falign-labels")
#pragma GCC optimize("-fcaller-saves")
#pragma GCC optimize("-fcrossjumping")
#pragma GCC optimize("-fcse-follow-jumps")
#pragma GCC optimize("-fcse-skip-blocks")
#pragma GCC optimize("-fdelete-null-pointer-checks")
#pragma GCC optimize("-fdevirtualize")
#pragma GCC optimize("-fexpensive-optimizations")
#pragma GCC optimize("-fgcse")
#pragma GCC optimize("-fgcse-lm")
#pragma GCC optimize("-fhoist-adjacent-loads")
#pragma GCC optimize("-finline-small-functions")
#pragma GCC optimize("-findirect-inlining")
#pragma GCC optimize("-fipa-sra")
#pragma GCC optimize("-foptimize-sibling-calls")
#pragma GCC optimize("-fpartial-inlining")
#pragma GCC optimize("-fpeephole2")
#pragma GCC optimize("-freorder-blocks")
#pragma GCC optimize("-freorder-functions")
#pragma GCC optimize("-frerun-cse-after-loop")
#pragma GCC optimize("-fsched-interblock")
#pragma GCC optimize("-fsched-spec")
#pragma GCC optimize("-fschedule-insns")
#pragma GCC optimize("-fschedule-insns2")
#pragma GCC optimize("-fstrict-aliasing")
#pragma GCC optimize("-fstrict-overflow")
#pragma GCC optimize("-ftree-switch-conversion")
#pragma GCC optimize("-ftree-tail-merge")
#pragma GCC optimize("-ftree-pre")
#pragma GCC optimize("-ftree-vrp")
#pragma GCC target("avx")
*/
#define ll long long
#define kebzaro ios_base::sync_with_stdio(0),cin.tie(0),cout.tie(0);
#define pb push_back
#define mp make_pair
#define all(x) x.begin(),x.end()
#define flush cout.flush()
using namespace std;
const ll mod = (ll)1e9 + 7, mod3 = 998244353, inf = (ll)2e15, P = 31;
void judge(string name){
if (name.size()){
freopen((name + ".in").c_str(), "r", stdin);
freopen((name + "out").c_str(), "w", stdout);
}
}
ll rnd(){
ll x = rand();
return rand() ^ x;
}
ll binpow (ll a, ll b){
ll ans = 1;
while(b){
if (b & 1){
ans *= a;
ans %= mod;
}
b >>= 1;
a *= a;
a %= mod;
}
return ans;
}
ll gcd(ll a, ll b){
return (b ? gcd(b, a % b) : a);
}
ll lcm(ll a, ll b){
return a / gcd(a, b) * b;
}
struct tr{
ll four, seven, inc, dec;
};
struct T{
ll l, r, val, ind;
};
ll n, m, k, a[500005], b[500005], dp[505][505];
string s;
double x[500005], y[500005];
void sarsen(){
cin >> n;
for (ll i = 1; i <= n; i++){
cin >> x[i] >> y[i];
}
sort (x + 1, x + n + 1);
sort (y + 1, y + n + 1);
reverse(x + 1, x + n + 1);
reverse(y + 1, y + n + 1);
double ans = 0;
ll l = 0, r = 0;
double sum1 = 0, sum2 = 0;
for (ll i = 1; i <= n * 2; i++){
if (sum1 < sum2 && l < n || r == n){
sum1 += x[++l];
}
else{
sum2 += y[++r];
}
ans = max(ans, min(sum1, sum2) - i);
}
cout << fixed;
cout.precision(4);
cout << ans;
}
int main(){
kebzaro
judge("");
ll TT = 1, tests = 1;
//cin >> TT;
while(TT--){
//cout << "Case " << tests << ": ";
sarsen();
//tests++;
}
return 0;
}
/**
2 6 4 3 1 5
1 5 4 3 2 6
*/
