int N;
int A[1000][2];
int dp[1<<18][18];

inline int dist(int i, int j) {
    return A[i][1] - A[i][0] + A[j][0];
}

{
    rd(N);
    REP(i, N) {
        int a, b; rd(a, b);
        A[i][0] = a;
        A[i][1] = b;
    }
    
    fill(dp[1], dp[1<<N], int(1e9));
    
    REP(i, N) {
        dp[0][i] = 0;
        dp[i<<1][i] = 0;
    }
    
    REP(b, 1, 1 << N) {
        REP(i, N) {
            if (~b & (1 << i)) continue;
            REP(j, N) {
                if (b & (1 << j)) continue; 
                dp[b | (1 << j)][j] = min(dp[b | (1 << j)][j],
                    max(dp[b][i], dist(i, j))
                );
            }
        }
    }
    
    int best = 1e9;
    REP(i, N){
        best = min(best, dp[(1<<N)-1][i]);
    }
    
    wt(best);
    
}