#pragma GCC optimize("Ofast,unroll-loops")
#include<bits/stdc++.h>

using namespace std;
using Int = long long;
typedef pair<int,int> Pi; typedef pair<Int,Int> Pl;
constexpr int mod = 1e9+7;
constexpr int INF32 = numeric_limits<int>::max();
constexpr Int INF64 = numeric_limits<Int>::max();
#define END {cout<<ans<<'\n'; return 0;}
#define ALL(v) (v).begin(),(v).end()
#define fi first
#define se second
#define mp make_pair
#define Pr(t) priority_queue<t>
#define gPr(t) priority_queue<t,vector<t>,greater<t>>
#define V(t) vector<t>
#define rer(i,l,r) for(int(i)=(l); (i)<(r); ++(i))
#define rep(i,n) rer(i,0,n)
#define gnr(i,l,r) for(int (i)=(r)-1; (i)>=(l); --(i))
#define eb emplace_back
#define pri(a) cout<<(a)<<'\n'
#define pri2(a,n) {rep(i,n-1)cout<<a[i]<<' '; if(n)cout<<a[n-1]<<'\n';}
#define prip(p) cout<<(p).fi<<' '<<(p).se<<'\n'
#define priend(a) {cout<<(a)<<'\n'; return 0;}
template<class T> inline bool cmax(T& a, T b) { if (a < b) { a = b; return 1; } return 0; }
template<class T> inline bool cmin(T& a, T b) { if (a > b) { a = b; return 1; } return 0; }

int n,m,_,x,y,q;
string s,sb;
bool ok;
char cf,cb;

int main(){
  cin.tie(nullptr); ios::sync_with_stdio(0);
  //cout<<fixed<<setprecision(20);
  
  cin>>n>>m; int a[n]; rep(i,n)cin>>a[i];
  double ans = 0;
  rep(i,pow(m,n)){
    V(int) v[m]; x=i;
    rep(j,n){
      v[x%m].eb(a[j]); x/=m;
    }
    double ave[m]={};
    rep(j,m){
      for(auto e:v[j])ave[j]+=e;
      ave[j]/=v[j].size();
    }
    cmax(ans, *max_element(ave, ave+m)-*min_element(ave, ave+m));
  }
  pri(floor(ans+0.5));
}