#include<bits/stdc++.h>
using namespace std;
#define task "CC"
#define pb push_back
#define fi first
#define se second
#define for1(i,j,k) for(int i=j;i<=k;i++)
#define for2(i,j,k) for(int i=j;i>=k;i--)
#define for3(i,j,k,l) for(int i=j;i<=k;i+=l)
#define bit(n,i) ((n>>i)&1)
#define all(x) x.begin(),x.end()
//#pragma GCC optimize("Ofast,unroll-loops")
//#pragma GCC target("avx,avx2,bmi,bmi2,sse,sse2,sse3,ssse3,sse4,popcnt")
//#define int long long
typedef long long ll;
typedef pair<int,int> pii;
typedef double ld;
typedef pair<ld,ld> pdd;
typedef pair<ll,ll> pll;
const ll maxn=1e7+1;
const ll N=1e5+2;
int n,nxt[maxn+1],par[N],x[N];
vector<pii> Edge[maxn];
int Find(int v) {
return (par[v] < 0) ?v: (par[v] = Find(par[v]));
}
bool Union(int x, int y) {
if ((x = Find(x)) == (y = Find(y)))
{
return false ;
}
if (par[y] < par[x])
{
swap(x, y);
}
par[x] += par[y];
par[y] = x;
return true;
}
void sol()
{
cin >> n;
int mx=-1;
for (int i=1;i<=n;i++)
{
cin >> x[i];
mx=max(mx,x[i]);
par[i]=-1;
}
sort(x+1,x+1+n);
for1(i,1,n)
{
if (nxt[x[i]]==0) nxt[x[i]]=i;
}
int ans=0;
nxt[(int)mx+1]=-1;
int cnt=0;
for2(i,mx,1)
{
if (nxt[i]==0) nxt[i]=nxt[i+1];
//cerr<< i<< ' '<<nxt[i]<<'\n';
}
// for (int i=1;i<=mx;i++)
// {
// //cerr<<"wtf\n";
// if (nxt[i]==i)
// {
// if (nxt[i+1]==-1) continue;
// Edge[nxt[i+1]-i].pb({nxt[i+1],i});
// for (int j=2*i;j<=mx;j+=i)
// {
// if (nxt[j]==-1) break;
// Edge[nxt[j]-j].pb({nxt[j],i});
// }
// }
// }
for1(k,1,n)
{
if (x[k]==x[k-1]) continue;
int i=x[k];
if (nxt[i+1]==-1) continue;
Edge[x[nxt[i+1]]-i].pb({nxt[i+1],k});
for (int j=2*i;j<=mx;j+=i)
{
if (nxt[j]==-1) break;
Edge[x[nxt[j]]-j].pb({nxt[j],k});
}
}
for (int i=0;i<=mx;i++)
{
for (auto x:Edge[i])
{
//cerr<<i<<' '<<x.se<<' '<<x.fi<<'\n';
if (!Union(x.fi,x.se)) continue;
ans+=i;
}
Edge[i].clear();
}
cout << ans;
}
int32_t main()
{
ios_base::sync_with_stdio(false);
cin.tie(NULL);
if (fopen(task".inp","r"))
{
freopen(task".inp","r",stdin);
freopen(task".out","w",stdout);
}
sol();
}
