#include <bits/stdc++.h>
using namespace std;
using ll = long long; using pii = pair<ll,ll>;
const ll Nm = 4e5+5;
ll f[Nm]; //forward
ll fsz[Nm]; //size
ll mval[Nm]; //max val of S
ll gt(ll x) {
if (f[x]==x) {
return x;
}
return (f[x]=gt(f[x])); //ackermann dsu impl
}
void fz(ll x, ll y) {
x = gt(x); y=gt(y);
if (fsz[x]>fsz[y]) {
swap(x,y);
}
f[x]=y;
fsz[y]+=fsz[x];
mval[y]=max(mval[y],mval[x]);
}
int main() {
ios_base::sync_with_stdio(false); cin.tie(0);
ll N,M,Q; cin >> N >> M >> Q;
ll S[N];
bool ispr[N];
vector<ll> adj[N];
vector<pii> ord;
ll ans = 0;
ll maxS = 0;
for (ll i=0;i<N;i++) {
ispr[i]=0;
cin >> S[i];
ans -= S[i];
maxS = max(maxS,S[i]);
f[i]=i;
fsz[i]=1;
mval[i]=S[i];
ord.push_back({S[i],i});
}
ans += maxS;
for (ll m=0;m<M;m++) {
ll a,b; cin >> a >> b;
a--; b--;
adj[a].push_back(b);
adj[b].push_back(a);
}
sort(ord.begin(),ord.end());
for (pii p1: ord) {
ll x = p1.second;
ispr[x]=1;
for (ll y: adj[x]) {
if (ispr[y]) {
//ans += (S[y]+S[x]-mval[gt(y)]);
ans += (S[y]+S[x]);
fz(x,y);
}
}
}
cout << ans << "\n";
}
//sum of mval[gt(y)] is just sum of all non-maximal S??!!?
