/*
#pragma GCC optimize("Ofast,unroll-loops")
#pragma GCC target("avx2,fma,bmi,bmi2,sse4.2,popcnt,lzcnt")
*/
#include <bits/stdc++.h>
#define taskname ""
#define all(x) x.begin(), x.end()
#define rall(x) x.rbegin(), x.rend()
#define i64 long long
#define isz(x) (int)x.size()
using namespace std;
template<bool ALLOW_NON_PREFIX_QUERY, class T, class F, class I>
struct fenwick_tree{
int n;
vector<T> data;
F TT;
T T_id;
I Tinv;
fenwick_tree(F TT, T T_id, I Tinv): TT(TT), T_id(T_id), Tinv(Tinv){ }
fenwick_tree &operator=(const fenwick_tree &fw){
n = fw.n;
data = fw.data;
}
// O(n)
void build(int n){
assert(n >= 0);
this->n = n;
data.assign(n, T_id);
}
// O(n)
void build(int n, T x){
assert(n >= 0);
this->n = n;
data.assign(n, x);
for(auto i = 1; i <= n; ++ i) if(i + (i & -i) <= n) data[i + (i & -i) - 1] = TT(data[i + (i & -i) - 1], data[i - 1]);
}
// O(n)
template<class U>
void build(const vector<U> &a){
n = (int)a.size();
data.resize(n);
copy(a.begin(), a.end(), data.begin());
for(auto i = 1; i <= n; ++ i) if(i + (i & -i) <= n) data[i + (i & -i) - 1] = TT(data[i + (i & -i) - 1], data[i - 1]);
}
// O(log(n))
void update(int p, T x){
assert(0 <= p && p < n);
for(++ p; p <= n; p += p & -p) data[p - 1] = TT(data[p - 1], x);
}
// O(log(n))
void set(int p, T x){
update(p, TT(x, Tinv(query(p))));
}
// O(log(n))
T prefix(int r) const{
assert(0 <= r && r <= n);
T s = T_id;
for(; r > 0; r -= r & -r) s = TT(s, data[r - 1]);
return s;
}
// O(log(n))
T query(int l, int r) const{
static_assert(ALLOW_NON_PREFIX_QUERY);
assert(0 <= l && l <= r && r <= n);
if(l == r) return T_id;
T sum_minus = T_id, sum_plus = T_id;
for(; l < r; r -= r & -r) sum_plus = TT(sum_plus, data[r - 1]);
for(; r < l; l -= l & -l) sum_minus = TT(sum_minus, data[l - 1]);
return TT(sum_plus, Tinv(sum_minus));
}
// O(log(n))
T query(int p) const{
static_assert(ALLOW_NON_PREFIX_QUERY);
return query(p, p + 1);
}
// O(log(n))
T query_all() const{
return prefix(n);
}
// pred(sum[0, r)) is T, T, ..., T, F, F, ..., F, returns max r with T
// O(log(n))
int max_pref(auto pred) const{
assert(pred(T_id));
int p = 0;
T sum = T_id;
for(auto i = __lg(n + 1); i >= 0; -- i) if(p + (1 << i) <= n && pred(TT(sum, data[p + (1 << i) - 1]))){
sum = TT(sum, data[p + (1 << i) - 1]);
p += 1 << i;
}
return p;
}
template<class output_stream>
friend output_stream &operator<<(output_stream &out, const fenwick_tree &fw){
out << "{";
for(auto i = 0; i < fw.n; ++ i){
out << fw.query(i);
if(i != fw.n - 1) out << ", ";
}
return out << '}';
}
};
template<class T, class F, class I>
auto make_fenwick_tree(F TT, T T_id, I Tinv){
return fenwick_tree<true, T, F, I>(TT, T_id, Tinv);
}
template<class T>
auto make_fenwick_tree_sum(){
return fenwick_tree<true, T, plus<>, negate<>>(plus<>(), T{0}, negate<>());
}
template<class T>
auto make_fenwick_tree_product(){
auto inverse = [](const T &x){ return 1 / x; };
return fenwick_tree<true, T, multiplies<>, decltype(inverse)>(multiplies<>(), T{1}, inverse);
}
template<class T>
auto make_fenwick_tree_min(){
auto TT = [&](const T &x, const T &y)->T{ return min(x, y); };
return fenwick_tree<false, T, decltype(TT), negate<>>(TT, numeric_limits<T>::max(), negate<>());
}
template<class T>
auto make_fenwick_tree_max(){
auto TT = [&](const T &x, const T &y)->T{ return max(x, y); };
return fenwick_tree<false, T, decltype(TT), negate<>>(TT, numeric_limits<T>::max(), negate<>());
}
const i64 oo = 1e18;
void solve() {
int n, k, q;
cin >> n >> k >> q;
vector<int> a(n), p(k);
vector<vector<int>> pos(k);
for (int i = 0; i < n; ++i) {
cin >> a[i];
pos[--a[i]].emplace_back(i);
}
iota(all(p), 0);
vector<int> qq(q);
vector<vector<pair<int, int>>> ntb(k);
for (int i = 0; i < q; ++i) {
cin >> qq[i]; --qq[i];
ntb[p[qq[i]]].emplace_back(p[qq[i] + 1], i);
ntb[p[qq[i] + 1]].emplace_back(p[qq[i]], ~i);
swap(p[qq[i]], p[qq[i] + 1]);
}
vector<int> ap(k);
iota(all(ap), 0);
sort(all(ap), [&](int x, int y) {
return isz(pos[x]) > isz(pos[y]);
});
auto fw = make_fenwick_tree_sum<int>();
fw.build(n);
vector<bool> vis(k);
vector<i64> vis2(k, oo), change(q);
for (auto val : ap) {
vis[val] = true;
for (auto idx : pos[val]) {
fw.update(idx, 1);
}
for (auto [val2, qidx] : ntb[val]) if (not vis[val2]) {
int rq = (qidx < 0 ? ~qidx : qidx);
if (vis2[val2] != oo) {
change[rq] = vis2[val2] * (qidx < 0 ? 1 : -1);
continue;
}
i64 cnt = 0;
for (auto idx : pos[val2]) {
cnt += fw.prefix(idx);
}
vis2[val2] = 1ll * isz(pos[val]) * isz(pos[val2]) - cnt * 2;
change[rq] = vis2[val2] * (qidx < 0 ? 1 : -1);
}
for (auto [val2, qidx] : ntb[val]) {
vis2[val2] = oo;
}
for (auto idx : pos[val]) {
fw.update(idx, -1);
}
}
i64 cres = 0;
for (int i = k - 1; i >= 0; --i) {
for (int j = isz(pos[i]) - 1; j >= 0; --j) {
cres += fw.prefix(pos[i][j]);
fw.update(pos[i][j], 1);
}
}
// cout << cres << endl;
for (int i = 0; i < q; ++i) {
cres = cres + change[i];
cout << cres << "\n";
}
}
signed main() {
#ifndef CDuongg
if (fopen(taskname".inp", "r"))
assert(freopen(taskname".inp", "r", stdin)), assert(freopen(taskname".out", "w", stdout));
#else
freopen("bai3.inp", "r", stdin);
freopen("bai3.out", "w", stdout);
auto start = chrono::high_resolution_clock::now();
#endif
ios_base::sync_with_stdio(false);
cin.tie(nullptr);
int t = 1; //cin >> t;
while(t--) solve();
#ifdef CDuongg
auto end = chrono::high_resolution_clock::now();
cout << "\n"; for(int i = 1; i <= 100; ++i) cout << '=';
cout << "\nExecution time: " << chrono::duration_cast<chrono::milliseconds> (end - start).count() << "[ms]" << endl;
#endif
}
