#include <bits/stdc++.h>
using namespace std;
using i32 = int;
using i64 = long long;
template <typename T>
using V = vector<T>;
template <typename T>
using VV = V<V<T>>;
#define OVERRIDE4(a, b, c, d, ...) d
#define REP2(i, n) for (i32 i = 0; i < (i32)(n); ++i)
#define REP3(i, l, r) for (i32 i = (i32)(l); i < (i32)(r); ++i)
#define PER2(i, n) for (i32 i = (i32)(n) - 1; i >= 0; --i)
#define PER3(i, l, r) for (i32 i = (i32)(r) - 1; i >= (i32)(l); --i)
#define REP(...) OVERRIDE4(__VA_ARGS__, REP3, REP2)(__VA_ARGS__)
#define PER(...) OVERRIDE4(__VA_ARGS__, PER3, PER2)(__VA_ARGS__)
#define LEN(x) (i32)size(x)
#define ALL(x) begin(x), end(x)
template <typename T>
bool chmin(T &x, const T &y) {
if (x > y) {
x = y;
return true;
}
return false;
}
template <typename T>
bool chmax(T &x, const T &y) {
if (x < y) {
x = y;
return true;
}
return false;
}
#include "meetings.h"
constexpr i32 INF = 1001001001;
constexpr i64 INF64 = 3003003003003003003;
i64 solve_naive(V<i32> h) {
i32 n = LEN(h);
V<i64> cost(n, 0);
V<pair<i64, i32>> stc;
stc.emplace_back(INF, -1);
i64 cur = 0;
REP(i, n) {
while (stc.back().first <= h[i]) {
i64 v;
i32 j;
tie(v, j) = stc.back();
stc.pop_back();
cur -= v * (j - stc.back().second);
}
cur += (i64)h[i] * (i - stc.back().second);
stc.emplace_back(h[i], i);
cost[i] += cur;
}
stc.clear();
stc.emplace_back(INF, n);
cur = 0;
PER(i, n) {
while (stc.back().first <= h[i]) {
i64 v;
i32 j;
tie(v, j) = stc.back();
stc.pop_back();
cur -= v * (stc.back().second - j);
}
cur += (i64)h[i] * (stc.back().second - i);
stc.emplace_back(h[i], i);
cost[i] += cur;
}
REP(i, n) {
cost[i] -= h[i];
}
i64 mn = INF64;
REP(i, n) {
chmin(mn, cost[i]);
}
return mn;
}
V<i64> minimum_costs(V<i32> h, V<i32> l, V<i32> r) {
i32 n = LEN(h);
i32 q = LEN(l);
assert(n <= 5000 && q <= 5000);
V<i64> ans(q);
REP(i, q) {
ans[i] = solve_naive(V<i32>(h.begin() + l[i], h.begin() + r[i] + 1));
}
return ans;
}
