結果
| 問題 | No.705 ゴミ拾い Hard |
| コンテスト | |
| ユーザー |
 drken1215
|
| 提出日時 | 2026-10-24 12:37:17 |
| 言語 | C++23 (gcc 15.3.0 + boost 1.92.0 + ACL) |
| 結果 |
AC
不安定
|
| 実行時間 | 255 ms / 1,500 ms |
| + 239µs | |
| コード長 | 43,667 bytes |
| 記録 | |
| コンパイル時間 | 5,694 ms |
| コンパイル使用メモリ | 400,912 KB |
| 実行使用メモリ | 25,152 KB |
| 最終ジャッジ日時 | 2026-10-24 12:37:31 |
| 合計ジャッジ時間 | 13,889 ms |
|
ジャッジサーバーID （参考情報） |
judge5_0 / judge3_0 |
（要ログイン）
| ファイルパターン | 結果 |
|---|---|
| sample | AC * 4 |
| other | AC * 39 |
ソースコード
//
// Monge 最短路問題に関するアルゴリズム全集
//
// verified
// COLOCON 2018 Final C - スペースエクスプローラー高橋君 (for Monge CHT)
// https://beta.atcoder.jp/contests/colopl2018-final-open/tasks/colopl2018_final_c
//
// AtCoder EDPC Z - Frog 3 (for simple LARSCH, LARSCH, monge CHT)
// https://atcoder.jp/contests/dp/tasks/dp_z
//
// Codeforces Round 189 (Div. 1) C. Kalila and Dimna in the Logging Industry (for simple LARSCH, LARSCH, monge CHT)
// https://codeforces.com/contest/319/problem/C
//
// yukicoder No.705 ゴミ拾い Hard (for simple LARSCH, LARSCH, monge CHT)
// https://yukicoder.me/problems/no/705
//
// ABC 218 H - Red and Blue Lamps (for Alien DP)
// https://atcoder.jp/contests/abc218/tasks/abc218_h
//
// yukicoder No.952 危険な火薬庫 (for SMAWK + D-enum)
// https://yukicoder.me/problems/no/952
//
// Codeforces Round 438 (Div. 1 + Div. 2 combined) F. Yet Another Minimization Problem
// 2-solutions (slide access Monotone Minima + D-enum, slide access simple LARSCH + Alien DP)
// https://codeforces.com/contest/868/problem/F
//
// ABC 480 G - Zonal Score Minimization (for anti-Monge D&D SMAWK, Monge CHT)
// https://atcoder.jp/contests/abc480/tasks/abc480_g
//
#include <bits/stdc++.h>
using namespace std;
//------------------------------//
// Template
//------------------------------//
using ll = long long;
using i128 = __int128_t;
using u128 = __uint128_t;
using pint = pair<int, int>;
using pll = pair<long long, long long>;
using tll = array<long long, 3>;
using fll = array<long long, 4>;
using vint = vector<int>;
using vll = vector<long long>;
using dint = deque<int>;
using dll = deque<long long>;
using vvint = vector<vector<int>>;
using vvll = vector<vector<long long>>;
using vpll = vector<pair<long long, long long>>;
template<class T> using min_priority_queue = priority_queue<T, vector<T>, greater<T>>;
template<class S, class T> inline bool chmax(S &a, T b) { return (a < b ? a = b, 1 : 0); }
template<class S, class T> inline bool chmin(S &a, T b) { return (a > b ? a = b, 1 : 0); }
template<class S, class T> inline auto maxll(S a, T b) { return max(ll(a), ll(b)); }
template<class S, class T> inline auto minll(S a, T b) { return min(ll(a), ll(b)); }
template<class T> auto max(const T &a) { return *max_element(a.begin(), a.end()); }
template<class T> auto min(const T &a) { return *min_element(a.begin(), a.end()); }
template<class T> auto argmax(const T &a) { return max_element(a.begin(), a.end()) - a.begin(); }
template<class T> auto argmin(const T &a) { return min_element(a.begin(), a.end()) - a.begin(); }
template<class T> auto accum(const vector<T> &a) { return accumulate(a.begin(), a.end(), T()); }
template<class T> auto accum(const deque<T> &a) { return accumulate(a.begin(), a.end(), T()); }
#define REP(i, a) for (long long i = 0; i < (long long)(a); i++)
#define REP2(i, a, b) for (long long i = a; i < (long long)(b); i++)
#define RREP(i, a) for (long long i = (a)-1; i >= (long long)(0); --i)
#define RREP2(i, a, b) for (long long i = (b)-1; i >= (long long)(a); --i)
#define EB emplace_back
#define PF push_front
#define PB push_back
#define MP make_pair
#define FI first
#define SE second
#define ALL(x) x.begin(), x.end()
// input stream
template<class T> istream& operator >> (istream &is, vector<T> &P)
{ for (int i = 0; i < (int)P.size(); ++i) cin >> P[i]; return is; }
template<class T> istream& operator >> (istream &is, deque<T> &P)
{ for (int i = 0; i < (int)P.size(); ++i) cin >> P[i]; return is; }
template<class T> istream& operator >> (istream &is, vector<vector<T>> &P)
{ for (int i = 0; i < (int)P.size(); ++i) cin >> P[i]; return is; }
// output stream
#define COUT(x) cout << #x << " = " << (x) << " (L" << __LINE__ << ")" << endl
template<class S, class T> ostream& operator << (ostream &s, const pair<S, T> &P)
{ return s << '<' << P.first << ", " << P.second << '>'; }
template<class T> ostream& operator << (ostream &s, const array<T, 2> &P)
{ return s << '<' << P[0] << "," << P[1] << '>'; }
template<class T> ostream& operator << (ostream &s, const array<T, 3> &P)
{ return s << '<' << P[0] << "," << P[1] << "," << P[2] << '>'; }
template<class T> ostream& operator << (ostream &s, const array<T, 4> &P)
{ return s << '<' << P[0] << "," << P[1] << "," << P[2] << "," << P[3] << '>'; }
template<class T> ostream& operator << (ostream &s, const vector<string> &P)
{ for (int i = 0; i < P.size(); ++i) { s << P[i] << endl; } return s; }
template<class T> ostream& operator << (ostream &s, const vector<T> &P)
{ for (int i = 0; i < P.size(); ++i) { if (i > 0) { s << " "; } s << P[i]; } return s; }
template<class T> ostream& operator << (ostream &s, const deque<T> &P)
{ for (int i = 0; i < P.size(); ++i) { if (i > 0) { s << " "; } s << P[i]; } return s; }
template<class T> ostream& operator << (ostream &s, const vector<vector<T>> &P)
{ for (int i = 0; i < P.size(); ++i) { s << endl << P[i]; } return s << endl; }
template<class T> ostream& operator << (ostream &s, const set<T> &P)
{ for (auto it : P) { s << "<" << it << "> "; } return s; }
template<class T> ostream& operator << (ostream &s, const multiset<T> &P)
{ for (auto it : P) { s << "<" << it << "> "; } return s; }
template<class T> ostream& operator << (ostream &s, const unordered_set<T> &P)
{ for (auto it : P) { s << "<" << it << "> "; } return s; }
template<class S, class T> ostream& operator << (ostream &s, const map<S, T> &P)
{ for (auto it : P) { s << "<" << it.first << "->" << it.second << "> "; } return s; }
template<class S, class T> ostream& operator << (ostream &s, const unordered_map<S, T> &P)
{ for (auto it : P) { s << "<" << it.first << "->" << it.second << "> "; } return s; }
//------------------------------//
// Monge
//------------------------------//
// check whether N x M matrix f is Monge or not, in O(NM)
template<class FUNC> bool is_monge
(int N, int M, const FUNC &f, bool upper_triangular = false, bool output_detail = true) {
assert(N > 0 && M > 0);
for (int i = 0; i + 1 < N; i++) {
for (int j = (upper_triangular ? i+2 : 0); j + 1 < M; j++) {
auto f00 = f(i, j), f01 = f(i, j + 1), f10 = f(i + 1, j), f11 = f(i + 1, j + 1);
if (f00 + f11 > f10 + f01) {
if (output_detail) {
cout << "Not Monge! " << endl;
cout << "f(" << i << ", " << j << ") = " << f(i, j) << ", ";
cout << "f(" << i << ", " << j + 1 << ") = " << f(i, j + 1) << endl;
cout << "f(" << i + 1 << ", " << j << ") = " << f(i + 1, j) << ", ";
cout << "f(" << i + 1 << ", " << j + 1 << ") = " << f(i + 1, j + 1) << endl;
}
return false;
}
}
}
return true;
}
// check whether N x M matrix f is anti-Monge or not, in O(NM)
template<class FUNC> bool is_anti_monge
(int N, int M, const FUNC &f, bool upper_triangular = false, bool output_detail = true) {
assert(N > 0 && M > 0);
for (int i = 0; i + 1 < N; i++) {
for (int j = (upper_triangular ? i+2 : 0); j + 1 < M; j++) {
auto f00 = f(i, j), f01 = f(i, j + 1), f10 = f(i + 1, j), f11 = f(i + 1, j + 1);
if (f00 + f11 < f10 + f01) {
if (output_detail) {
cout << "Not anti-Monge! " << endl;
cout << "f(" << i << ", " << j << ") = " << f(i, j) << ", ";
cout << "f(" << i << ", " << j + 1 << ") = " << f(i, j + 1) << endl;
cout << "f(" << i + 1 << ", " << j << ") = " << f(i + 1, j) << ", ";
cout << "f(" << i + 1 << ", " << j + 1 << ") = " << f(i + 1, j + 1) << endl;
}
return false;
}
}
}
return true;
}
// Convex Hull Trick
/*
Func: Monge 性を仮定
注意点：SMAWK や LARSCH とは異なり、i < j なる領域だけでなく、全領域での Monge 性が必要！！！
- MIN: クエリ x の取りうる最小値
- MAX: クエリ x の取りうる最大値 (f_i(MAX) がオーバーフローしないように注意）
- INF: 最大値
- insert (Func f_i): add f_i, O(log X)
- query (x): min_i{ f_i(x) }, O(log X)
*/
template<class T> struct MongeCHT {
using Func = function<T(T)>;
struct IndexedFunc {
Func f;
int id;
IndexedFunc(Func f, int id) : f(std::move(f)), id(id) {}
T eval(T x) const { return f(x); };
};
struct Node {
IndexedFunc func;
unique_ptr<Node> left, right;
explicit Node(IndexedFunc f) : func(std::move(f)) {}
};
const T MIN, MAX, INF = numeric_limits<T>::max() / 2;
unique_ptr<Node> root;
int next_id = 0;
MongeCHT() {}
MongeCHT(T MIN, T MAX, T INF = numeric_limits<T>::max() / 2) : MIN(MIN), MAX(MAX), INF(INF) {
assert(MIN <= MAX);
}
// insert
void insert(unique_ptr<Node> &p, T low, T high, IndexedFunc &f) {
if (!p) {
p = make_unique<Node>(std::move(f));
return;
}
if (p->func.eval(low) <= f.eval(low) && p->func.eval(high) <= f.eval(high)) return;
if (p->func.eval(low) >= f.eval(low) && p->func.eval(high) >= f.eval(high)) {
p->func = std::move(f);
return;
}
T mid = (low + high) / 2;
if (p->func.eval(mid) > f.eval(mid)) swap(p->func, f);
if (low == high) return;
if (p->func.eval(low) >= f.eval(low)) insert(p->left, low, mid, f);
else insert(p->right, mid + 1, high, f);
}
int insert(Func f) {
int id = next_id++;
IndexedFunc func(std::move(f), id);
insert(root, MIN, MAX, func);
return id;
}
void clear() {
root.reset();
next_id = 0;
}
// query
pair<T, int> query(const unique_ptr<Node>& p, T low, T high, T x) const {
if (!p) return {INF, -1};
pair<T, int> res = {p->func.eval(x), p->func.id};
if (low == high) return res;
T mid = (low + high) / 2;
if (x <= mid) return min(res, query(p->left, low, mid, x));
else return min(res, query(p->right, mid + 1, high, x));
}
pair<T, int> query(T x) const {
assert(MIN <= x && x <= MAX);
return query(root, MIN, MAX, x);
}
};
// find shortest path on DAG with Monge or anti-Monge cost, by Convex Hull Trick, O(N log N)
// vertex: 0, 1, 2, ..., N
// f(i, j) must be Monge or anti-Monge
// domain: j >= start[i], start is non-decreasing
template<class VAL, class FUNC> vector<pair<VAL, int>> MongeShortestPathByCHT(int N, const FUNC &f, const vector<int> &start) {
assert((int)start.size() == N);
VAL INF = numeric_limits<VAL>::max() / 2;
vector<pair<VAL, int>> dp(N + 1, make_pair(numeric_limits<VAL>::max() / 2, -1));
MongeCHT<VAL> cht(1, N, INF);
dp[0].first = 0;
int i = 0;
for (int j = 1; j <= N; j++) {
while (i < N && j >= start[i]) {
auto f2 = [&, i](int j) -> long long { return dp[i].first + f(i, j); };
cht.insert(f2);
i++;
}
dp[j] = cht.query(j);
};
return dp;
}
template<class VAL, class FUNC> vector<pair<VAL, int>> MongeShortestPathByCHT(int N, const FUNC &f) {
vector<int> start(N);
for (int i = 0; i < N; i++) start[i] = i + 1;
return MongeShortestPathByCHT<VAL>(N, f, start);
}
// find min_j f(i, j) for all i, by Monotone Minima, O(H + W log H)
// f(i, j) must be monotone (argmin is not decreasing)
// slide access OK
template<class VAL, class FUNC> void MonotoneMinimaRec
(int HL, int HR, int WL, int WR, const FUNC &f, vector<pair<VAL, int>> &res) {
if (HR - HL <= 0) return;
int HM = (HL + HR) / 2;
res[HM].second = WL;
for (int i = WL; i < WR; i++) {
VAL val = f(HM, i);
if (res[HM].first > val) res[HM] = make_pair(val, i);
}
MonotoneMinimaRec(HL, HM, WL, res[HM].second + 1, f, res);
MonotoneMinimaRec(HM + 1, HR, res[HM].second, WR, f, res);
}
template<class VAL, class FUNC> vector<pair<VAL, int>> MonotoneMinima(int H, int W, const FUNC &f) {
vector<pair<VAL, int>> res(H, make_pair(numeric_limits<VAL>::max() / 2, -1));
MonotoneMinimaRec(0, H, 0, W, f, res);
return res;
}
// find shortest path on DAG with monotone cost, by D&D Monotone Minima, O(N (log N)^2)
// vertex: 0, 1, 2, ..., N
// f(i, j) must be monotone (argmin is not decreasing)
// slide access OK
template<class VAL, class FUNC> vector<pair<VAL, int>> DDMonotoneMinima(int N, const FUNC &f) {
vector<pair<VAL, int>> res(N + 1, make_pair(numeric_limits<VAL>::max() / 2, -1));
res[0].first = VAL(0);
auto f2 = [&](int i, int j) -> VAL { return res[j].first + f(j, i); };
auto rec = [&](auto &&rec, int left, int right) -> void {
if (right - left <= 1) return;
int mid = (left + right) / 2;
rec(rec, left, mid);
MonotoneMinimaRec(mid, right, left, mid, f2, res);
rec(rec, mid, right);
};
rec(rec, 0, N + 1);
return res;
}
// find min_j f(i, j) for all i, by SMAWK, O(H + W)
// f(i, j) must be totally monotone
// slide access NG
template<class VAL, class FUNC> void SMAWKRec
(const vector<int> &X, const vector<int> &Y, const FUNC &f, vector<pair<VAL, int>> &res) {
if (X.empty()) return;
// Reduce Step
vector<int> X2, Y2;
for (auto y : Y) {
while (!Y2.empty()) {
int py = Y2.back(), x = X[(int)Y2.size() - 1];
if (f(x, y) >= f(x, py)) break;
Y2.pop_back();
}
if (Y2.size() < X.size()) Y2.emplace_back(y);
}
// Recurse Step
for (int i = 1; i < (int)X.size(); i += 2) X2.emplace_back(X[i]);
SMAWKRec(X2, Y2, f, res);
// Interpolate Step
int p = 0;
for (int i = 0; i < (int)X.size(); i += 2) {
int lim = (i + 1 < (int)X.size() ? res[X[i + 1]].second : Y.back()), best = Y[p];
while (Y[p] < lim) {
p++;
if (f(X[i], Y[p]) < f(X[i], best)) best = Y[p];
}
res[X[i]] = {f(X[i], best), best};
}
}
template<class VAL, class FUNC> vector<pair<VAL, int>> SMAWK(int H, int W, const FUNC &f) {
if (H == 0) return {};
assert(W > 0);
vector<pair<VAL, int>> res(H, make_pair(numeric_limits<VAL>::max() / 2, -1));
vector<int> X(H), Y(W);
for (int i = 0; i < H; i++) X[i] = i;
for (int j = 0; j < W; j++) Y[j] = j;
SMAWKRec(X, Y, f, res);
return res;
}
// find shortest path on DAG with totally monotone cost, by D&D SMAWK, O(N log N)
// vertex: 0, 1, 2, ..., N
// f(i, j) must be totally monotone
template<class VAL, class FUNC> vector<pair<VAL, int>> DDSMAWK(int N, const FUNC &f) {
vector<pair<VAL, int>> res(N + 1, make_pair(numeric_limits<VAL>::max() / 2, -1));
vector<pair<VAL, int>> tmp(N + 1);
res[0].first = VAL(0);
auto f2 = [&](int i, int j) -> VAL { return res[j].first + f(j, i); };
auto rec = [&](auto &&rec, int left, int right) -> void {
if (right - left <= 1) return;
int mid = (left + right) / 2;
vector<int> X(right - mid), Y(mid - left);
for (int i = mid; i < right; i++) X[i - mid] = i;
for (int j = left; j < mid; j++) Y[j - left] = j;
rec(rec, left, mid);
SMAWKRec(X, Y, f2, tmp);
for (auto x : X) if (tmp[x].first < res[x].first) res[x] = tmp[x];
rec(rec, mid, right);
};
rec(rec, 0, N + 1);
return res;
}
// find shortest path on DAG with anti-Monge cost, by D&D SMAWK, O(NK log N)
// vertex: 0, 1, 2, ..., N
// f(i, j) must be anti-Monge, j >= i + K
template<class VAL, class FUNC> vector<pair<VAL, int>> antiMongeDDSMAWK(int N, const FUNC &f, int K = 1) {
assert(K >= 1);
const VAL INF = numeric_limits<VAL>::max() / 4;
vector<pair<VAL, int>> res(N + 1, make_pair(INF, -1));
vector<pair<VAL, int>> tmp(N + 1, {INF, -1});
res[0].first = VAL(0);
auto relax = [&](int i, int j) -> void {
if (i != 0 && res[i].second == -1) return;
VAL val = res[i].first + f(i, j);
if (val < res[j].first) res[j] = {val, i};
};
auto rec = [&](auto &&rec, int left, int right) -> void {
if (right - left <= 1) return;
int mid = (left + right) / 2, border = min(right, mid + K - 1);
rec(rec, left, mid);
for (int j = mid; j < border; j++) {
for (int i = left; i < mid && i <= j - K; i++) relax(i, j);
}
if (border < right) {
vector<int> X, Y;
for (int j = border; j < right; j++) X.push_back(j);
for (int i = mid - 1; i >= left; i--) {
if (i == 0 || res[i].second != -1) Y.push_back(i);
}
if (!Y.empty()) {
vector<int> cols(Y.size());
iota(cols.begin(), cols.end(), 0);
auto f2 = [&](int j, int k) -> VAL { return res[Y[k]].first + f(Y[k], j); };
SMAWKRec(X, cols, f2, tmp);
for (auto x : X) {
if (tmp[x].first < res[x].first) res[x] = {tmp[x].first, Y[tmp[x].second]};
}
}
}
rec(rec, mid, right);
};
rec(rec, 0, N + 1);
return res;
}
// noshi's simplified LARSCH
// find shortest path from vertex 0 on DAG in O(N log N)
// vertex: 0, 1, 2, ..., N, f(i, j) must be Monge
// slide access OK
template<class VAL> struct SimpleLARSCH {
VAL INF = numeric_limits<VAL>::max() / 2;
int CNT_INF = numeric_limits<int>::max() / 2;
// results
vector<VAL> dp;
vector<int> cnt, prev;
// solver (random access ver)
template<class FUNC> vector<pair<VAL, int>> solve(int N, const FUNC &f, bool minimize_cnt = true) {
dp.assign(N + 1, INF);
cnt.assign(N + 1, CNT_INF);
prev.assign(N + 1, 0);
dp[0] = 0, cnt[0] = 0;
auto relax = [&](int l, int r) -> void {
VAL val = dp[l] + f(l, r);
int cn = cnt[l] + 1;
if (dp[r] > val || (dp[r] == val && (minimize_cnt ? cn < cnt[r] : cn > cnt[r]))) {
dp[r] = val;
cnt[r] = cn;
prev[r] = l;
}
};
auto rec = [&](auto &&rec, int l, int r) -> void {
if (r - l <= 1) return;
int m = (l + r) / 2;
for (int k = prev[l]; k <= prev[r]; k++) relax(k, m);
rec(rec, l, m);
for (int k = l + 1; k <= m; k++) relax(k, r);
rec(rec, m, r);
};
if (N > 0) relax(0, N), rec(rec, 0, N);
vector<pair<VAL, int>> res(N + 1, make_pair(numeric_limits<VAL>::max() / 2, -1));
res[0].first = VAL(0);
for (int i = 1; i <= N; i++) res[i] = {dp[i], prev[i]};
return res;
}
// solver (slide access ver)
template<class STATE, class ADD, class DEL, class GETCOST> vector<pair<VAL, int>> solve
(int N, const STATE &ini, const ADD &add, const DEL &del, const GETCOST &get, bool minimize_cnt = true) {
dp.assign(N + 1, INF);
cnt.assign(N + 1, CNT_INF);
prev.assign(N + 1, 0);
dp[0] = 0, cnt[0] = 0;
STATE win[2] = { ini, ini };
int cl[2] = {0, 0}, cr[2] = {0, 0};
auto move_cursor = [&](int c, int l, int r) {
while (cr[c] < r) add(win[c], cr[c]++);
while (cl[c] > l) add(win[c], --cl[c]);
while (cr[c] > r) del(win[c], --cr[c]);
while (cl[c] < l) del(win[c], cl[c]++);
};
auto relax = [&](int c, int l, int r) {
move_cursor(c, l, r);
VAL val = dp[l] + get(win[c]);
int cn = cnt[l] + 1;
if (dp[r] > val || (dp[r] == val && (minimize_cnt ? cn < cnt[r] : cn > cnt[r]))) {
dp[r] = val;
cnt[r] = cn;
prev[r] = l;
}
};
auto rec = [&](auto &&rec, int l, int r) -> void {
if (r - l <= 1) return;
int m = (l + r) / 2;
for (int k = prev[l]; k <= prev[r]; k++) relax(0, k, m);
rec(rec, l, m);
for (int k = l + 1; k <= m; k++) relax(1, k, r);
rec(rec, m, r);
};
if (N > 0) relax(0, 0, N), rec(rec, 0, N);
vector<pair<VAL, int>> res(N + 1, make_pair(numeric_limits<VAL>::max() / 2, -1));
res[0].first = VAL(0);
for (int i = 1; i <= N; i++) res[i] = {dp[i], prev[i]};
return res;
}
vector<int> reconstruct() {
int N = (int)dp.size() - 1;
vector<int> path;
for (int v = N; v > 0; v = prev[v]) path.emplace_back(v);
path.emplace_back(0);
reverse(path.begin(), path.end());
return path;
}
};
// LARSCH, in O(N)
// vertex: 0, 1, 2, ..., N, f(i, j) must be Monge
template<class VAL, class FUNC> struct LARSCH {
struct RowReduce;
struct ColReduce;
struct ColMap {
const ColMap* parent = nullptr;
const vector<int> *v = nullptr;
constexpr int mapping(int i) const {
int x = (v ? (*v)[i] : i);
return (parent ? parent->mapping(x) : x);
}
};
struct Eval {
const FUNC *f = nullptr;
long long a = 1, b = 0; // row = a * i + b
const ColMap *cm = nullptr;
constexpr VAL operator () (int i, int j) const {
int i2 = int(a * i + b), j2 = (cm ? cm -> mapping(j) : j);
return (*f)(i2, j2);
}
};
struct RowReduce {
int N;
Eval e;
int cur_row = 0, state = 0;
unique_ptr<ColReduce> rec;
constexpr RowReduce(int N_, const Eval &e_) : N(N_), e(e_) {
int M = N / 2;
if (M > 0) {
Eval e2 = e;
e2.a = e.a * 2, e2.b = e.a + e.b;
rec = make_unique<ColReduce>(M, e2);
}
}
constexpr void reset() {
cur_row = 0, state = 0;
if (rec) rec->reset();
}
int get_argmin() {
int i = cur_row++;
if (!(i & 1)) {
int prev = state, next = (i + 1 == N ? N - 1 : rec->get_argmin());
state = next;
int res = prev;
for (int j = prev + 1; j <= next; j++) {
if (e(i, res) > e(i, j)) res = j;
}
return res;
} else {
return (e(i, state) <= e(i, i) ? state : i);
}
}
};
struct ColReduce {
int N;
Eval e;
int cur_row = 0;
vector<int> cols;
ColMap cm;
RowReduce rec;
constexpr ColReduce(int N_, const Eval &e_)
: N(N_), e(e_), cols(), cm{e.cm, &cols}, rec(N_, Eval{e.f, e.a, e.b, &cm}) {
cols.reserve(N);
}
constexpr void reset() {
cur_row = 0;
cols.clear();
rec.reset();
}
constexpr void push_col(int i, int j) {
while (!cols.empty()) {
int siz = (int)cols.size();
if (siz == i) break;
int last = cols.back();
if (e(siz - 1, last) > e(siz - 1, j)) cols.pop_back();
else break;
}
if ((int)cols.size() != N) cols.emplace_back(j);
}
constexpr int get_argmin() {
int i = cur_row++;
if (i == 0) {
cols.clear();
cols.emplace_back(0);
} else {
push_col(i, i * 2 - 1);
push_col(i, i * 2);
}
return cols[rec.get_argmin()];
}
};
FUNC f;
ColMap root_cm;
Eval root_eval;
unique_ptr<RowReduce> base;
explicit LARSCH(int N_, const FUNC &f_)
: f(std::move(f_)), root_cm{nullptr, nullptr}, root_eval{&f, 1, 0, &root_cm} {
base = make_unique<RowReduce>(N_, root_eval);
}
constexpr void reset() {
base->reset();
}
constexpr int get_argmin() {
return base->get_argmin();
}
};
// Monge Shortest Path Wrapper
template<class VAL> struct MongeShortestPath {
VAL INF = numeric_limits<VAL>::max() / 2;
int CNT_INF = numeric_limits<int>::max() / 2;
// results
vector<VAL> dp;
vector<int> cnt, prev;
// solver (random access ver)
template<class FUNC> vector<pair<VAL, int>> solve(int N, const FUNC &f
, const string solver = "simple_larsch") {
vector<pair<VAL, int>> res(N + 1, make_pair(numeric_limits<VAL>::max() / 2, -1));
if (solver == "simple_larsch") {
SimpleLARSCH<VAL> slar;
res = slar.solve(N, f, true);
dp = slar.dp, cnt = slar.cnt, prev = slar.prev;
} else if (solver == "larsch") {
dp.assign(N + 1, INF);
cnt.assign(N + 1, CNT_INF);
prev.assign(N + 1, 0);
dp[0] = 0, cnt[0] = 0;
auto f2 = [&](int i, int j) -> VAL {
i++;
if (i <= j) return INF;
else return dp[j] + f(j, i);
};
LARSCH<VAL, decltype(f2)> lar(N, f2);
for (int r = 1; r <= N; r++) {
int l = lar.get_argmin();
dp[r] = dp[l] + f(l, r);
prev[r] = l;
cnt[r] = cnt[l] + 1;
}
res[0].first = VAL(0);
for (int r = 1; r <= N; r++) res[r] = {dp[r], prev[r]};
} else if (solver == "monotone") {
res = DDMonotoneMinima<VAL>(N, f);
} else if (solver == "smawk") {
res = DDSMAWK<VAL>(N, f);
} else if (solver == "cht") {
res = MongeShortestPathByCHT<VAL>(N, f);
} else {
cout << "no solver" << endl;
assert(true);
}
return res;
}
// solver (slide access ver)
template<class STATE, class ADD, class DEL, class GETCOST> vector<pair<VAL, int>> solve
(int N, const STATE &ini, const ADD &add, const DEL &del, const GETCOST &get
, const string solver = "simple_larsch") {
vector<pair<VAL, int>> res;
if (solver == "simple_larsch") {
SimpleLARSCH<VAL> sl;
res = sl.solve(N, ini, add, del, get, true);
dp = sl.dp, cnt = sl.cnt, prev = sl.prev;
} else {
cout << "no solver" << endl;
assert(true);
}
return res;
}
vector<int> reconstruct() {
int N = (int)dp.size() - 1;
vector<int> path;
for (int v = N; v > 0; v = prev[v]) path.emplace_back(v);
path.emplace_back(0);
reverse(path.begin(), path.end());
return path;
}
};
// find the d-edges shortest path for d = 1, 2, ..., D, in O(ND)
// vertex: 0, 1, 2, ..., N
// you can choose SMAWK (random access) or Monotone Minima (slide access)
template<class VAL> struct MongeShortestPathWithDEdges {
VAL INF = numeric_limits<VAL>::max() / 2;
// solver
template<class FUNC> vector<VAL> solve(int N, const FUNC &f, int D, const string &solver = "smawk") {
assert(D <= N);
vector<VAL> res{VAL(0)}, dp(N + 1, INF);
dp[0] = 0;
auto f2 = [&](int i, int j) -> VAL { return (j < i ? dp[j] + f(j, i) : INF); };
for (int d = 1; d <= D; d++) {
vector<pair<VAL, int>> tmp;
if (solver == "smawk") tmp = SMAWK<VAL>(N + 1, N + 1, f2);
else if (solver == "monotone") tmp = MonotoneMinima<VAL>(N + 1, N + 1, f2);
for (int i = d; i <= N; i++) dp[i] = tmp[i].first;
res.emplace_back(dp[N]);
}
return res;
}
};
/*
Alien's Trick (by ラグランジュ緩和)
min_{x}: f(x) s.t. g(x) = K
= max_{λ は整数}: min_{x} (f(x) + λ(g(x) - K))
max_{x}: f(x) s.t. g(x) = K
= min_{λ は整数}: max_{x} (f(x) + λ(g(x) - K))
【仮定】
・f(x), g(x): 0 以上の整数値をとる関数
・K: 整数
・g(x) = p となる x についての f(x) の最小値を h(p) とおくと、h(p) は p について下に凸な関数
・h(K) < ∞ (g(x) = K となる x が存在)
【インターフェース】
・lag(λ): x✳︎ = argmin_{x} {f(x) + λ(g(x) - K)} としたときのペア値 {f(x✳︎), g(x✳︎)} を返す関数
・返り値: {最適値, そのときの λ}
*/
template<class VAL> struct AliensTrick {
// find lo such that g(lo) >= R
template<class FUNC> tuple<VAL, VAL, VAL> find_low(const FUNC &lag, VAL R) {
auto [f0, g0] = lag(VAL(0));
VAL lo = 0, hi = 0, flo = f0, glo = g0;
if (glo >= R) return tie(lo, flo, glo);
lo = VAL(-1);
tie(flo, glo) = lag(lo);
while (glo < R) {
assert(lo < hi);
VAL dif = hi - lo;
hi = lo, lo -= VAL(2) * dif;
tie(flo, glo) = lag(lo);
}
return tie(lo, flo, glo);
}
// find hi such that g(hi) <= L
template<class FUNC> tuple<VAL, VAL, VAL> find_high(const FUNC &lag, VAL L) {
auto [f0, g0] = lag(VAL(0));
VAL lo = 0, hi = 0, fhi = f0, ghi = g0;
if (ghi <= L) return tie(hi, fhi, ghi);
hi = VAL(1);
tie(fhi, ghi) = lag(hi);
while (ghi > L) {
assert(lo < hi);
VAL dif = hi - lo;
lo = hi, hi += VAL(2) * dif;
tie(fhi, ghi) = lag(hi);
}
return tie(hi, fhi, ghi);
}
// solver
template<class FUNC> pair<VAL, VAL> solve(const FUNC &lag, VAL K) {
// zero check
auto [f0, g0] = lag(VAL(0));
if (g0 == K) return {f0, VAL(0)};
// find an initial value of binary search (by doubling)
auto [lo, flo, glo] = find_low(lag, K);
auto [hi, fhi, ghi] = find_high(lag, K);
if (glo == K) return {flo, lo};
if (ghi == K) return {fhi, hi};
assert(glo > K && ghi < K);
// binary search
while (hi - lo > VAL(1)) {
VAL mid = (lo + hi) / 2;
auto [fmid, gmid] = lag(mid);
if (gmid == K) return {fmid, mid};
if (gmid > K) lo = mid, flo = fmid, glo = gmid;
else hi = mid, fhi = fmid, ghi = gmid;
}
// in case: h(p) has a collinear part (= could not achieve g(x) = K)
// return the Lagrange value
VAL reslo = flo + lo * (glo - K), reshi = fhi + hi * (ghi - K);
if (reslo >= reshi) return {reslo, lo};
else return {reshi, hi};
}
template<class FUNC> pair<VAL, VAL> minimize(const FUNC &lag, VAL K) {
return solve(lag, K);
}
template<class FUNC> pair<VAL, VAL> maximize(const FUNC &lag, VAL K) {
auto reverse_lag = [&](VAL lambda) -> pair<VAL, VAL> {
auto [f, g] = lag(-lambda);
return {-f, g};
};
auto [f, g] = solve(reverse_lag, K);
return {-f, -g};
}
// in case: L <= g(x) <= R
template<class FUNC> pair<VAL, VAL> solve(const FUNC &lag, VAL L, VAL R) {
assert(L <= R);
auto [f0, g0] = lag(VAL(0));
if (L <= g0 && g0 <= R) return {f0, VAL(0)};
if (g0 > R) return solve(lag, R);
else return solve(lag, L);
}
template<class FUNC> pair<VAL, VAL> minimize(const FUNC &lag, VAL L, VAL R) {
return solve(lag, L, R);
}
template<class FUNC> pair<VAL, VAL> maximize(const FUNC &lag, VAL L, VAL R) {
auto reverse_lag = [&](VAL lambda) -> pair<VAL, VAL> {
auto [f, g] = lag(-lambda);
return {-f, g};
};
auto [f, g] = solve(reverse_lag, L, R);
return {-f, -g};
}
};
// Alien DP
// find the d-edges shortest path from vertex 0 on DAG by Lagrange relaxation
// vertex: 0, 1, 2, ..., N, f(i, j) must be Monge
template<class VAL> struct AlienDP {
// inner values
MongeShortestPath<VAL> msp;
AliensTrick<VAL> at;
// solver (random access ver.)
template<class FUNC> pair<VAL, VAL> solve(int N, const FUNC &f, int D
, const string &solver = "simple_larsch") {
auto lag = [&](VAL lambda) -> pair<VAL, VAL> {
auto fg = [&](int i, int j) -> VAL { return f(i, j) + lambda; };
const auto &obj = msp.solve(N, fg, solver);
auto cnt = msp.cnt[N];
auto res = obj.back().first - lambda * cnt;
return {res, cnt};
};
return at.solve(lag, D);
}
// solver (slide access ver.)
template<class STATE, class ADD, class DEL, class GETCOST> pair<VAL, int> solve
(int N, const STATE &ini, const ADD &add, const DEL &del, const GETCOST &get, int D
, const string &solver = "simple_larsch") {
auto lag = [&](VAL lambda) -> pair<VAL, VAL> {
auto get_lag = [&](STATE &st) -> VAL { return get(st) + lambda; };
const auto &obj = msp.solve(N, ini, add, del, get_lag, solver);
auto cnt = msp.cnt[N];
auto res = obj.back().first - lambda * cnt;
return {res, cnt};
};
return at.solve(lag, D);
}
// debugger
template<class FUNC> void debug(int N, const FUNC &f, VAL start = 0, VAL goal = 10) {
auto lag = [&](VAL lambda) -> void {
auto fg = [&](int i, int j) -> VAL { return f(i, j) + lambda; };
const auto &dp = msp.solve(N, fg);
auto cnt = msp.cnt[N];
auto res = dp.back().first - lambda * cnt;
cout << lambda << ": " << res << "(" << cnt << "); (";
for (int i = 0; i < (int)dp.size(); i++) {
if (i) cout << " ";
cout << "<" << dp[i].first << ", " << dp[i].second << ">";
}
cout << ")" << endl;
};
for (VAL lambda = start; lambda <= goal; lambda++) lag(lambda);
}
};
// f: Monge, interval monotone ((N+1) x (N+1))
// dp[i][j] = min_{i <= k < j} (dp[i][k] + dp[k][j]) + f(i, j), in O(N^2)
template<class VAL, class FUNC> VAL KnuthYao(int N, const FUNC &f) {
VAL INF = numeric_limits<VAL>::max() / 2;
vector<vector<VAL>> dp(N+1, vector<VAL>(N+1, INF));
vector<vector<int>> K(N+1, vector<int>(N+1, -1));
for (int i = 0; i <= N; i++) {
dp[i][i] = 0, K[i][i] = i;
if (i < N) dp[i][i+1] = f(i, i+1), K[i][i+1] = i;
}
for (int between = 2; between <= N; between++) {
for (int i = 0; i + between <= N; i++) {
int j = i + between;
for (int k = K[i][j-1]; k <= K[i+1][j]; k++) {
VAL tmp = dp[i][k] + dp[k][j] + f(i, j);
if (dp[i][j] > tmp) {
dp[i][j] = tmp;
K[i][j] = k;
}
}
}
}
return dp[0][N];
}
//------------------------------//
// Examples
//------------------------------//
// COLOCON 2018 Final C - スペースエクスプローラー高橋君
void COLOCON_2018_final_C() {
long long N;
cin >> N;
vector<long long> a(N), res(N);
for (int i = 0; i < N; ++i) cin >> a[i];
MongeCHT<long long> cht(0, 210000, 1LL<<60);
for (long long i = 0; i < N; ++i) {
auto func = [i, &a](long long x) -> long long { return (x - i) * (x - i) + a[i]; };
cht.insert(func);
}
for (long long i = 0; i < N; ++i) res[i] = cht.query(i).first;
for (int i = 0; i < N; ++i) cout << res[i] << endl;
}
// AtCoder EDPC Z - Frog 3
/*
H は単調増加数列
chmin(dp[j], dp[i] + (H[j] - H[i])^2 + C)
i -> j のコスト：(H[j] - H[i])^2 ...... 差の凸関数は Monge
スタート: 0, ゴール: N-1
*/
void EDPC_Z(const string solver = "simple_larsch") {
long long N, C;
cin >> N >> C;
vector<long long> H(N);
for (long long i = 0; i < N; i++) cin >> H[i];
auto func = [&](int i, int j) -> long long {
return (H[j] - H[i]) * (H[j] - H[i]) + C;
};
MongeShortestPath<long long> msp;
const auto &res = msp.solve(N-1, func, solver);
cout << res[N-1].first << endl;
}
// Codeforces Round 189 (Div. 1) C. Kalila and Dimna in the Logging Industry
/*
A: 単調増加, B: 単調減少, ともに長さ N
i -> j のコストが、B[i] × A[j] で与えられる ..... 単調増加 × 単調減少は Monge
スタート: 0, ゴール: N-1
*/
void Codeforces_189_C(const string solver = "simple_larsch") {
long long N;
cin >> N;
vector<long long> A(N), B(N);
for (int i = 0; i < N; i++) cin >> A[i];
for (int i = 0; i < N; i++) cin >> B[i];
auto func = [&](int i, int j) -> long long {
return B[i] * A[j];
};
MongeShortestPath<long long> msp;
auto res = msp.solve(N-1, func, solver);
cout << res[N-1].first << endl;
}
// yukicoder No.705 ゴミ拾い Hard
/*
A, X, Y: N 個
これらを区間に分割していく
dp[j] = min_{0 ≦ i < j}(dp[i] + |A[j-1] - X[i]|^3 + |-Y[i]|^3)
i -> j のコスト：|A[j-1] - X[i]|^3 + |-Y[i]|^3 ...... 差の凸関数 (Monge) + 縞々 (Monge) -> Monge
スタート: 0, ゴール: N
*/
void yukicoder_705(const string solver = "simple_larsch") {
int N;
cin >> N;
vector<long long> A(N), X(N), Y(N);
for (int i = 0; i < N; i++) cin >> A[i];
for (int i = 0; i < N; i++) cin >> X[i];
for (int i = 0; i < N; i++) cin >> Y[i];
auto func = [&](int i, int j) -> long long {
long long dx = abs(A[j-1] - X[i]), dy = abs(Y[i]);
return dx * dx * dx + dy * dy * dy;
};
MongeShortestPath<long long> msp;
auto res = msp.solve(N, func, solver);
cout << res[N].first << endl;
}
// ABC 218 H - Red and Blue Lamps
void ABC_218_H() {
long long N, R;
cin >> N >> R;
vector<long long> A(N - 1);
for (int i = 0; i < N - 1; i++) cin >> A[i];
auto cost = [&](int i, int j) -> long long {
if (j - i <= 1) return 0;
if (i == 0) return -A[j - 2];
else return -A[i - 1] - A[j - 2];
};
AlienDP<long long> ad;
//ad.debug(N, cost);
auto [resR, cntR] = ad.solve(N, cost, R); // both "larsch" and "simple_larsch" are OK
auto [resB, cntB] = ad.solve(N, cost, N - R);
auto res = max(-resR, -resB);
cout << res << endl;
}
// yukicoder No.952 危険な火薬庫
/*
末尾にドアを追加する。
ノード 0, 1, ..., N, N+1
f(i, j) = (S[j-1] - S[i])^2
選ぶドアが k 個 → 区間の個数は、N+1-k
*/
void yukicoder_952() {
int N;
cin >> N;
vector<long long> A(N + 1, 0), S(N + 2, 0);
for (int i = 0; i < N; i++) cin >> A[i], S[i + 1] = S[i] + A[i];
auto f = [&](int i, int j) -> long long {
return (S[j - 1] - S[i]) * (S[j - 1] - S[i]);
};
MongeShortestPathWithDEdges<long long> msp;
auto res = msp.solve(N + 1, f, N + 1);
for (int k = 1; k <= N; k++) cout << res[N + 1 - k] << endl;
}
// Codeforces Round 438 (Div. 1 + Div. 2 combined) F. Yet Another Minimization Problem
// by Monge 単一始点 d-辺最短路の d = 1, 2, ..., D における列挙 (by SMAWK 法, in O(ND))
void Codeforces438_F_enum() {
long long N, K;
cin >> N >> K;
vector<long long> A(N);
for (int i = 0; i < N; i++) cin >> A[i];
vector<long long> nums(N + 1, 0);
long long left = 0, right = 0, score = 0;
auto add = [&](int i) -> void {
score += nums[A[i]];
nums[A[i]]++;
};
auto del = [&](int i) -> void {
nums[A[i]]--;
score -= nums[A[i]];
};
auto f = [&](int i, int j) -> long long {
while (left < i) del(left++);
while (i < left) add(--left);
while (right < j) add(right++);
while (j < right) del(--right);
return score;
};
MongeShortestPathWithDEdges<long long> msp;
auto res = msp.solve(N, f, K, "monotone"); // we should use slide access
cout << res[K] << endl;
}
// Codeforces Round 438 (Div. 1 + Div. 2 combined) F. Yet Another Minimization Problem
// by Alien DP (not the assumed solution)
void Codeforces438_F_alien() {
long long N, K;
cin >> N >> K;
vector<long long> A(N);
for (int i = 0; i < N; i++) cin >> A[i];
struct STATE {
vector<long long> nums;
long long cost;
STATE(int N) : nums(N + 1, 0), cost(0) {}
};
STATE ini(N);
auto add = [&](STATE &st, int i) -> void {
st.cost += st.nums[A[i]];
st.nums[A[i]]++;
};
auto del = [&](STATE &st, int i) -> void {
st.nums[A[i]]--;
st.cost -= st.nums[A[i]];
};
auto get = [&](STATE &st) -> long long {
return st.cost;
};
AlienDP<long long> adp;
auto res = adp.solve(N, ini, add, del, get, K); // we should use slide access
cout << res.first << endl;
}
// ABC 480 G - Zonal Score Minimization
template<class MeetSemiLattice> struct SparseTable {
using Func = function<MeetSemiLattice(MeetSemiLattice, MeetSemiLattice)>;
// core member
Func OP = [](const MeetSemiLattice &l, const MeetSemiLattice &r) {
return min(l, r);
};
vector<vector<MeetSemiLattice>> dat;
vector<int> height;
SparseTable() {}
SparseTable(const vector<MeetSemiLattice> &vec) {
init(vec);
}
SparseTable(const vector<MeetSemiLattice> &vec, const Func &op) {
init(vec, op);
}
void init(const vector<MeetSemiLattice> &vec) {
int n = (int)vec.size(), h = 1;
while ((1<<h) <= n) ++h;
dat.assign(h, vector<MeetSemiLattice>(1<<h));
height.assign(n+1, 0);
for (int i = 2; i <= n; i++) height[i] = height[i>>1]+1;
for (int i = 0; i < n; ++i) dat[0][i] = vec[i];
for (int i = 1; i < h; ++i) {
for (int j = 0; j < n; ++j)
dat[i][j] = OP(dat[i-1][j], dat[i-1][min(j+(1<<(i-1)),n-1)]);
}
}
void init(const vector<MeetSemiLattice> &vec, const Func &op) {
OP = op;
init(vec);
}
MeetSemiLattice get(int a, int b) {
return OP(dat[height[b-a]][a], dat[height[b-a]][b-(1<<height[b-a])]);
}
};
void ABC_480_G_SMAWK() {
int N;
cin >> N;
vector<long long> A(N);
for (int i = 0; i < N; ++i) cin >> A[i];
SparseTable<long long> stmi(A);
SparseTable<long long> stma(A, [&](long long p, long long q){return max(p, q);});
auto func = [&](int l, int r) -> long long {
assert(r - l >= 2);
return stma.get(l, r) - stmi.get(l, r);
};
auto res = antiMongeDDSMAWK<long long>(N, func, 2);
cout << res.back().first << endl;
}
void ABC_480_G_CHT() {
int N;
cin >> N;
long long INF = numeric_limits<long long>::max() / 2;
long long M = numeric_limits<long long>::max() / (N * 3);
vector<long long> A(N);
for (int i = 0; i < N; ++i) cin >> A[i];
SparseTable<long long> stmi(A);
SparseTable<long long> stma(A, [&](long long p, long long q){return max(p, q);});
auto func = [&](int l, int r) -> long long {
if (r < l + 2) return -M * ((l + 1 - r) * 2 + 1);
else return stma.get(l, r) - stmi.get(l, r);
};
//assert(is_anti_monge(N + 1, N + 1, func));
vector<int> start(N);
for (int i = 0; i < N; i++) start[i] = i + 2; // 辺 (i, i + 1) は禁止
auto res = MongeShortestPathByCHT<long long>(N, func, start);
cout << res.back().first << endl;
}
int main() {
//COLOCON_2018_final_C();
//EDPC_Z("simple_larsch");
//EDPC_Z("larsch");
//EDPC_Z("cht");
//Codeforces_189_C("simple_larsch");
//Codeforces_189_C("larsch");
//Codeforces_189_C("cht");
//yukicoder_705("simple_larsch");
//yukicoder_705("larsch");
yukicoder_705("cht");
//ABC_218_H();
//yukicoder_952();
//Codeforces438_F_enum();
//Codeforces438_F_alien();
//ABC_480_G_SMAWK();
//ABC_480_G_CHT();
}