#include //#include //using namespace atcoder; #pragma GCC target ("avx2") #pragma GCC optimization ("O3") #pragma GCC optimization ("unroll-loops") using namespace std; typedef vector VI; typedef vector VVI; typedef vector VS; typedef pair PII; typedef pair pii; typedef pair PLL; typedef pair TIII; typedef long long ll; typedef long double ld; typedef unsigned long long ull; #define FOR(i, s, n) for (int i = s; i < (int)n; ++i) #define REP(i, n) FOR(i, 0, n) #define rep(i, a, b) for (int i = a; i < (b); ++i) #define trav(a, x) for (auto &a : x) #define all(x) x.begin(), x.end() #define MOD 1000000007 template inline bool chmax(T1 &a, T2 b) {if (a < b) {a = b; return true;} return false;} template inline bool chmin(T1 &a, T2 b) {if (a > b) {a = b; return true;} return false;} const double EPS = 1e-9, PI = acos(-1); const double pi = 3.141592653589793238462643383279; //ここから編集 typedef string::const_iterator State; ll GCD(ll a, ll b){ return (b==0)?a:GCD(b, a%b); } ll LCM(ll a, ll b){ return a/GCD(a, b) * b; } template< int mod > struct ModInt { int x; ModInt() : x(0) {} ModInt(int64_t y) : x(y >= 0 ? y % mod : (mod - (-y) % mod) % mod) {} ModInt &operator+=(const ModInt &p) { if((x += p.x) >= mod) x -= mod; return *this; } ModInt &operator-=(const ModInt &p) { if((x += mod - p.x) >= mod) x -= mod; return *this; } ModInt &operator*=(const ModInt &p) { x = (int) (1LL * x * p.x % mod); return *this; } ModInt &operator/=(const ModInt &p) { *this *= p.inverse(); return *this; } ModInt operator-() const { return ModInt(-x); } ModInt operator+(const ModInt &p) const { return ModInt(*this) += p; } ModInt operator-(const ModInt &p) const { return ModInt(*this) -= p; } ModInt operator*(const ModInt &p) const { return ModInt(*this) *= p; } ModInt operator/(const ModInt &p) const { return ModInt(*this) /= p; } bool operator==(const ModInt &p) const { return x == p.x; } bool operator!=(const ModInt &p) const { return x != p.x; } ModInt inverse() const { int a = x, b = mod, u = 1, v = 0, t; while(b > 0) { t = a / b; swap(a -= t * b, b); swap(u -= t * v, v); } return ModInt(u); } ModInt pow(int64_t n) const { ModInt ret(1), mul(x); while(n > 0) { if(n & 1) ret *= mul; mul *= mul; n >>= 1; } return ret; } friend ostream &operator<<(ostream &os, const ModInt &p) { return os << p.x; } friend istream &operator>>(istream &is, ModInt &a) { int64_t t; is >> t; a = ModInt< mod >(t); return (is); } static int get_mod() { return mod; } }; using modint = ModInt< 998244353 >; template< typename T > struct Combination { vector< T > _fact, _rfact, _inv; Combination(int sz) : _fact(sz + 1), _rfact(sz + 1), _inv(sz + 1) { _fact[0] = _rfact[sz] = _inv[0] = 1; for(int i = 1; i <= sz; i++) _fact[i] = _fact[i - 1] * i; _rfact[sz] /= _fact[sz]; for(int i = sz - 1; i >= 0; i--) _rfact[i] = _rfact[i + 1] * (i + 1); for(int i = 1; i <= sz; i++) _inv[i] = _rfact[i] * _fact[i - 1]; } inline T fact(int k) const { return _fact[k]; } inline T rfact(int k) const { return _rfact[k]; } inline T inv(int k) const { return _inv[k]; } T P(int n, int r) const { if(r < 0 || n < r) return 0; return fact(n) * rfact(n - r); } T C(int p, int q) const { if(q < 0 || p < q) return 0; return fact(p) * rfact(q) * rfact(p - q); } T H(int n, int r) const { if(n < 0 || r < 0) return (0); return r == 0 ? 1 : C(n + r - 1, r); } }; struct WaveletMatrix { struct BitVector{ // 大ブロックを 256, 小ブロックを8としておく const uint64_t BLOCK_BITNUM = 256; const uint64_t TABLE_INTERVAL = 8; const uint64_t NOT_FOUND = 0xFFFFFFFFFFFFFFFFLLU; /* selectで使う */ uint64_t length; vector bits; /* 元のbit列を圧縮したもの */ vector blocks; /* 1つ目の補助データに該当するもの */ vector tables; /* 2つ目の補助データに該当するもの */ BitVector() : length(0){} BitVector(uint64_t N){ init(N); } void init(uint64_t N) { length = N; uint64_t block_num = (N + BLOCK_BITNUM - 1) / (BLOCK_BITNUM); blocks.assign(block_num+1,0); uint64_t bits_len = (N + TABLE_INTERVAL - 1) / (TABLE_INTERVAL); bits.assign(bits_len+1,0); tables.assign(bits_len+1,0); //cerr << blocks.size() << " " << bits.size() << endl; } void build(){ uint64_t block_ind, offset, table_ind, sz = (length + BLOCK_BITNUM - 1) / (BLOCK_BITNUM); blocks[0] = 0; for(block_ind=0; block_ind < sz; block_ind++){ tables[block_ind * BLOCK_BITNUM / TABLE_INTERVAL] = 0; table_ind = (block_ind * BLOCK_BITNUM) / TABLE_INTERVAL; for(offset=0; offset= 0 && pos < length); uint64_t block_ind = pos / TABLE_INTERVAL; uint64_t offset = pos % TABLE_INTERVAL; if(b == 1) bits[block_ind] |= (1UL << offset); else bits[block_ind] &= ~(1UL << offset); } int access(uint64_t pos){ assert(pos >= 0 && pos < length); uint64_t block_ind = pos / TABLE_INTERVAL; uint64_t offset = pos % TABLE_INTERVAL; return (bits[block_ind] >> offset) & 1; } /* [0,pos)の1の数 */ uint64_t rank(uint64_t bit, uint64_t pos){ uint64_t block_ind = pos / BLOCK_BITNUM; uint64_t table_ind = pos / TABLE_INTERVAL; uint64_t offset = pos % TABLE_INTERVAL; uint64_t m = (bits[table_ind] & ((1UL << offset) - 1)); return bit ? blocks[block_ind] + tables[table_ind] + __builtin_popcountll(m) : pos - rank(1,pos); } uint64_t rank(uint64_t bit, uint64_t l, uint64_t r){ return rank(bit, r) - rank(bit, l); } }; uint64_t length; uint64_t height; uint64_t sigma; vector B; /* 各高さの索引 */ vector zerosum; /* 各高さにおける0の数 */ vector> cums; WaveletMatrix(){} WaveletMatrix(vector v, uint64_t _sigma) :length(v.size()), sigma(_sigma) { height = (sigma == 1) ? 1 : (64 - __builtin_clzll(sigma-1)); B.resize(height); zerosum.resize(height), cums.resize(height); for(int i=0; i v0, v1; for(int j=0; j> (height - 1 - i)) & 1; B[i].set(bi_j, j); if(bi_j == 0) v0.emplace_back(v[j]); else v1.emplace_back(v[j]); cums[i][j+1] = cums[i][j] + v[j]; } zerosum[i] = v0.size(); for(int j=0; j= 0 && pos < length); uint64_t res = 0; for(int i=0; i= sigma) return 0; int l = 0; for(int i=0; i> (height - 1 - i)) & 1); l = (bit?zerosum[i]:0) + B[i].rank(bit, l); r = (bit?zerosum[i]:0) + B[i].rank(bit, r); } return r-l; } /* [l, r)のcの数 */ int rank(uint64_t c, int l, int r){ if(c >= sigma) return 0; for(int i=0; i> (height - 1 - i)) & 1); l = (bit?zerosum[i]:0) + B[i].rank(bit, l); r = (bit?zerosum[i]:0) + B[i].rank(bit, r); } return r-l; } /* [l, r)でk(=0,1,...,r-l-1)番目に小さい値を返す */ /* verify:https://old.yosupo.jp/problem/range_kth_smallest*/ uint64_t kth_smallest(int l, int r, int k) { uint64_t res = 0; for(int i=0; i> (height - 1 - i) & 1); if(bit == 0){ l = B[i].rank(0, l); r = B[i].rank(0, r); }else{ res += B[i].rank(0, l, r); l = zerosum[i] + B[i].rank(1, l); r = zerosum[i] + B[i].rank(1, r); } } return res; } /* [l, r)中でx以上y未満の要素の数を返す */ /* verify:https://codeforces.com/contest/1042/problem/D */ uint64_t range_freq(int l, int r, uint64_t x, uint64_t y) { return rank_lower(l, r, y) - rank_lower(l, r, x); } /* [l, r)中でv < xを満たす最大のvを返す */ /* verigy:http://judge.u-aizu.ac.jp/onlinejudge/description.jsp?id=1549*/ uint64_t prev_value(int l, int r, uint64_t x) { uint64_t cnt = rank_lower(l, r, x); return cnt == 0 ? 1e9 : kth_smallest(l, r, cnt - 1); } /* [l, r)中でx <= vを満たす最小のvを返す */ /* verigy:http://judge.u-aizu.ac.jp/onlinejudge/description.jsp?id=1549*/ uint64_t next_value(int l, int r, uint64_t x) { uint64_t cnt = rank_lower(l, r, x); return cnt == r-l ? 1e9 : kth_smallest(l, r, cnt); } }; template struct BIT{ vector node; int N; BIT(int n){ N = n; node.resize(N+1); } void add(int i, T x){ i++; while(i <= N){ node[i] += x; i += (i & -i); } } T sum(int i){ T res = 0; while(i > 0){ res += node[i]; i -= (i & -i); } return res; } }; int main() { cin.tie(0); ios::sync_with_stdio(false); cout << fixed << setprecision(12); int N, K; cin >> N >> K; vector a(N); BIT sums(1010101), tot(1010101); REP(i,N) cin >> a[i]; vector v; REP(i,N){ v.push_back(a[i]); } sort(all(v)); v.erase(unique(all(v)), v.end()); WaveletMatrix wm(a, *max_element(all(a)) + 1); REP(i,N) a[i] = lower_bound(all(v), a[i]) - v.begin(); for(int i=0; i