#432764 (C++14) No.995 タピオカオイシクナーレ

提出ソース
結果

問題	No.995 タピオカオイシクナーレ
ユーザー	otamay6
提出日時	2020-02-21 21:54:18
言語	C++14 (gcc 13.2.0 + boost 1.83.0)
結果	AC
実行時間	18 ms / 2,000 ms
コード長	43,126 bytes
コンパイル時間	3,627 ms
コンパイル使用メモリ	215,532 KB
実行使用メモリ	5,376 KB
最終ジャッジ日時	2024-04-17 08:05:09
合計ジャッジ時間	4,278 ms
ジャッジサーバーID （参考情報）	judge3 / judge5
このコードへのチャレンジ
（要ログイン）
テストケース

テストケース表示
入力	結果	実行時間実行使用メモリ
testcase_00	AC	2 ms 5,248 KB
testcase_01	AC	2 ms 5,376 KB
testcase_02	AC	2 ms 5,376 KB
testcase_03	AC	2 ms 5,376 KB
testcase_04	AC	2 ms 5,376 KB
testcase_05	AC	2 ms 5,376 KB
testcase_06	AC	1 ms 5,376 KB
testcase_07	AC	2 ms 5,376 KB
testcase_08	AC	2 ms 5,376 KB
testcase_09	AC	2 ms 5,376 KB
testcase_10	AC	2 ms 5,376 KB
testcase_11	AC	2 ms 5,376 KB
testcase_12	AC	2 ms 5,376 KB
testcase_13	AC	2 ms 5,376 KB
testcase_14	AC	2 ms 5,376 KB
testcase_15	AC	2 ms 5,376 KB
testcase_16	AC	18 ms 5,376 KB
testcase_17	AC	18 ms 5,376 KB
testcase_18	AC	17 ms 5,376 KB
testcase_19	AC	17 ms 5,376 KB
testcase_20	AC	16 ms 5,376 KB
testcase_21	AC	17 ms 5,376 KB
testcase_22	AC	17 ms 5,376 KB
testcase_23	AC	17 ms 5,376 KB
testcase_24	AC	18 ms 5,376 KB
testcase_25	AC	18 ms 5,376 KB
権限があれば一括ダウンロードができます
ソースコード

raw source code
#include <bits/stdc++.h>
#define REP(i,n) for(int i=0,i##_len=int(n);i<i##_len;++i)
#define RREP(i,n) for(int i=int(n)-1;i>=0;--i)
#define rep(i,a,b) for(int i=int(a);i<int(b);++i)
#define rrep(i,a,b) for(int i=int(a)-1;i>=int(b);--i)
#define All(x) (x).begin(),(x).end()
#define rAll(x) (x).rbegin(),(x).rend()
#define ITR(i,x) for(auto i=(x).begin();i!=(x).end();++i)
using std::cin;
using std::cout;
using std::cerr;
using std::endl;
using std::vector;
using std::string;


typedef long long ll;
typedef std::pair<ll, ll> P;
typedef vector<int> vi;
typedef vector<vi> vvi;
typedef vector<ll> vl;
typedef vector<vl> vvl;
constexpr ll mod = 1e9+7;
constexpr double eps = 1e-9;
const double PI = acos(-1);
void solve();

namespace Math{
    template<typename T>
    bool chmax(T &a,T b){
        if(a<b){
            a=b;
            return true;
        }
        return false;
    }
    template<typename T>
    bool chmin(T &a,T b){
        if(b<a){
            a=b;
            return true;
        }
        return false;
    }
    ll bisearch(ll ok,ll ng,std::function<bool(ll)> check){
        while(llabs(ok-ng)>1){
            ll mid=ng-((ng-ok)>>1);
            if(check(mid)) ok=mid;
            else ng=mid;
        }
        return ok;
    }
    ll sqrt(ll n){ll s=1; while(s*s>n||n>=(s+1)*(s+1)){ s=(n/s+s)/2; } return s;}
    ll roundup(ll n,ll div){
        if(n>0) return (n-1)/div+1;
        else return n/div;    
    }
    bool square(ll a){ll n=Math::sqrt(a);return a==n*n;}
    template<typename T>
    T pow(T x, ll n){
        T ans = T(1);
        while(n != 0){
            if(n&1) ans = ans*x;
            x = x*x;
            n = n >> 1;
        }
        return ans;
    }
    int digitsum(ll N,int a){
        if(N==0) return 0;
        int ret=0;
        ret+=digitsum(N/a,a)+N%a;
        return ret;
    }
    ll gcd(ll x,ll y){return y ? gcd(y,x%y) : x;};
    ll lcm(ll x,ll y){return x/Math::gcd(x,y)*y;}
    ll manhattan(const P &a,const P &b){return llabs(a.first-b.first)+llabs(a.second-b.second);}
}
namespace Solution{
    using Graph = vector<vector<int>>;
    ll knapsack(int kinds,int MAX_W,const vl &weight,const vl &cost){
        vector<vector<ll>> dp(kinds+1,vector<ll>(MAX_W+1,0));
        REP(i,kinds) REP(j,MAX_W+1){
            if(j<weight[i]) dp[i+1][j]=dp[i][j];
            else dp[i+1][j]=std::max(dp[i][j],dp[i][j-weight[i]]+cost[i]);
        }
        return dp[kinds][MAX_W];
    }
    ll cost_based_knapsack(int kinds,int MAX_W,const vl &weight,const vl &cost){
        int MAX_V=0;
        REP(i,kinds) Math::chmax(MAX_V,(int)cost[i]);
        vvl dp(kinds+1,vl(kinds*MAX_V+1,1LL<<58));
        dp[0][0] = 0;
        REP(i,kinds) REP(j,kinds*MAX_V+1){
            if(j<cost[i]) dp[i+1][j]=dp[i][j];
            else dp[i+1][j] = std::min(dp[i][j],dp[i][j-cost[i]]+weight[i]);
        }
        int res=0;
        REP(i,kinds*MAX_V+1) if(dp[kinds][i]<=MAX_W) res=i;
        return res;
    }
    ll unlimited_knapsack(int kinds,int MAX_W,const vl &weight,const vl &cost){
        vector<ll> dp(MAX_W+1);
        REP(i,kinds) for(int j=weight[i];j<=MAX_W;++j){
            dp[j] = std::max(dp[j],dp[j-weight[i]]+cost[i]);
        }
        return dp[MAX_W];
    }
    ll huge_knapsack(int kinds,ll MAX_W,const vl &weight,const vl &cost){
        int n2=kinds/2;
        vector<P> ps(1<<(kinds/2));
        REP(i,1<<n2){
            ll sw=0,sv=0;
            REP(j,n2){
                if(i>>j&1){
                    sw += weight[j];
                    sv += cost[j];
                }
            }
            ps[i] = P(sw,sv);
        }
        std::sort(ps.begin(),ps.begin() + (1<<n2) );
        int m=1;
        rep(i,1,1<<n2){
            if(ps[m-1].second<ps[i].second) ps[m++] = ps[i];
        }

        ll res=0;
        REP(i,1<<(kinds-n2)){
            ll sw=0,sv=0;
            REP(j,kinds-n2){
                if((i>>j)&1){
                    sw += weight[n2+j];
                    sv += cost[n2+j];
                }
            }
            if(sw<=MAX_W){
                ll tv = (lower_bound(ps.begin(),ps.begin()+m,P(MAX_W-sw,LLONG_MAX))-1)->second;
                Math::chmax(res,sv+tv);
            }
        }
        return res;
    }
    ll choose_MonN(int N,int M,const vl &cost){
        vvl dp(N+1,vl(M+1,-1LL<<58));
        REP(i,N+1) dp[i][0]=0;
        REP(i,N) REP(j,M){
            if(j>i) break;
            dp[i+1][j+1]=std::max(dp[i][j+1],dp[i][j]+cost[i]);
        }
        return dp[N][M];
    }
    ll Partition_Func(int n,int k){
        vector<vector<ll>> dp(k+1,vector<ll> (n+1,0));
        dp[0][0]=1;
        rep(i,1,k+1) REP(j,n+1){
            if(j-i>=0) dp[i][j]=(dp[i][j-i]+dp[i-1][j]);
            else dp[i][j]=dp[i-1][j];
        }
        return dp[k][n];
    }
    int LCS(string s,string t){
        int n=s.length(),m=s.length();
        vector<vector<int>> dp(n+1,vector<int>(m+1));
        REP(i,n) REP(j,m){
            if (s[i] == t[j]) dp[i+1][j+1] = dp[i][j] + 1;
            else dp[i+1][j+1] = std::max(dp[i][j+1], dp[i+1][j]);
        }
        return dp[n][m];
    }
    int LIS(const vector<ll> a){
        int n=a.size();
        ll INF=1LL<<50;
        vector<ll> dp(n+1,INF);
        REP(i,n) *std::lower_bound(All(dp),a[i])=a[i];
        int k=std::lower_bound(All(dp),INF)-dp.begin();
        return k;
    }
    
    int max_flow(int s,int t,const vector<vector<P>> &g){
        struct edge_{int to,cap, rev;};
        vector<bool> used(g.size(),false);
        vector<vector<edge_>> G(g.size());
        REP(i,g.size()) REP(j,g[i].size()){
            int from = i, to = g[i][j].second;
            int cap = g[i][j].first;
            G[from].push_back((edge_){to,cap,(int)G[to].size()});
            G[to].push_back((edge_){from,0,(int)G[from].size()-1});
        }
        auto dfs = [&](auto&& f,int v,int t,int fl)->int{
            if(v==t) return fl;
            used[v] = true;
            REP(i,G[v].size()){
                edge_ &e = G[v][i];
                if(!used[e.to] && e.cap>0){
                    int d = f(f, e.to,t,std::min(fl,e.cap));
                    if(d>0){
                        e.cap -= d;
                        G[e.to][e.rev].cap += d;
                        return d;
                    }
                }
            }
            return 0;
        };
        int flow=0;
        while(1){
            used.assign(used.size(),false);
            int f = dfs(dfs,s,t,INT_MAX);
            if(f==0) return flow;
            flow += f;
        }
    }
    int bipartite_matching_calculate(const Graph &g){
        int V = g.size();
        vi match(V,-1);
        vector<bool> used(V,false);
        auto dfs = [&](auto&& f,int v)->bool{
            used[v]=true;
            REP(i,g[v].size()){
                int u = g[v][i], w = match[u];
                if(w<0 || (!used[w] && f(f,w))){
                    match[v]=u;
                    match[u]=v;
                    return true;
                }
            }
            return false;
        };
        int res=0;
        REP(v,V){
            if(match[v] < 0){
                used.assign(V,false);
                if(dfs(dfs,v)) res++;
            }
        }
        return res;
    }
    int bipartite_matching(int l,int r,std::function<bool(int,int)> F){
        int V = l+r;
        Graph G(V);
        REP(i,l) REP(j,r) if(F(i,j)){
            G[i].push_back(l+j);
            G[l+j].push_back(i);
        }
        return bipartite_matching_calculate(G);
    }
    ll dinic(int s,int t,const vector<vector<P>> &graph){
        struct max_flow {
            struct edge_ { int to;ll cap;int rev; };
            int V;
            vector<vector<edge_>> G;
            vector<int> itr, level;
            max_flow(int V) : V(V) { G.assign(V,vector<edge_>()); }

            void add_edge(int from, int to, ll cap) {
                G[from].push_back((edge_) {to, cap, (int) G[to].size()});
                G[to].push_back((edge_) {from, 0, (int) G[from].size()-1});
            }
            void bfs(int s) {
                level.assign(V,-1);
                std::queue<int> q;
                level[s] = 0; q.push(s);
                while (!q.empty()) {
                    int v = q.front(); q.pop();
                    for(auto &e: G[v]) if(e.cap > 0 and level[e.to] < 0) {
                            level[e.to] = level[v] + 1;
                            q.push(e.to);
                    }
                }
            }

            ll dfs(int v, int t, ll f) {
                if(v == t) return f;
                for(int& i=itr[v]; i < (int) G[v].size(); ++i) {
                    edge_& e = G[v][i];
                    if (e.cap > 0 and level[v] < level[e.to]) {
                        int d = dfs(e.to, t, std::min(f, e.cap));
                        if (d > 0) {
                            e.cap -= d;
                            G[e.to][e.rev].cap += d;
                            return d;
                        }
                    }
                }
                return 0;
            }

            int run(int s, int t) {
                int ret = 0, f;
                while(bfs(s), level[t] >= 0) {
                    itr.assign(V,0);
                    while ((f = dfs(s, t, 1LL<<59)) > 0) ret += f;
                }
                return ret;
            }
        };
        max_flow d(graph.size());
        REP(i,graph.size()) for(auto e:graph[i]){
            d.add_edge(i,e.second,e.first);
        }
        return d.run(s,t);
    }
}
namespace NTT {
    using uint = uint_fast32_t;

    // NTT_PRIMES {{{
    constexpr ll NTT_PRIMES[][2] = {
    {1224736769, 3}, // 2^24 * 73 + 1,
    {1053818881, 7}, // 2^20 * 3 * 5 * 67 + 1
    {1051721729, 6}, // 2^20 * 17 * 59 + 1
    {1045430273, 3}, // 2^20 * 997 + 1
    {1012924417, 5}, // 2^21 * 3 * 7 * 23 + 1
    {1007681537, 3}, // 2^20 * 31^2 + 1
    {1004535809, 3}, // 2^21 * 479 + 1
    {998244353, 3},  // 2^23 * 7 * 17 + 1
    {985661441, 3},  // 2^22 * 5 * 47 + 1
    {976224257, 3},  // 2^20 * 7^2 * 19 + 1
    {975175681, 17}, // 2^21 * 3 * 5 * 31 + 1
    {962592769, 7},  // 2^21 * 3^3 * 17 + 1
    {950009857, 7},  // 2^21 * 4 * 151 + 1
    {943718401, 7},  // 2^22 * 3^2 * 5^2 + 1
    {935329793, 3},  // 2^22 * 223 + 1
    {924844033, 5},  // 2^21 * 3^2 * 7^2 + 1
    {469762049, 3},  // 2^26 * 7 + 1
    {167772161, 3},  // 2^25 * 5 + 1
    };
    ll extgcd(ll a, ll b, ll &x, ll &y) {
        ll d;
        return b == 0 ? (x = a < 0 ? -1 : 1, y = 0, a < 0 ? -a : a)
                : (d = extgcd(b, a % b, y, x), y -= a / b * x, d);
    }
    ll modinv(ll a, ll mod) {
        ll x, y;
        extgcd(a, mod, x, y);
        x %= mod;
        return x < 0 ? x + mod : x;
    }
    ll modpow(ll a, ll b, ll mod) {
        ll r = 1;
        a %= mod;
        while(b) {
            if(b & 1) r = r * a % mod;
            a = a * a % mod;
            b >>= 1;
        }
        return r;
    }

    // NTT Core {{{
    template < int MAX_H >
    struct Pool {
        static ll *tmp, *A, *B;
    };
    template < int MAX_H >
    ll *Pool< MAX_H >::tmp = new ll[1 << MAX_H];
    template < int MAX_H >
    ll *Pool< MAX_H >::A = new ll[1 << MAX_H];
    template < int MAX_H >
    ll *Pool< MAX_H >::B = new ll[1 << MAX_H];

    template < int MAX_H, ll mod, ll primitive >
    class Core {
      public:
        static_assert((mod & ((1 << MAX_H) - 1)) == 1, "mod is too small; comment out");
        // ord zetaList[i] = 2^(i + 1)
        ll zetaList[MAX_H], zetaInvList[MAX_H];
        // constexpr
        Core() {
            zetaList[MAX_H - 1] = modpow(primitive, (mod - 1) / (1 << MAX_H), mod);
            zetaInvList[MAX_H - 1] = modinv(zetaList[MAX_H - 1], mod);
            for(int ih = MAX_H - 2; ih >= 0; --ih) {
                zetaList[ih] = zetaList[ih + 1] * zetaList[ih + 1] % mod;
                zetaInvList[ih] = zetaInvList[ih + 1] * zetaInvList[ih + 1] % mod;
            }
        }
        void fft(ll *a, uint n, uint nh, bool inverse) const {
            ll *tmp = Pool< MAX_H >::tmp;
            uint mask = n - 1;
            for(uint i = n >> 1, ih = nh - 1; i >= 1; i >>= 1, --ih) {
                ll zeta = inverse ? zetaInvList[nh - 1 - ih] : zetaList[nh - 1 - ih];
                ll powZeta = 1;
                for(uint j = 0; j < n; j += i) {
                    for(uint k = 0; k < i; ++k) {
                        tmp[j | k] =
                        (a[((j << 1) & mask) | k] + powZeta * a[(((j << 1) | i) & mask) | k]) % mod;
                    }
                powZeta = powZeta * zeta % mod;
                }
                std::swap(a, tmp);
            }
            if(nh & 1) {
                std::swap(a, tmp);
                for(uint i = 0; i < n; ++i) a[i] = tmp[i];
            }
            if(inverse) {
                ll invN = modinv(n, mod);
                for(uint i = 0; i < n; ++i) a[i] = a[i] * invN % mod;
            }
        }
        vector< ll > conv(const vector< ll > &a, const vector< ll > &b) const {
            uint t = a.size() + b.size() - 1;
            uint n = 1, nh = 0;
            while(n < t) n <<= 1, ++nh;
            return convStrict(a, b, n, nh);
        }
        vector< ll > convStrict(const vector< ll > &a, const vector< ll > &b, uint n,
                          uint nh) const {
            ll *A = Pool< MAX_H >::A, *B = Pool< MAX_H >::B;
            for(uint i = 0; i < n; ++i) A[i] = B[i] = 0;
            copy(a.begin(), a.end(), A);
            copy(b.begin(), b.end(), B);
            fft(A, n, nh, 0), fft(B, n, nh, 0);
            for(uint i = 0; i < n; ++i) A[i] = A[i] * B[i] % mod;
            fft(A, n, nh, 1);
            return vector< ll >(A, A + n);
        }
    };

    // Convolution With Garner {{{
    template < int MAX_H, int I >
    class ConvolutionWithGarnerCore {
      public:
        static void conv_for(uint n, uint nh, const vector< ll > &a, const vector< ll > &b,
                       vector< ll > &mods, vector< ll > &coeffs,
                       vector< vector< ll > > &constants) {
            static const Core< MAX_H, NTT_PRIMES[I][0], NTT_PRIMES[I][1] > ntt;
            auto c = ntt.convStrict(a, b, n, nh);
            mods[I] = NTT_PRIMES[I][0];
            ConvolutionWithGarnerCore< MAX_H, I - 1 >::conv_for(
                n, nh, a, b, mods, coeffs, constants);
            // garner
            for(size_t i = 0; i < c.size(); ++i) {
                ll v = (c[i] - constants[I][i]) * modinv(coeffs[I], mods[I]) % mods[I];
                if(v < 0) v += mods[I];
                for(size_t j = I + 1; j < mods.size(); ++j) {
                    constants[j][i] = (constants[j][i] + coeffs[j] * v) % mods[j];
                }
            }
            for(size_t j = I + 1; j < mods.size(); ++j) {
                coeffs[j] = (coeffs[j] * mods[I]) % mods[j];
            }
        }
    };

    template < int MAX_H >
    class ConvolutionWithGarnerCore< MAX_H, -1 > {
      public:
        static void conv_for(uint, uint, const vector< ll > &, const vector< ll > &,
                       vector< ll > &, vector< ll > &, vector< vector< ll > > &) {}
    };

    template < int MAX_H >
    class ConvolutionWithGarner {
      public:
        template < int USE >
        static vector< ll > conv(const vector< ll > &a, const vector< ll > &b, ll mod) {
            static_assert(USE >= 1, "USE must be positive");
            static_assert(USE <= sizeof(NTT_PRIMES) / sizeof(*NTT_PRIMES), "USE is too big");
            uint nt = a.size() + b.size() - 1;
            uint n = 1, nh = 0;
            while(n < nt) n <<= 1, ++nh;
            vector< ll > coeffs(USE + 1, 1);
            vector< vector< ll > > constants(USE + 1, vector< ll >(n));
            vector< ll > mods(USE + 1, mod);
            ConvolutionWithGarnerCore< MAX_H, USE - 1 >::conv_for(
                n, nh, a, b, mods, coeffs, constants);
            return constants.back();
        }
    };

} 
// 1st param is MAX_H
NTT::Core< 18, NTT::NTT_PRIMES[0][0], NTT::NTT_PRIMES[0][1] > nttBig;
NTT::Core< 18, 998244353, 5 > ntt;
using nttconv = NTT::ConvolutionWithGarner< 18 >;
// nttconv::conv< USE >(a, b, mod)

template <class iter>
std::pair<std::complex<double>, double> min_ball(iter left, iter right, int seed = 1333) {
    const int n = right - left;
    using P=std::complex<double>;
    using ld=double;
    
    assert(n >= 1);
    if (n == 1) {
        return {*left, ld(0)};
    }

    std::mt19937 mt(seed);
    std::shuffle(left, right, mt);
    // std::random_shuffle(left, right); // simple but deprecated

    iter ps = left;
    using circle = std::pair<P, ld>;
    auto cross=[](const P& a, const P& b) { return a.real()*b.imag() - a.imag()*b.real(); };
    auto dot=[](const P& a, const P& b) { return a.real()*b.real() + a.imag()*b.imag(); };
    auto make_circle_3 = [&](const P &a, const P &b, const P &c) -> circle {
        ld A = std::norm(b - c), B = std::norm(c - a), C = std::norm(a - b),
           S = cross(b - a, c - a);
        P p = (A * (B + C - A) * a + B * (C + A - B) * b + C * (A + B - C) * c) / (4 * S * S);
        ld r2 = std::norm(p - a);
        return {p, r2};
    };

    auto make_circle_2 = [](const P &a, const P &b) -> circle {
        P c = (a + b) / (ld)2;
        ld r2 = std::norm(a - c);
        return {c, r2};
    };

    auto in_circle = [](const P &a, const circle &c) -> bool {
        const double eps=1e-9;
        return std::norm(a - c.first) <= c.second + eps;
    };

    circle c = make_circle_2(ps[0], ps[1]);

    // MiniDisc
    for (int i = 2; i < n; ++i) {
        if (!in_circle(ps[i], c)) {
            // MiniDiscWithPoint
            c = make_circle_2(ps[0], ps[i]);
            for (int j = 1; j < i; ++j) {
                if (!in_circle(ps[j], c)) {
                    // MiniDiscWith2Points
                    c = make_circle_2(ps[i], ps[j]);
                    for (int k = 0; k < j; ++k) {
                        if (!in_circle(ps[k], c)) {
                            c = make_circle_3(ps[i], ps[j], ps[k]);
                        }
                    }
                }
            }
        }
    }
    return c;
}

template<typename V>
size_t KMP(const V &Search,const V &Word){
    size_t i=2,j=0;
    std::vector<int> Table(Search.size());
    Table[0]=-1;Table[1]=0;
    while(i<Word.size()){
        if(Word[i-1]==Word[j]){
            Table[i]=j+1;
            ++i;++j;
        }
        else if(j>0) j=Table[j];
        else{
            Table[i]=0;
            ++i;
        }
    }
    i=0;j=0;
    while(j+i<Search.size()){
        if(Word[i]==Search[j+i]){
            ++i;
            if(i==Word.size()) return j;
        }
        else{
            j+=i-Table[i];
            if(i>0) i=Table[i];
        }
    }
    return Search.size();
}
template<class V> 
vector<int> Zalgo(V s){
    vector<int> A(s.size(),0);
    A[0]=s.size();
    int i = 1, j = 0;
    while (i < s.size()) {
        while (i+j < s.size() && s[j] == s[i+j]) ++j;
        A[i] = j;
    if (j == 0) { ++i; continue;}
    int k = 1;
    while (i+k < s.size() && k+A[k] < j) A[i+k] = A[k], ++k;
    i += k; j -= k;
    }
    return A;
}
template<typename V>
struct rolling_hash {
    int n;
    const long long MOD[2] = {999999937LL, 1000000007LL}, base = 9973;
    vector<long long> hs[2], pw[2];
    rolling_hash(){}
    rolling_hash(const V &s) {
        n = s.size();
        for (int i = 0; i < 2; i++) {
            hs[i].assign(n+1,0);
            pw[i].assign(n+1,0);
            hs[i][0] = 0;
            pw[i][0] = 1;
            for (int j = 0; j < n; j++) {
                pw[i][j+1] = pw[i][j]*base%MOD[i];
                hs[i][j+1] = (hs[i][j]*base+s[j])%MOD[i];
            }
        }
    }

    long long hash(int l, int r, int i) { return ((hs[i][r]-hs[i][l]*pw[i][r-l])%MOD[i]+MOD[i])%MOD[i]; }

    bool match(int l1, int r1, int l2, int r2) {
        bool ret = 1;
        for (int i = 0; i < 2; i++) ret &= hash(l1,r1,i)==hash(l2,r2,i);
        return ret;
    }

    bool match(int l, int r, long long h[]) {
        bool ret = 1;
        for (int i = 0; i < 2; i++) ret &= hash(l,r,i)==h[i];
        return ret;
    }
};
template<typename V>
struct RH{
    using u64 = std::uint_fast64_t;
    const u64 MASK30 = (1<<30)-1,MASK31 = (1LL<<31) -1, MOD = (1LL<<61)-1;
    const u64 Posint = MOD*3;
    const u64 base = 9973;
    std::vector<u64> Hash,POW;
    RH(V s){
        int n=s.size();
        Hash.resize(n+1);
        POW.resize(n+1,1);
        for(size_t i=0;i<n;++i) {
            Hash[i+1] = CalcMod(Mul(Hash[i],base)+s[i]);
            POW[i+1] = CalcMod(Mul(POW[i],base));
        }
    }
    u64 hash(size_t l,size_t r){return CalcMod( Hash[r] + Posint - Mul(Hash[l],POW[r-l]) );}
    bool match(size_t l1,size_t r1,size_t l2,size_t r2){return hash(l1,r1)==hash(l2,r2);}
    u64 Mul(u64 a,u64 b){
        u64 au = a>>31;
        u64 ad = a&MASK31;
        u64 bu = b>>31;
        u64 bd = b&MASK31;
        u64 mid = ad*bu + au*bd;
        u64 midu = mid>>30;
        u64 midd = mid&MASK30;
        return au*bu*2 + midu + (midd<<31) + ad*bd;
    }
    u64 CalcMod(u64 val){
        val = (val & MOD) + (val>>61);
        if(val>=MOD) val-=MOD;
        return val;
    }
};

class mint {
 private:
  ll _num,_mod=mod;
  mint set(ll num){ 
      _num = num ;
      if(_num<0){
          if(_num>=-_mod)_num=_mod+_num;
          else _num=_mod-(-_num)%_mod;
      }
      else if(_num>=_mod) _num%=_mod;
      return *this;
  }
  ll imod()const{
    ll n=_mod-2;
    ll ans = 1,x=_num;
    while(n != 0){
        if(n&1) ans = ans*x%_mod;
        x = x*x%_mod;
        n = n >> 1;
    }
    return ans;
  }
 public:
  explicit mint(){ _num = 0; }
  explicit mint(ll num){
      _num = num;
      if(_num<0){
          if(_num>=-_mod)_num=_mod+_num;
          else _num=_mod-(-_num)%_mod;
      }
      else if(_num>=_mod) _num%=_mod;
  }
  explicit mint(ll num,ll M){
      _mod=M;
      _num=num;
      if(_num<0){
          if(_num>=-_mod)_num=_mod+_num;
          else _num=_mod-(-_num)%_mod;
      }
      else if(_num>=_mod) _num%=_mod;
  }
  mint(const mint &cp){_num=cp._num;_mod=cp._mod;}
  
  mint operator+ (const mint &x)const{ return mint(_num + x._num , _mod); }
  mint operator- (const mint &x)const{ return mint(_num - x._num , _mod);}
  mint operator* (const mint &x)const{ return mint(_num * x._num , _mod); }
  mint operator/ (const mint &x)const{ return mint(_num * x.imod() , _mod);}
  
  mint operator+=(const mint &x){ return set(_num + x._num); }
  mint operator-=(const mint &x){ return set(_num - x._num); }
  mint operator*=(const mint &x){ return set(_num * x._num); }
  mint operator/=(const mint &x){ return set(_num * x.imod());}

  mint operator= (const ll x){ return set(x); }
  mint operator+ (const ll x)const{return *this + mint(x,_mod); }
  mint operator- (const ll x)const{ return *this - mint(x,_mod); }
  mint operator* (const ll x)const{ return *this * mint(x,_mod); }
  mint operator/ (const ll x)const{ return *this/mint(x);}

  mint operator+=(const ll x){ *this = *this + x;return *this; }
  mint operator-=(const ll x){ *this = *this - x;return *this; }
  mint operator*=(const ll x){ *this = *this * x;return *this;}
  mint operator/=(const ll x){ *this = *this / x;return *this;}

  bool operator==(const mint &x)const{return _num==x._num;}
  bool operator!=(const mint &x)const{return _num!=x._num;}

  friend mint operator+(ll x,const mint &m){return mint(m._num + x , m._mod);}
  friend mint operator-(ll x,const mint &m){return mint( x - m._num , m._mod);}
  friend mint operator*(ll x,const mint &m){return mint(m._num * (x % m._mod) , m._mod);}
  friend mint operator/(ll x,const mint &m){return mint(m.imod() * (x % m._mod) , m._mod);}

  explicit operator ll() { return _num; }
  explicit operator int() { return (int)_num; }
  
  friend std::ostream& operator<<(std::ostream &os, const mint &x){ os << x._num; return os; }
  friend std::istream& operator>>(std::istream &is, mint &x){ll val; is>>val; x.set(val); return is;}
};
ll inv_mod(ll a,ll _mod){return (ll)Math::pow(mint(a,_mod),_mod-2);}
class Factorial{
 private:
    vector<ll> fac;
    vector<ll> ifac;
 public:
    
    Factorial(ll N){
        fac.reserve(N+1);
        fac.push_back(1);
        REP(i,N) fac.push_back(fac[i]*(i+1)%mod);
        ifac.resize(N+1);
        ifac[N]=inv_mod(fac[N],mod);
        REP(i,N) ifac[N-1-i]=(ifac[N-i]*(N-i))%mod;
    }

    ll fact(ll a){return fac[a];}
    ll ifact(ll a){return ifac[a];}

    ll cmb(ll a,ll b){
        if(a==0&&b==0) return 1;
        if(a<b||a<0||b<0) return 0;
        ll tmp =ifact(a-b)*ifact(b)%mod;
        return tmp*fac[a]%mod;
    }
    ll per(ll a,ll b){
        if(a==0&&b==0) return 1;
        if(a<b||a<0||b<0) return 0;
        return fac[a]*ifac[a-b]%mod;
    }
};
struct rational {
  ll p, q;
  void normalize() { // keep q positive
    if (q < 0) p *= -1, q *= -1;
    ll d = Math::gcd(p < 0 ? -p : p, q);
    if (d == 0) p = 0,  q = 1;
    else        p /= d, q /= d;
  }
  rational(ll p, ll q = 1) : p(p), q(q) {
    normalize();
  }
  rational &operator+=(const rational &a){p = a.q * p + a.p * q; q = a.q * q; normalize();return *this;}
  rational &operator-=(const rational &a){p = a.q * p - a.p * q; q = a.q * q; normalize();return *this;}
  rational &operator*=(const rational &a){p *= a.p; q *= a.q;normalize();return *this;}
  rational &operator/=(const rational &a){p *= a.q; q *= a.p; normalize();return *this;}
  rational &operator-(){p *= -1;return *this;}
  friend rational operator+(const rational &a, const rational &b){return rational(a) += b;}
  friend rational operator*(const rational &a, const rational &b){return rational(a) *= b;}
  friend rational operator-(const rational &a, const rational &b){return rational(a)-=b;}
  friend rational operator/(const rational &a, const rational &b){return rational(a) /= b;}
  friend bool operator<(const rational &a, const rational &b){ // avoid overflow 
    return (long double) a.p * b.q < (long double) a.q * b.p;
  }
  friend bool operator<=(const rational &a, const rational &b){return !(b < a);}
  friend bool operator>(const rational &a, const rational &b){return b < a;}
  friend bool operator>=(const rational &a, const rational &b){return !(a < b);}
  friend bool operator==(const rational &a, const rational &b){return !(a < b) && !(b < a);}
  friend bool operator!=(const rational &a, const rational &b){return (a < b) || (b < a);}
  friend std::ostream& operator<<(std::ostream &os, const rational &x){ printf("%.16f",(double)x.p/(double)x.q); return os; }
  friend std::istream& operator>>(std::istream &is, rational &x){is>>x.p>>x.q; x.normalize(); return is;}
};
template<typename T> class MAT{
 private:
    int row,col;
    vector<vector<T>> _A;
    double eps = 1e-9;
    MAT set(vector<vector<T>> A){_A = A ; return *this;}
 public:
    MAT(){ }
    MAT(int n,int m=0,T x=T(0)){
        if(n<1 || m<0){cout << "err Matrix::Matrix" <<endl;exit(1);}
        row = n;
        col = m?m:n;//return E if m=0
        REP(i,row){
            vector<T> a(col,x);
            _A.push_back(a);
        }
        if(m==0) REP(i,n) _A[i][i]=1.0;
    }
    MAT(vector<vector<T>> A){row=A.size();col=A[0].size();_A=A;}
    MAT(const MAT &cp){_A=cp._A;row=cp.row;col=cp.col;}
    T* operator[] (int i){return _A[i].data();}
    MAT operator= (vector<vector<T>> x) {return set(x);}
    MAT operator+ (MAT x) const {
        if(row!=x.row || col!=x.col){
            cerr << "err Matrix::operator+" <<endl;
            cerr << "  not equal matrix size" <<endl;
            exit(0);
        }
        MAT r(row, col);
        REP(i,row) REP(j,col) r[i][j]=_A[i][j]+x[i][j];
        return r;
    }
    MAT operator- (MAT x) const {
        if(row!=x.row || col!=x.col){
            cerr << "err Matrix::operator-" <<endl;
            cerr << "  not equal matrix size" <<endl;
            exit(0);
        }
        MAT r(row, col);
        REP(i,row) REP(j,col) r[i][j]=_A[i][j]-x[i][j];
        return r;
    }
    MAT operator* (MAT x) const {
        if(x.col==1&&x.row==1) return x[0][0]*MAT(_A);
        if(row==1&&col==1) return _A[0][0]*x;
        if(col!=x.row){
            cerr << "err Matrix::operator*" <<endl;
            cerr << "  not equal matrix size" <<endl;
            exit(0);
        }
        MAT r(row, x.col);
        REP(i,row) REP(j,x.col) REP(k,col) r[i][j]+=_A[i][k]*x[k][j];
        return r;
    }
    MAT operator/(MAT x)const{*this = *this * x.inverse(); return *this;}
    MAT operator/ (T a)const{
        MAT r(row,col);
        REP(i,row) REP(j,col) r[i][j]=_A[i][j]/a;
        return r;
    }
    MAT operator+= (MAT x) {*this = *this + x;return *this;}
    MAT operator-= (MAT x) {*this = *this - x; return *this;}
    MAT operator*=(T a){*this = a*(*this); return this;}
    MAT operator/=(MAT x){*this = *this/x;return *this;}
    MAT operator/=(T a){*this = *this/a; return *this;}
    friend MAT operator* (T n,MAT x){
        MAT r(x.row,x.col);
        REP(i,x.row) REP(j,x.col) r[i][j]=n*x[i][j];
        return r;
    }
    friend MAT operator* (MAT x,T n){
        MAT r(x.row,x.col);
        REP(i,x.row) REP(j,x.col) r[i][j]=n*x[i][j];
        return r;
    }
    explicit operator vector<vector<T>>(){return _A;}
    friend std::ostream &operator<<(std::ostream &os,const MAT &x){ REP(i,x.row) REP(j,x.col) os<<x._A[i][j]<<" \n"[j==x.col-1]; return os;}
    friend std::istream &operator>>(std::istream &is,MAT &x){REP(i,x.row) REP(j,x.col) is>>x._A[i][j];return is;}
    size_t size_row()const{return row;}
    size_t size_col()const{return col;}
    MAT transpose()const{
        MAT r(col,row);
        REP(i,col) REP(j,row) r[i][j]=_A[j][i];
        return r;
    }
    MAT inverse()const{
        T buf;
        MAT<T> inv_a(row,0);
        vector<vector<T>> a=_A;
        //row reduction
        REP(i,row){
            buf=1/a[i][i];
            REP(j,row){
                a[i][j]*=buf;
                inv_a[i][j]*=buf;
            }
            REP(j,row){
                if(i!=j){
                    buf=a[j][i];
                    REP(k,row){
                        a[j][k]-=a[i][k]*buf;
                        inv_a[j][k]-=inv_a[i][k]*buf;
                    }
                }
            }
        }
        return inv_a;
    }
    MAT Jacobi(MAT b)const{//ヤコビ法によって解を求める
        size_t sz=row;
        MAT D(sz,sz),inD(sz,sz),H(sz,sz),N(sz,sz);
        MAT c(sz,1),x(sz,1),tmp(sz,1);
        //cout<<"initialized"<<endl;
        REP(i,sz){//値の初期化、Aを対角要素とそれ以外に分解する(できてる)
            REP(j,sz){
                H[i][j] = 0;
                if(i==j){
                    D[i][j] = _A[i][j];
                    inD[i][j] = 1/_A[i][j];
                    N[i][j]=0;
                }
                else if(i!=j){
                    D[i][j] = 0;
                    inD[i][j] = 0;
                    N[i][j]=_A[i][j];
                }
            }
            c[i][0] = 0;
            x[i][0] = 1;
        }
        c=inD*b;
        H=inD*N;
        while(1){//反復法ステップ1→2→1...
            tmp=x;
            x=c-H*x;
            T r=T(0);
            for(int i=0;i<row;++i){
                r+=(x[i][0]-tmp[i][0])*(x[i][0]-tmp[i][0]);
            }
            if(r<eps) break;
        }
        return x;
    }
    MAT Gauss(MAT b)const{//ガウス・ザイデル法によって解を求める
        MAT<T> DL(row),U(row),inDL(row),H(row),c(row,1),x(row,1),tmp(row,1);
        for(int i=0;i<row;i++){
            for(int j=0;j<col;j++){
                H[i][j] = 0;
                if(i>=j){
                    DL[i][j] = _A[i][j];
                    U[i][j] = 0;
                }
                else{
                    DL[i][j] = 0;
                    U[i][j] = _A[i][j];
                }
            }
            x[i][0] = 1;
        }
        inDL=DL.inverse();
        c=inDL*b;
        H=inDL*U;
        int n=0;
        while(1){
            tmp=x;
            x=c-H*x;
            T r = T(0);
            for(int i=0;i<row;++i){
                r+=(x[i][0]-tmp[i][0])*(x[i][0]-tmp[i][0]);
            }
            n++;
            if(r<eps) break;
        }
        return x;
    }
    int rank()const{// O( n^3 )
        vector<vector<T>> A=_A;
        const int n = row, m = col;
        int r = 0;
        for(int i = 0; r < n && i < m; ++i) {
            int pivot = r;
            for(int j = r+1; j < n; ++j) if(fabs(A[j][i]) > fabs(A[pivot][i])) pivot = j;
            swap(A[pivot], A[r]);
            if(fabs(A[r][i]) < eps) continue;
            for (int k = m-1; k >= i; --k) A[r][k] /= A[r][i];
            rep(j,r+1,n) rep(k,i,m) A[j][k] -= A[r][k] * A[j][i];
            ++r;
        }
        return r;
    }
};

class UnionFind{
 private:
    vector<int> Parent,es;
    vector<ll> diff_weight;
 public:
    UnionFind(int N){
        es.resize(N,0);
        Parent.resize(N,-1);
        diff_weight.resize(N,0LL);
    }

    int root(int A){
        if(Parent[A]<0) return A;
        else{ 
            int r = root(Parent[A]);
            diff_weight[A] += diff_weight[Parent[A]]; // 累積和をとる
            return Parent[A]=r;
        }
    }
    bool issame(int A,int B){
        return root(A)==root(B);
    }
    ll weight(int x) {
        root(x); // 経路圧縮
        return diff_weight[x];
    }
    ll diff(int x, int y) {
        return weight(y) - weight(x);
    }
    int size(int A){
        return -Parent[root(A)];
    }
    int eize(int A){
        return es[root(A)];
    }
    bool connect(int A,int B){
        A=root(A); B=root(B);
        if(A==B) return false;
        if(size(A)<size(B)) std::swap(A,B);
        Parent[A]+=Parent[B];
        es[A]+=es[B]+1;
        Parent[B]=A;
        return true;
    }
    void unite(int A,int B){
        A=root(A); B=root(B);
        if(A==B){ 
            es[A]++;
            return;
        }
        if(size(A)<size(B)) std::swap(A,B);
        Parent[A]+=Parent[B];
        es[A]+=es[B]+1;
        Parent[B]=A;
        return;
    }
    bool merge(int A, int B, ll w) {
        // x と y それぞれについて、 root との重み差分を補正
        w += weight(A); w -= weight(B); 
        A=root(A); B=root(B);
        if(A==B) return false;
        if(size(A)<size(B)) std::swap(A,B),w=-w;
        Parent[A]+=Parent[B];
        Parent[B]=A;
        // x が y の親になるので、x と y の差分を diff_weight[y] に記録
        diff_weight[B] = w; 
        return true;
    }
};
struct edge{
    int from;int to;ll cost;
    void push(int a,int b,int c){
        from=a;to=b;cost=c;
    }
    bool operator<(const edge& y)const{
        if(cost!=y.cost) return cost<y.cost;
        else if(to!=y.to) return to<y.to;
        else return from<y.from;}
    bool operator>(const edge& y)const{
        if(cost!=y.cost) return cost>y.cost;
        else if(to!=y.to) return to>y.to;
        else return from>y.from;}
    bool operator==(const edge& y) const{return !(*this<y)&&!(*this>y);}
};
class lca {
  public:
    using Graph = vector<vector<int>>;
    const int n = 0;
    const int log2_n = 0;
    std::vector<std::vector<int>> parent;
    std::vector<int> depth;

    lca() {}

    lca(const Graph &g, int root)
        : n(g.size()), log2_n(log2(n) + 1), parent(log2_n, std::vector<int>(n)), depth(n) {
        dfs(g, root, -1, 0);
        for (int k = 0; k + 1 < log2_n; k++) {
            for (int v = 0; v < (int)g.size(); v++) {
                if (parent[k][v] < 0)
                    parent[k + 1][v] = -1;
                else
                    parent[k + 1][v] = parent[k][parent[k][v]];
            }
        }
    }

    void dfs(const Graph &g, int v, int p, int d) {
        parent[0][v] = p;
        depth[v] = d;
        REP(j,g[v].size()) {
            if (g[v][j] != p) dfs(g, g[v][j], v, d + 1);
        }
    }

    int get(int u, int v) {
        if (depth[u] > depth[v]) std::swap(u, v);
        for (int k = 0; k < log2_n; k++) {
            if ((depth[v] - depth[u]) >> k & 1) {
                v = parent[k][v];
            }
        }
        if (u == v) return u;
        for (int k = log2_n - 1; k >= 0; k--) {
            if (parent[k][u] != parent[k][v]) {
                u = parent[k][u];
                v = parent[k][v];
            }
        }
        return parent[0][u];
    }
};

class SOSU{
 private:
    vector<int> Prime_Number;
    vector<bool> isp;
 public:
    SOSU(int N){
        isp.resize(N+1,true);
        isp[0]=isp[1]=false;
        rep(i,2,N+1) if(isp[i]){
            Prime_Number.push_back(i);
            for(int j=2*i;j<=N;j+=i) isp[j]=false;
        }
    }
    int operator[](int i){return Prime_Number[i];}
    int size(){return Prime_Number.size();}
    int back(){return Prime_Number.back();}
    bool isPrime(int q){return isp[q];}
};
class Divisor{//素因数分解をしてから約数列挙、分解結果はＰ(底,指数)でpfacにまとめている
  private:
    vector<ll> F;
    vector<P> pfactorize;
  public:
    Divisor(ll N){
        for(ll i = 1; i * i <= N; i++) {
            if(N % i == 0) {
                F.push_back(i);
                if(i * i != N) F.push_back(N / i);
            }
        }
        sort(begin(F), end(F));
		SOSU p(Math::sqrt(N)+1);
    	REP(i,p.size()){
    		pfactorize.push_back(P(p[i],0));
    		while(N%p[i]==0){
    			N/=p[i];
    			pfactorize.back().second++;
    		}
            if(pfactorize.back().second==0) pfactorize.pop_back();
    	}
    	if(N>1) pfactorize.push_back(P(N,1));
    }
    int size(){return F.size();}
    vector<P> pfac(){return pfactorize;}
    ll operator[](int k){return F[k];}
};

template<typename T>
struct compress{
    std::map<T,int> zip;
    vector<T> unzip;
    compress(vector<T> x){
        unzip=x;
        sort(All(x));
        x.erase(unique(All(x)),x.end());
        REP(i,x.size()){
            zip[x[i]]=i;
        }
    }
    size_t operator[](int k){return zip[unzip[k]];}
};

template<typename T> class SegmentTree{
  private:
    typedef std::function<T(T,T)> F;
    int n;
    T d0;
    vector<T> vertex;
    F f;
    F g;
  public:
    SegmentTree(F f,F g,T d):d0(d),f(f),g(g){}
    void init(int _n){
        n=1;
        while(n<_n) n*=2;
        vertex.resize(2*n-1,d0);
    }
    void build(const vector<T> &v){
        int n_=v.size();
        init(n_);
        for(int i=0;i<n_;i++) vertex[n+i-1]=v[i];
        for(int i=n-2;i>=0;i--)
        vertex[i]=f(vertex[2*i+1],vertex[2*i+2]);
    }
    void update(int i,T x){
        int k=i+n-1;
        vertex[k]=g(vertex[k],x);
        while(k>0){
            k=(k-1)/2;
            vertex[k]=f(vertex[2*k+1],vertex[2*k+2]);
        }
        return;
    }
    T query(int l,int r){
        T vl=d0,vr=d0;
        l+=n-1;
        r+=n-1;
        for(;l<=r;l/=2,r=r/2-1){
            if(l%2==0) vl=f(vl,vertex[l]);
            if(r&1) vr=f(vr,vertex[r]);
        }
        return f(vl,vr);
    }
};
template <typename T,typename E>
struct LazySegmentTree{
  using F = std::function<T(T,T)>;
  using G = std::function<T(T,E)>;
  using H = std::function<E(E,E)>;
  int n,height;
  F f;
  G g;
  H h;
  T ti;
  E ei;
  vector<T> dat;
  vector<E> laz;
  LazySegmentTree(F f,G g,H h,T ti,E ei):
    f(f),g(g),h(h),ti(ti),ei(ei){}

  void init(int n_){
    n=1;height=0;
    while(n<n_) n<<=1,height++;
    dat.assign(2*n,ti);
    laz.assign(2*n,ei);
  }

  void build(const vector<T> &v){
    int n_=v.size();
    init(n_);
    for(int i=0;i<n_;i++) dat[n+i]=v[i];
    for(int i=n-1;i;i--)
      dat[i]=f(dat[(i<<1)|0],dat[(i<<1)|1]);
  }

  inline T reflect(int k){
    return laz[k]==ei?dat[k]:g(dat[k],laz[k]);
  }

  inline void propagate(int k){
    if(laz[k]==ei) return;
    laz[(k<<1)|0]=h(laz[(k<<1)|0],laz[k]);
    laz[(k<<1)|1]=h(laz[(k<<1)|1],laz[k]);
    dat[k]=reflect(k);
    laz[k]=ei;
  }

  inline void thrust(int k){
    for(int i=height;i;i--) propagate(k>>i);
  }

  inline void recalc(int k){
    while(k>>=1)
      dat[k]=f(reflect((k<<1)|0),reflect((k<<1)|1));
  }

  void update(int a,int b,E x){
    thrust(a+=n);
    thrust(b+=n-1);
    for(int l=a,r=b+1;l<r;l>>=1,r>>=1){
      if(l&1) laz[l]=h(laz[l],x),l++;
      if(r&1) --r,laz[r]=h(laz[r],x);
    }
    recalc(a);
    recalc(b);
  }

  void set_val(int a,T x){
    thrust(a+=n);
    dat[a]=x;laz[a]=ei;
    recalc(a);
  }

  T query(int a,int b){
    thrust(a+=n);
    thrust(b+=n-1);
    T vl=ti,vr=ti;
    for(int l=a,r=b+1;l<r;l>>=1,r>>=1) {
      if(l&1) vl=f(vl,reflect(l++));
      if(r&1) vr=f(reflect(--r),vr);
    }
    return f(vl,vr);
  }

  template<typename C>
  int find(int st,C &check,T &acc,int k,int l,int r){
    if(l+1==r){
      acc=f(acc,reflect(k));
      return check(acc)?k-n:-1;
    }
    propagate(k);
    int m=(l+r)>>1;
    if(m<=st) return find(st,check,acc,(k<<1)|1,m,r);
    if(st<=l&&!check(f(acc,dat[k]))){
      acc=f(acc,dat[k]);
      return -1;
    }
    int vl=find(st,check,acc,(k<<1)|0,l,m);
    if(~vl) return vl;
    return find(st,check,acc,(k<<1)|1,m,r);
  }

  template<typename C>
  int find(int st,C &check){
    T acc=ti;
    return find(st,check,acc,1,0,n);
  }
};

template<typename T>
class FFT{
  private:
    using Complex = std::complex<double>;
    std::vector<Complex> C;
    void DFT(std::vector<Complex> &F,size_t n,int sig=1){
        if(n==1) return;
        std::vector<Complex> f0(n/2),f1(n/2);
        for(size_t i=0;i<n/2;++i){
            f0[i]=F[2*i];
            f1[i]=F[2*i+1];
        }
        DFT(f0,n/2,sig);
        DFT(f1,n/2,sig);
        Complex z(cos(2.0*PI/n),sin(2.0*PI/n)*sig),zi=1;
        for(size_t i=0;i<n;++i){
            if(i<n/2) F[i]=f0[i]+zi*f1[i];
            else F[i]=f0[i-n/2]+zi*f1[i-n/2];
            zi*=z;
        }
        return;
    }
    void invDFT(std::vector<Complex> &f,size_t n){
        DFT(f,n,-1);
        for(size_t i=0;i<n;++i){
            f[i]/=n;
        }
        return;
    }
  public:
    FFT(const std::vector<T> &A,const std::vector<T> &B){
        size_t n=1;
        while(n<=A.size()+B.size()){
            n*=2;
        }
        std::vector<Complex> g(n,Complex(0));
        C.resize(n,Complex(0));
        size_t i_len=std::max(A.size(),B.size());
        for(size_t i=0;i<i_len;++i){
            if(i<A.size()) C[i]=Complex(A[i]);
            if(i<B.size()) g[i]=Complex(B[i]);
        }
        DFT(C,n);
        DFT(g,n);
        for(size_t i=0;i<n;++i) C[i]=C[i]*g[i];
        invDFT(C,n);
        for(size_t i=0;i<n;++i) if(T(C[i].real())!=C[i].real()){
            C[i]=Complex(C[i].real()+0.5);
        }
    }
    T operator[](int k)const{return T(C[k].real());}
};

int main(){
    cin.tie(nullptr);
    std::ios::sync_with_stdio(false);
    solve();
    return 0;
}


void solve(){
    ll N,M,K;
    mint p,q;
    cin>>N>>M>>K>>p>>q;
    vector<mint> b(N);
    REP(i,N) cin>>b[i];
    mint ans(0);
    mint z=(Math::pow((q-2*p)/q,K)+1)/mint(2);
    REP(i,N){
        if(i<M) ans+=b[i]*z;
        else ans+=b[i]*(1-z);
    }
    cout<<ans<<endl;
}
yukicoder

結果

テストケース

ソースコード
