#939039 (C++17(gcc12)) No.2575 Almost Increasing Sequence

提出ソース
結果

問題	No.2575 Almost Increasing Sequence
コンテスト
ユーザー	maroon_kuri
提出日時	2023-12-18 22:42:15
言語	C++17(gcc12) (gcc 12.3.0 + boost 1.87.0)
結果	AC
実行時間	2,606 ms / 10,000 ms
コード長	31,879 bytes
コンパイル時間	7,086 ms
コンパイル使用メモリ	325,660 KB
実行使用メモリ	87,900 KB
スコア	600,000
最終ジャッジ日時	2023-12-18 22:43:19
合計ジャッジ時間	63,602 ms
ジャッジサーバーID （参考情報）	judge15 / judge12
	純コード判定しない問題か言語
このコードへのチャレンジ
（要ログイン）
ファイルパターン	結果
sample	AC * 1
other	AC * 19
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
#ifndef LOCAL
#pragma GCC optimize ("Ofast")
#pragma GCC optimize ("unroll-loops")
#endif

#include <bits/stdc++.h>
using namespace std;

using ll=long long;
#define int ll

#define rng(i,a,b) for(int i=int(a);i<int(b);i++)
#define rep(i,b) rng(i,0,b)
#define gnr(i,a,b) for(int i=int(b)-1;i>=int(a);i--)
#define per(i,b) gnr(i,0,b)
#define pb push_back
#define eb emplace_back
#define a first
#define b second
#define bg begin()
#define ed end()
#define all(x) x.bg,x.ed
#define si(x) int(x.size())
#ifdef LOCAL
#define dmp(x) cerr<<__LINE__<<" "<<#x<<" "<<x<<endl
#else
#define dmp(x) void(0)
#endif

template<class t,class u> bool chmax(t&a,u b){if(a<b){a=b;return true;}else return false;}
template<class t,class u> bool chmin(t&a,u b){if(b<a){a=b;return true;}else return false;}

template<class t> using vc=vector<t>;
template<class t> using vvc=vc<vc<t>>;

using pi=pair<int,int>;
using vi=vc<int>;

template<class t,class u>
ostream& operator<<(ostream& os,const pair<t,u>& p){
	return os<<"{"<<p.a<<","<<p.b<<"}";
}

template<class t> ostream& operator<<(ostream& os,const vc<t>& v){
	os<<"{";
	for(auto e:v)os<<e<<",";
	return os<<"}";
}

#define mp make_pair
#define mt make_tuple
#define one(x) memset(x,-1,sizeof(x))
#define zero(x) memset(x,0,sizeof(x))
#ifdef LOCAL
void dmpr(ostream&os){os<<endl;}
template<class T,class... Args>
void dmpr(ostream&os,const T&t,const Args&... args){
	os<<t<<" ";
	dmpr(os,args...);
}
#define dmp2(...) dmpr(cerr,__LINE__,##__VA_ARGS__)
#else
#define dmp2(...) void(0)
#endif

using uint=unsigned;
using ull=unsigned long long;

template<class t,size_t n>
ostream& operator<<(ostream&os,const array<t,n>&a){
	return os<<vc<t>(all(a));
}

template<int i,class T>
void print_tuple(ostream&,const T&){
}

template<int i,class T,class H,class ...Args>
void print_tuple(ostream&os,const T&t){
	if(i)os<<",";
	os<<get<i>(t);
	print_tuple<i+1,T,Args...>(os,t);
}

template<class ...Args>
ostream& operator<<(ostream&os,const tuple<Args...>&t){
	os<<"{";
	print_tuple<0,tuple<Args...>,Args...>(os,t);
	return os<<"}";
}

ll read(){
	ll i;
	cin>>i;
	return i;
}

vi readvi(int n,int off=0){
	vi v(n);
	rep(i,n)v[i]=read()+off;
	return v;
}

pi readpi(int off=0){
	int a,b;cin>>a>>b;
	return pi(a+off,b+off);
}

template<class t>
void print_single(t x,int suc=1){
	cout<<x;
	if(suc==1)
		cout<<"\n";
	if(suc==2)
		cout<<" ";
}

template<class t,class u>
void print_single(const pair<t,u>&p,int suc=1){
	print_single(p.a,2);
	print_single(p.b,suc);
}

template<class T>
void print_single(const vector<T>&v,int suc=1){
	rep(i,v.size())
		print_single(v[i],i==int(v.size())-1?suc:2);
}

template<class T>
void print_offset(const vector<T>&v,ll off,int suc=1){
	rep(i,v.size())
		print_single(v[i]+off,i==int(v.size())-1?suc:2);
}

template<class T,size_t N>
void print_single(const array<T,N>&v,int suc=1){
	rep(i,N)
		print_single(v[i],i==int(N)-1?suc:2);
}

template<class T>
void print(const T&t){
	print_single(t);
}

template<class T,class ...Args>
void print(const T&t,const Args&...args){
	print_single(t,2);
	print(args...);
}

string readString(){
	string s;
	cin>>s;
	return s;
}

template<class T>
T sq(const T& t){
	return t*t;
}

void YES(bool ex=true){
	cout<<"YES\n";
	if(ex)exit(0);
	#ifdef LOCAL
	cout.flush();
	#endif
}
void NO(bool ex=true){
	cout<<"NO\n";
	if(ex)exit(0);
	#ifdef LOCAL
	cout.flush();
	#endif
}
void Yes(bool ex=true){
	cout<<"Yes\n";
	if(ex)exit(0);
	#ifdef LOCAL
	cout.flush();
	#endif
}
void No(bool ex=true){
	cout<<"No\n";
	if(ex)exit(0);
	#ifdef LOCAL
	cout.flush();
	#endif
}
//#define CAPITAL
/*
void yes(bool ex=true){
	#ifdef CAPITAL
	cout<<"YES"<<"\n";
	#else
	cout<<"Yes"<<"\n";
	#endif
	if(ex)exit(0);
	#ifdef LOCAL
	cout.flush();
	#endif
}
void no(bool ex=true){
	#ifdef CAPITAL
	cout<<"NO"<<"\n";
	#else
	cout<<"No"<<"\n";
	#endif
	if(ex)exit(0);
	#ifdef LOCAL
	cout.flush();
	#endif
}*/
void possible(bool ex=true){
	#ifdef CAPITAL
	cout<<"POSSIBLE"<<"\n";
	#else
	cout<<"Possible"<<"\n";
	#endif
	if(ex)exit(0);
	#ifdef LOCAL
	cout.flush();
	#endif
}
void impossible(bool ex=true){
	#ifdef CAPITAL
	cout<<"IMPOSSIBLE"<<"\n";
	#else
	cout<<"Impossible"<<"\n";
	#endif
	if(ex)exit(0);
	#ifdef LOCAL
	cout.flush();
	#endif
}

constexpr ll ten(int n){
	return n==0?1:ten(n-1)*10;
}

const ll infLL=LLONG_MAX/3;

#ifdef int
const int inf=infLL;
#else
const int inf=INT_MAX/2-100;
#endif

int topbit(signed t){
	return t==0?-1:31-__builtin_clz(t);
}
int topbit(ll t){
	return t==0?-1:63-__builtin_clzll(t);
}
int topbit(ull t){
	return t==0?-1:63-__builtin_clzll(t);
}
int botbit(signed a){
	return a==0?32:__builtin_ctz(a);
}
int botbit(ll a){
	return a==0?64:__builtin_ctzll(a);
}
int botbit(ull a){
	return a==0?64:__builtin_ctzll(a);
}
int popcount(signed t){
	return __builtin_popcount(t);
}
int popcount(ll t){
	return __builtin_popcountll(t);
}
int popcount(ull t){
	return __builtin_popcountll(t);
}
int bitparity(ll t){
	return __builtin_parityll(t);
}
bool ispow2(int i){
	return i&&(i&-i)==i;
}
ll mask(int i){
	return (ll(1)<<i)-1;
}
ull umask(int i){
	return (ull(1)<<i)-1;
}
ll minp2(ll n){
	if(n<=1)return 1;
	else return ll(1)<<(topbit(n-1)+1);
}

bool inc(int a,int b,int c){
	return a<=b&&b<=c;
}

template<class t> void mkuni(vc<t>&v){
	sort(all(v));
	v.erase(unique(all(v)),v.ed);
}

ll rand_int(ll l, ll r) { //[l, r]
	//#ifdef LOCAL
	static mt19937_64 gen;
	/*#else
	static mt19937_64 gen(chrono::steady_clock::now().time_since_epoch().count());
	#endif*/
	return uniform_int_distribution<ll>(l, r)(gen);
}

ll rand_int(ll k){ //[0,k)
	return rand_int(0,k-1);
}

template<class t>
void myshuffle(vc<t>&a){
	rep(i,si(a))swap(a[i],a[rand_int(0,i)]);
}

template<class t,class u>
int lwb(const vc<t>&v,const u&a){
	return lower_bound(all(v),a)-v.bg;
}
template<class t,class u>
bool bis(const vc<t>&v,const u&a){
	return binary_search(all(v),a);
}

vvc<int> readGraph(int n,int m){
	vvc<int> g(n);
	rep(i,m){
		int a,b;
		cin>>a>>b;
		//sc.read(a,b);
		a--;b--;
		g[a].pb(b);
		g[b].pb(a);
	}
	return g;
}

vvc<int> readTree(int n){
	return readGraph(n,n-1);
}

template<class t>
vc<t> presum(const vc<t>&a){
	vc<t> s(si(a)+1);
	rep(i,si(a))s[i+1]=s[i]+a[i];
	return s;
}
vc<ll> presum(const vi&a){
	vc<ll> s(si(a)+1);
	rep(i,si(a))s[i+1]=s[i]+a[i];
	return s;
}
//BIT で数列を管理するときに使う (CF850C)
template<class t>
vc<t> predif(vc<t> a){
	gnr(i,1,si(a))a[i]-=a[i-1];
	return a;
}
template<class t>
vvc<ll> imos(const vvc<t>&a){
	int n=si(a),m=si(a[0]);
	vvc<ll> b(n+1,vc<ll>(m+1));
	rep(i,n)rep(j,m)
		b[i+1][j+1]=b[i+1][j]+b[i][j+1]-b[i][j]+a[i][j];
	return b;
}

//verify してないや
void transvvc(int&n,int&m){
	swap(n,m);
}
template<class t,class... Args>
void transvvc(int&n,int&m,vvc<t>&a,Args&...args){
	assert(si(a)==n);
	vvc<t> b(m,vi(n));
	rep(i,n){
		assert(si(a[i])==m);
		rep(j,m)b[j][i]=a[i][j];
	}
	a.swap(b);
	transvvc(n,m,args...);
}
//CF854E
void rotvvc(int&n,int&m){
	swap(n,m);
}
template<class t,class... Args>
void rotvvc(int&n,int&m,vvc<t>&a,Args&...args){
	assert(si(a)==n);
	vvc<t> b(m,vi(n));
	rep(i,n){
		assert(si(a[i])==m);
		rep(j,m)b[m-1-j][i]=a[i][j];
	}
	a.swap(b);
	rotvvc(n,m,args...);
}

//ソートして i 番目が idx[i]
//CF850C
template<class t>
vi sortidx(const vc<t>&a){
	int n=si(a);
	vi idx(n);iota(all(idx),0);
	sort(all(idx),[&](int i,int j){return a[i]<a[j];});
	return idx;
}
//vs[i]=a[idx[i]]
//例えば sortidx で得た idx を使えば単にソート列になって返ってくる
//CF850C
template<class t>
vc<t> a_idx(const vc<t>&a,const vi&idx){
	int n=si(a);
	assert(si(idx)==n);
	vc<t> vs(n);
	rep(i,n)vs[i]=a[idx[i]];
	return vs;
}
//CF850C
vi invperm(const vi&p){
	int n=si(p);
	vi q(n);
	rep(i,n)q[p[i]]=i;
	return q;
}

template<class t,class s=t>
s SUM(const vc<t>&a){
	return accumulate(all(a),s(0));
}
template<class t,size_t K,class s=t>
s SUM(const array<t,K>&a){
	return accumulate(all(a),s(0));
}

template<class t>
t MAX(const vc<t>&a){
	return *max_element(all(a));
}

template<class t>
pair<t,int> MAXi(const vc<t>&a){
	auto itr=max_element(all(a));
	return mp(*itr,itr-a.bg);
}

template<class A>
auto MIN(const A&a){
	return *min_element(all(a));
}

template<class t>
pair<t,int> MINi(const vc<t>&a){
	auto itr=min_element(all(a));
	return mp(*itr,itr-a.bg);
}

vi vid(int n){
	vi res(n);iota(all(res),0);
	return res;
}

template<class S>
void soin(S&s){
	sort(all(s));
}

template<class S,class F>
void soin(S&s,F&&f){
	sort(all(s),forward<F>(f));
}

template<class S>
S soout(S s){
	soin(s);
	return s;
}

template<class S>
void rein(S&s){
	reverse(all(s));
}

template<class S>
S reout(S s){
	rein(s);
	return s;
}

template<class t,class u>
pair<t,u>&operator+=(pair<t,u>&a,pair<t,u> b){
	a.a+=b.a;a.b+=b.b;return a;}
template<class t,class u>
pair<t,u>&operator-=(pair<t,u>&a,pair<t,u> b){
	a.a-=b.a;a.b-=b.b;return a;}
template<class t,class u>
pair<t,u> operator+(pair<t,u> a,pair<t,u> b){return mp(a.a+b.a,a.b+b.b);}
template<class t,class u>
pair<t,u> operator-(pair<t,u> a,pair<t,u> b){return mp(a.a-b.a,a.b-b.b);}

template<class t>
t gpp(vc<t>&vs){
	assert(si(vs));
	t res=move(vs.back());
	vs.pop_back();
	return res;
}

template<class t,class u>
void pb(vc<t>&a,const vc<u>&b){
	a.insert(a.ed,all(b));
}

template<class t,class...Args>
vc<t> cat(vc<t> a,Args&&...b){
	(pb(a,forward<Args>(b)),...);
	return a;
}

template<class t,class u>
vc<t>& operator+=(vc<t>&a,u x){
	for(auto&v:a)v+=x;
	return a;
}

template<class t,class u>
vc<t> operator+(vc<t> a,u x){
	return a+=x;
}

template<class t,class u>
vc<t>& operator-=(vc<t>&a,u x){
	for(auto&v:a)v-=x;
	return a;
}

template<class t,class u>
vc<t>& operator-(vc<t> a,u x){
	return a-=x;
}

template<class t,class u>
void remval(vc<t>&a,const u&v){
	a.erase(remove(all(a),v),a.ed);
}

template<class VS,class u>
void fila(VS&vs,const u&a){
	fill(all(vs),a);
}

template<class t,class u>
int findid(const vc<t>&vs,const u&a){
	auto itr=find(all(vs),a);
	if(itr==vs.ed)return -1;
	else return itr-vs.bg;
}

template<class t>
void rtt(vc<t>&vs,int i){
	rotate(vs.bg,vs.bg+i,vs.ed);
}

//mint107 は verify してねえ
//#define DYNAMIC_MOD

struct modinfo{uint mod,root;
#ifdef DYNAMIC_MOD
constexpr modinfo(uint m,uint r):mod(m),root(r),im(0){set_mod(m);}
ull im;
constexpr void set_mod(uint m){
	mod=m;
	im=ull(-1)/m+1;
}
uint product(uint a,uint b)const{
	ull z=ull(a)*b;
	uint x=((unsigned __int128)z*im)>>64;
	uint v=uint(z)-x*mod;
	return v<mod?v:v+mod;
}
#endif
};
template<modinfo const&ref>
struct modular{
	static constexpr uint const &mod=ref.mod;
	static modular root(){return modular(ref.root);}
	uint v;
	//modular(initializer_list<uint>ls):v(*ls.bg){}
	modular(ll vv=0){s(vv%mod+mod);}
	modular& s(uint vv){
		v=vv<mod?vv:vv-mod;
		return *this;
	}
	modular operator-()const{return modular()-*this;}
	modular& operator+=(const modular&rhs){return s(v+rhs.v);}
	modular&operator-=(const modular&rhs){return s(v+mod-rhs.v);}
	modular&operator*=(const modular&rhs){
		#ifndef DYNAMIC_MOD
		v=ull(v)*rhs.v%mod;
		#else
		v=ref.product(v,rhs.v);
		#endif
		return *this;
	}
	modular&operator/=(const modular&rhs){return *this*=rhs.inv();}
	modular operator+(const modular&rhs)const{return modular(*this)+=rhs;}
	modular operator-(const modular&rhs)const{return modular(*this)-=rhs;}
	modular operator*(const modular&rhs)const{return modular(*this)*=rhs;}
	modular operator/(const modular&rhs)const{return modular(*this)/=rhs;}
	modular pow(ll n)const{
		if(n<0)return inv().pow(-n);
		modular res(1),x(*this);
		while(n){
			if(n&1)res*=x;
			x*=x;
			n>>=1;
		}
		return res;
	}
	modular inv()const{return pow(mod-2);}
	/*modular inv()const{
		int x,y;
		int g=extgcd<ll>(v,mod,x,y);
		assert(g==1);
		if(x<0)x+=mod;
		return modular(x);
	}*/
	friend modular operator+(ll x,const modular&y){
		return modular(x)+y;
	}
	friend modular operator-(ll x,const modular&y){
		return modular(x)-y;
	}
	friend modular operator*(ll x,const modular&y){
		return modular(x)*y;
	}
	friend modular operator/(ll x,const modular&y){
		return modular(x)/y;
	}
	friend ostream& operator<<(ostream&os,const modular&m){
		return os<<m.v;
	}
	friend istream& operator>>(istream&is,modular&m){
		ll x;is>>x;
		m=modular(x);
		return is;
	}
	bool operator<(const modular&r)const{return v<r.v;}
	bool operator==(const modular&r)const{return v==r.v;}
	bool operator!=(const modular&r)const{return v!=r.v;}
	explicit operator bool()const{
		return v;
	}
};

#define USE_GOOD_MOD

//size of input must be a power of 2
//output of forward fmt is bit-reversed
//output elements are in the range [0,mod*4)
//input of inverse fmt should be bit-reversed
template<class mint>
void inplace_fmt(const int n,mint*const f,bool inv){
	static constexpr uint mod=mint::mod;
	static constexpr uint mod2=mod*2;
	static constexpr int L=30;
	static mint g[L],ig[L],p2[L];
	if(g[0].v==0){
		rep(i,L){
			mint w=-mint::root().pow(((mod-1)>>(i+2))*3);
			g[i]=w;
			ig[i]=w.inv();
			p2[i]=mint(1<<i).inv();
		}
	}
	if(!inv){
		int b=n;
		if(b>>=1){//input:[0,mod)
			rep(i,b){
				uint x=f[i+b].v;
				f[i+b].v=f[i].v+mod-x;
				f[i].v+=x;
			}
		}
		if(b>>=1){//input:[0,mod*2)
			mint p=1;
			for(int i=0,k=0;i<n;i+=b*2){
				rng(j,i,i+b){
					uint x=(f[j+b]*p).v;
					f[j+b].v=f[j].v+mod-x;
					f[j].v+=x;
				}
				p*=g[__builtin_ctz(++k)];
			}
		}
		while(b){
			if(b>>=1){//input:[0,mod*3)
				mint p=1;
				for(int i=0,k=0;i<n;i+=b*2){
					rng(j,i,i+b){
						uint x=(f[j+b]*p).v;
						f[j+b].v=f[j].v+mod-x;
						f[j].v+=x;
					}
					p*=g[__builtin_ctz(++k)];
				}
			}
			if(b>>=1){//input:[0,mod*4)
				mint p=1;
				for(int i=0,k=0;i<n;i+=b*2){
					rng(j,i,i+b){
						uint x=(f[j+b]*p).v;
						f[j].v=(f[j].v<mod2?f[j].v:f[j].v-mod2);
						f[j+b].v=f[j].v+mod-x;
						f[j].v+=x;
					}
					p*=g[__builtin_ctz(++k)];
				}
			}
		}
	}else{
		int b=1;
		if(b<n/2){//input:[0,mod)
			mint p=1;
			for(int i=0,k=0;i<n;i+=b*2){
				rng(j,i,i+b){
					ull x=f[j].v+mod-f[j+b].v;
					f[j].v+=f[j+b].v;
					f[j+b].v=x*p.v%mod;
				}
				p*=ig[__builtin_ctz(++k)];
			}
			b<<=1;
		}
		for(;b<n/2;b<<=1){
			mint p=1;
			for(int i=0,k=0;i<n;i+=b*2){
				rng(j,i,i+b/2){//input:[0,mod*2)
					ull x=f[j].v+mod2-f[j+b].v;
					f[j].v+=f[j+b].v;
					f[j].v=(f[j].v)<mod2?f[j].v:f[j].v-mod2;
					f[j+b].v=x*p.v%mod;
				}
				rng(j,i+b/2,i+b){//input:[0,mod)
					ull x=f[j].v+mod-f[j+b].v;
					f[j].v+=f[j+b].v;
					f[j+b].v=x*p.v%mod;
				}
				p*=ig[__builtin_ctz(++k)];
			}
		}
		if(b<n){//input:[0,mod*2)
			rep(i,b){
				uint x=f[i+b].v;
				f[i+b].v=f[i].v+mod2-x;
				f[i].v+=x;
			}
		}
		mint z=p2[__lg(n)];
		rep(i,n)f[i]*=z;
	}
}

template<class mint>
void inplace_fmt(vector<mint>&f,bool inv){
	inplace_fmt(si(f),f.data(),inv);
}

//size of input must be a power of 2
//output elements are in the range [0,mod*4)
template<class mint>
void half_fmt(const int n,mint*const f){
	static constexpr uint mod=mint::mod;
	static constexpr uint mod2=mod*2;
	static const int L=30;
	static mint g[L],h[L];
	if(g[0].v==0){
		rep(i,L){
			g[i]=-mint::root().pow(((mod-1)>>(i+2))*3);
			h[i]=mint::root().pow((mod-1)>>(i+2));
		}
	}
	int b=n;
	int lv=0;
	if(b>>=1){//input:[0,mod)
		mint p=h[lv++];
		for(int i=0,k=0;i<n;i+=b*2){
			rng(j,i,i+b){
				uint x=(f[j+b]*p).v;
				f[j+b].v=f[j].v+mod-x;
				f[j].v+=x;
			}
			p*=g[__builtin_ctz(++k)];
		}
	}
	if(b>>=1){//input:[0,mod*2)
		mint p=h[lv++];
		for(int i=0,k=0;i<n;i+=b*2){
			rng(j,i,i+b){
				uint x=(f[j+b]*p).v;
				f[j+b].v=f[j].v+mod-x;
				f[j].v+=x;
			}
			p*=g[__builtin_ctz(++k)];
		}
	}
	while(b){
		if(b>>=1){//input:[0,mod*3)
			mint p=h[lv++];
			for(int i=0,k=0;i<n;i+=b*2){
				rng(j,i,i+b){
					uint x=(f[j+b]*p).v;
					f[j+b].v=f[j].v+mod-x;
					f[j].v+=x;
				}
				p*=g[__builtin_ctz(++k)];
			}
		}
		if(b>>=1){//input:[0,mod*4)
			mint p=h[lv++];
			for(int i=0,k=0;i<n;i+=b*2){
				rng(j,i,i+b){
					uint x=(f[j+b]*p).v;
					f[j].v=(f[j].v<mod2?f[j].v:f[j].v-mod2);
					f[j+b].v=f[j].v+mod-x;
					f[j].v+=x;
				}
				p*=g[__builtin_ctz(++k)];
			}
		}
	}
}

template<class mint>
void half_fmt(vector<mint>&f){
	half_fmt(si(f),f.data());
}

#ifdef USE_GOOD_MOD

template<class mint>
vc<mint> multiply(vc<mint> x,const vc<mint>&y,bool same=false){
	int n=si(x)+si(y)-1;
	int s=1;
	while(s<n)s*=2;
	x.resize(s);inplace_fmt(x,false);
	if(!same){
		static vc<mint> z;
		z.clear();z.resize(s);
		rep(i,si(y))z[i]=y[i];
		inplace_fmt(z,false);
		rep(i,s)x[i]*=z[i];
	}else{
		rep(i,s)x[i]*=x[i];
	}
	inplace_fmt(x,true);x.resize(n);
	return x;
}
template<class mint>
vc<mint> multiply_givenlength(vc<mint> x,const vc<mint>&y,bool same=false){
	int s=si(x);
	assert(ispow2(s));
	assert(si(y));
	x.resize(s);inplace_fmt(x,false);
	if(!same){
		static vc<mint> z;
		z.clear();z.resize(s);
		rep(i,si(y))z[i]=y[i];
		inplace_fmt(z,false);
		rep(i,s)x[i]*=z[i];
	}else{
		rep(i,s)x[i]*=x[i];
	}
	inplace_fmt(x,true);
	return x;
}

#else

//59501818244292734739283969-1=5.95*10^25 までの値を正しく計算
//最終的な列の大きさが 2^24 までなら動く
//最終的な列の大きさが 2^20 以下のときは，下の 3 つの素数を使ったほうが速い（は？）
//VERIFY: yosupo
//Yukicoder No980 (same=true)
namespace arbitrary_convolution{
	constexpr modinfo base0{167772161,3};//2^25 * 5 + 1
	constexpr modinfo base1{469762049,3};//2^26 * 7 + 1
	constexpr modinfo base2{754974721,11};//2^24 * 45 + 1
	//extern constexpr modinfo base0{1045430273,3};//2^20 * 997 + 1
	//extern constexpr modinfo base1{1051721729,6};//2^20 * 1003 + 1
	//extern constexpr modinfo base2{1053818881,7};//2^20 * 1005 + 1
	using mint0=modular<base0>;
	using mint1=modular<base1>;
	using mint2=modular<base2>;
	template<class t,class mint>
	vc<t> sub(const vc<mint>&x,const vc<mint>&y,bool same=false){
		int n=si(x)+si(y)-1;
		int s=1;
		while(s<n)s*=2;
		vc<t> z(s);rep(i,si(x))z[i]=x[i].v;
		inplace_fmt(z,false);
		if(!same){
			vc<t> w(s);rep(i,si(y))w[i]=y[i].v;
			inplace_fmt(w,false);
			rep(i,s)z[i]*=w[i];
		}else{
			rep(i,s)z[i]*=z[i];
		}
		inplace_fmt(z,true);z.resize(n);
		return z;
	}
	template<class mint>
	vc<mint> multiply(const vc<mint>&x,const vc<mint>&y,bool same=false){
		auto d0=sub<mint0>(x,y,same);
		auto d1=sub<mint1>(x,y,same);
		auto d2=sub<mint2>(x,y,same);
		int n=si(d0);
		vc<mint> res(n);
		static const mint1 r01=mint1(mint0::mod).inv();
		static const mint2 r02=mint2(mint0::mod).inv();
		static const mint2 r12=mint2(mint1::mod).inv();
		static const mint2 r02r12=r02*r12;
		static const mint w1=mint(mint0::mod);
		static const mint w2=w1*mint(mint1::mod);
		rep(i,n){
			ull a=d0[i].v;
			ull b=(d1[i].v+mint1::mod-a)*r01.v%mint1::mod;
			ull c=((d2[i].v+mint2::mod-a)*r02r12.v+(mint2::mod-b)*r12.v)%mint2::mod;
			res[i].v=(a+b*w1.v+c*w2.v)%mint::mod;
		}
		return res;
	}
	template<class t,class mint>
	vc<t>&sub_givenlength(const vc<mint>&x,const vc<mint>&y,bool same=false){
		int s=si(x);
		assert(ispow2(s));
		assert(si(y)==s);
		static vc<t> z;
		z.clear();z.resize(s);
		rep(i,si(x))z[i]=x[i].v;
		inplace_fmt(z,false);
		if(!same){
			static vc<t> w;
			w.clear();w.resize(s);
			rep(i,si(y))w[i]=y[i].v;
			inplace_fmt(w,false);
			rep(i,s)z[i]*=w[i];
		}else{
			rep(i,s)z[i]*=z[i];
		}
		inplace_fmt(z,true);
		return z;
	}
	template<class mint>
	vc<mint> multiply_givenlength(vc<mint> x,const vc<mint>&y,bool same=false){
		auto&d0=sub_givenlength<mint0>(x,y,same);
		auto&d1=sub_givenlength<mint1>(x,y,same);
		auto&d2=sub_givenlength<mint2>(x,y,same);
		int n=si(d0);
		x.resize(n);
		static const mint1 r01=mint1(mint0::mod).inv();
		static const mint2 r02=mint2(mint0::mod).inv();
		static const mint2 r12=mint2(mint1::mod).inv();
		static const mint2 r02r12=r02*r12;
		static const mint w1=mint(mint0::mod);
		static const mint w2=w1*mint(mint1::mod);
		rep(i,n){
			ull a=d0[i].v;
			ull b=(d1[i].v+mint1::mod-a)*r01.v%mint1::mod;
			ull c=((d2[i].v+mint2::mod-a)*r02r12.v+(mint2::mod-b)*r12.v)%mint2::mod;
			x[i].v=(a+b*w1.v+c*w2.v)%mint::mod;
		}
		return x;
	}
}
using arbitrary_convolution::multiply;
using arbitrary_convolution::multiply_givenlength;

#endif

//UTPC2021 C
namespace integer_convolution{
	extern constexpr modinfo base0{1045430273,3};//2^20 * 997 + 1
	extern constexpr modinfo base1{1051721729,6};//2^20 * 1003 + 1
	//extern constexpr modinfo base0{469762049,3};//2^26 * 7 + 1
	//extern constexpr modinfo base1{754974721,11};//2^24 * 45 + 1
	using mint0=modular<base0>;
	using mint1=modular<base1>;
	template<class t>
	vc<t> sub(const vi&x,const vi&y,bool same=false){
		int n=si(x)+si(y)-1;
		int s=1;
		while(s<n)s*=2;
		vc<t> z(s);rep(i,si(x))z[i]=x[i];
		inplace_fmt(z,false);
		if(!same){
			vc<t> w(s);rep(i,si(y))w[i]=y[i];
			inplace_fmt(w,false);
			rep(i,s)z[i]*=w[i];
		}else{
			rep(i,s)z[i]*=z[i];
		}
		inplace_fmt(z,true);z.resize(n);
		return z;
	}
	vi multiply(const vi&x,const vi&y,bool same=false){
		auto d0=sub<mint0>(x,y,same);
		auto d1=sub<mint1>(x,y,same);
		const mint1 r=mint1(mint0::mod).inv();
		const ll v=ll(mint0::mod)*mint1::mod;
		int n=si(d0);
		vi res(n);
		rep(i,n){
			res[i]=d0[i].v+(r*(d1[i]-d0[i].v)).v*(ull)mint0::mod;
			if(res[i]>v/2)res[i]-=v;
		}
		return res;
	}
}

//最大で 1<<mx のサイズの fft が登場！
template<class mint>
vc<mint> large_convolution(const vc<mint>&a,const vc<mint>&b,int mx){
	int n=si(a),m=si(b);
	vc<mint> c(n+m-1);
	int len=1<<(mx-1);
	for(int i=0;i<n;i+=len){
		for(int j=0;j<n;j+=len){
			int x=min(len,n-i),y=min(len,m-j);
			auto d=multiply(vc<mint>(a.bg+i,a.bg+i+x),vc<mint>(b.bg+j,b.bg+j+y));
			rep(k,si(d))
				c[i+j+k]+=d[k];
		}
	}
	return c;
}

//input A: N 次,B ?,M
//output D: M 次多項式
//C を M 次多項式として
//[x^N] A*B*C = [x^M] D*C
//となるような D を返す
//CF796F
template<class mint>
vc<mint> transpose_advance(const vc<mint>&a,const vc<mint>&b,int m){
	int n=si(a)-1;
	auto d=multiply(a,b);
	vc<mint> res(m+1);
	if(n>=m){
		rep(i,m+1)res[i]=d[i+n-m];
	}else{
		rng(i,m-n,m+1)res[i]=d[i+n-m];
	}
	return res;
}

//Yukicoder 2166
template<class mint>
void chmult(vc<mint>&x,const vc<mint>&y,int s){
	x=multiply(move(x),y);
	x.resize(s);
}

#ifndef DYNAMIC_MOD
extern constexpr modinfo base{998244353,3};
//extern constexpr modinfo base{1000000007,0};
//extern constexpr modinfo base{2147483579,0};//2^31 未満の最大の安全素数
//modinfo base{1,0};
#ifdef USE_GOOD_MOD
static_assert(base.mod==998244353);
#endif
#else
modinfo base(1,0);
extern constexpr modinfo base107(1000000007,0);
using mint107=modular<base107>;
#endif
using mint=modular<base>;

mint parity(int i){
	return i%2==0?1:-1;
}

#ifdef LOCAL
const int vmax=10010;
#else
const int vmax=(1<<21)+10;
#endif
mint fact[vmax],finv[vmax],invs[vmax];
void initfact(){
	fact[0]=1;
	rng(i,1,vmax){
		fact[i]=fact[i-1]*i;
	}
	finv[vmax-1]=fact[vmax-1].inv();
	for(int i=vmax-2;i>=0;i--){
		finv[i]=finv[i+1]*(i+1);
	}
	for(int i=vmax-1;i>=1;i--){
		invs[i]=finv[i]*fact[i-1];
	}
}
mint choose(int n,int k){
	return inc(0,k,n)?fact[n]*finv[n-k]*finv[k]:0;
}
mint binom(int a,int b){
	return 0<=a&&0<=b?fact[a+b]*finv[a]*finv[b]:0;
}
mint catalan(int n){
	return binom(n,n)-(n-1>=0?binom(n-1,n+1):0);
}
//対角線を超えず (x,y) に至る方法の数
mint catalan(int x,int y){
	assert(y<=x);
	return binom(x,y)-binom(x+1,y-1);
}
//y=x+c を超えず (x,y) に至る方法の数
mint catalan(int x,int y,int c){
	assert(y<=x+c);
	return binom(x,y)-binom(x+c+1,y-c-1);
}

/*
const int vmax=610;
mint fact[vmax+1],binbuf[vmax+1][vmax+1];
mint choose(int n,int k){
	return 0<=k&&k<=n?binbuf[n-k][k]:0;
}
mint binom(int a,int b){
	return 0<=a&&0<=b?binbuf[a][b]:0;
}
void initfact(int n){
	fact[0]=1;
	rep(i,n)fact[i+1]=fact[i]*(i+1);
	rep(i,n+1)rep(j,n+1){
		if(i==0&&j==0){
			binbuf[i][j]=1;
		}else{
			binbuf[i][j]=0;
			if(i)binbuf[i][j]+=binbuf[i-1][j];
			if(j)binbuf[i][j]+=binbuf[i][j-1];
		}
	}
}
*/

mint p2[vmax],p2inv[vmax];
void initp2(){
	p2[0]=1;
	rep(i,vmax-1)p2[i+1]=p2[i]*2;
	p2inv[vmax-1]=p2[vmax-1].inv();
	per(i,vmax-1)p2inv[i]=p2inv[i+1]*2;
}

template<class F>
void array_rec(int n,int m,F f){
	vi a(n);
	function<void(int)> rec=[&](int i)->void{
		if(i==n)f(a);
		else{
			rep(nx,m){
				a[i]=nx;
				rec(i+1);
			}
		}
	};
	rec(0);
}

//FFT の結果をキャッシュして持っておく
//rw に生データ，buf に fft 後のデータ
//どちらかが空でもよい
//si(buf) は 0 or si(rw) 以上の 2 冪
//Codechef 2021 January Lunchtime EXPGROUP
//yosupo product of polynomial sequence
//Yukicoder 2166
//メモリ食いまくり
//ABC278H
//Osijek Day6 A
struct F{
	int n;
	vc<mint> rw,buf;
	F():n(0){}
	F(const vc<mint>&given):rw(given){
		n=si(rw);
		assert(n>0);
	}
	F(initializer_list<mint> init):rw(all(init)){
		n=si(rw);
		assert(n>0);
	}
	int size()const{return n;}
	bool empty()const{return n==0;}
	void assume_have(){
		if(rw.empty()){
			int s=minp2(n);
			assert(si(buf)>=s);
			rw.resize(s);
			rep(i,s)rw[i]=buf[i].v;
			inplace_fmt(rw,true);
			rw.resize(n);
		}
		assert(si(rw)==n);
	}
	vc<mint> getrw(){
		if(n==0)return {};
		assume_have();
		return rw;
	}
	void prepare(int len){
		if(si(buf)<len)assume_have();
		if(buf.empty()){
			int s=minp2(n);
			buf.resize(s);
			rep(i,n)buf[i]=rw[i];
			inplace_fmt(buf,false);
		}
		while(si(buf)<len){
			int s=si(buf);
			buf.resize(s*2);
			rep(i,n)buf[s+i]=rw[i];
			half_fmt(s,buf.data()+s);
		}
	}
	void copy_from(F&a){
		n=a.n;
		rw=a.rw;
		if(si(a.buf)){
			int s=minp2(n);
			buf.resize(s);
			rep(i,s)buf[i]=a.buf[i];
		}else buf.clear();
	}
	void init_from_sum(F&a,F&b){
		if(a.empty())return copy_from(b);
		if(b.empty())return copy_from(a);
		n=max(a.n,b.n);
		if(si(a.rw)&&si(b.rw)){
			rw.resize(n);
			rep(i,n){
				rw[i]=0;
				if(i<a.n)rw[i]+=a.rw[i];
				if(i<b.n)rw[i]+=b.rw[i];
			}
		}else rw.clear();
		int s=minp2(n);
		if(si(a.buf)>=s&&si(b.buf)>=s){
			buf.resize(s);
			rep(i,s)buf[i]=mint(a.buf[i].v)+mint(b.buf[i].v);
		}else buf.clear();
		if(rw.empty()&&buf.empty()){
			/*
			a.prepare(n);
			b.prepare(n);
			buf.resize(s);
			rep(i,s)buf[i]=mint(a.buf[i].v)+mint(b.buf[i].v);
			*/
			//こっちのほうが微小にメモリが少なくなる?
			a.assume_have();
			b.assume_have();
			rw.assign(n,0);
			rep(i,a.n)rw[i]+=a.rw[i];
			rep(i,b.n)rw[i]+=b.rw[i];
		}
	}
	//Osijek Day6 A (n=0,middle=true)
	void init_from_product(F&a,F&b,bool middle=false){
		if(a.n==0||b.n==0){
			n=0;
			rw.clear();
			buf.clear();
			return;
		}
		assert(a.n>0);
		assert(b.n>0);
		if(middle){
			assert(a.n>=b.n);
			n=a.n;
		}else{
			n=a.n+b.n-1;
		}
		rw.clear();
		int s=minp2(n);
		a.prepare(n);
		b.prepare(n);
		buf.resize(s);
		rep(i,s)buf[i]=a.buf[i]*b.buf[i];
	}
	F operator*(F&b){
		F res;
		res.init_from_product(*this,b);
		return res;
	}
	F operator*(F&&b){
		F res;
		res.init_from_product(*this,b);
		return res;
	}
	F operator+(F&b){
		F res;
		res.init_from_sum(*this,b);
		return res;
	}
	F operator+(F&&b){
		F res;
		res.init_from_sum(*this,b);
		return res;
	}
	F& operator*=(F&b){
		return *this=(*this)*b;
	}
	F& operator+=(F&b){
		return *this=(*this)+b;
	}
	F& operator+=(F&&b){
		return *this=(*this)+b;
	}
	void freememory(){
		n=0;
		vc<mint>().swap(rw);
		vc<mint>().swap(buf);
	}
	//middle product の情報をつくる
	//a,b,c が x の関数だとする
	//ans+=[x^deg(a)] abc だとしよう
	//ab の係数上位 deg(c)+1 項を d とおけば
	//ans+=[x^deg(d)] dc とできるので，この d を求めよう
	//単にサイズを広げない prod をするだけ
	//deg(a)>=deg(b)+deg(c) でないと破綻する
	F middle(F&b){
		F res;
		res.init_from_product(*this,b,true);
		return res;
	}
	//middle product の情報が buf に入っているとする
	//サイズ s に縮める
	//[n-s,n) が重要で，それ以外のパートにはゴミが入っているかもしれない
	//Osijek Day6 A
	void middle_shrink(int s){
		if(empty())return;
		assert(s<=n);
		if(s<n){
			assume_have();
			rep(i,s)rw[i]=rw[n-s+i];
			rw.resize(n=s);
			buf.clear();
		}
	}
	//転置原理+分割統治と組み合わせるといい感じ
	//ありがちなパターン
	//for i=0,1,...: ans[i]=[x^(n-1)]v; v=m_i*v
	//m_i が（高々）一次なら，dfs(l,r) では v[k].middle_shrink(r-l)
	//m_i の要素の次数はバラバラでも大丈夫
	//空の prod で空が返るようにしてあるので m_i に空を入れてもいい
	//行列とかで戻していくとき，middle 側は空or全部同じ次数という形にしておく
};
bool dbg=false;

template<class t,class u>
//vc<pi> (val,freq) みたいなやつの val を unique する
void unique_freq(vc<pair<t,u>>&a){
	sort(all(a));
	int s=0;
	rep(i,si(a)){
		auto [p,c]=a[i];
		if(i==0||a[s-1].a<p){
			a[s++]=a[i];
		}else{
			a[s-1].b+=c;
		}
	}
	a.resize(s);
}
template<class t,class u,class s>
int lwb_by_first(const vc<pair<t,u>>&a,const s&v){
	auto itr=lower_bound(all(a),v,[&](const pair<t,u>&p,const s&x){return p.a<x;});
	if(itr==a.ed||itr->a!=v)return -1;
	return itr-a.bg;
}

const int N=30000+4;

struct S{
	F f;
	int off;
	void shift(int k){
		assert(k>=0);
		if(k==0||f.empty())return;
		vc<mint> rw=f.getrw();
		vc<mint> cur(si(rw)+k);
		rep(i,si(rw))cur[i+k]=rw[i];
		f=F(cur);
		off-=k;
	}
};

S multS(S&a,S&b){
	return {a.f*b.f,a.off+b.off};
}

void addS(S&a,S b){
	if(b.f.empty())return;
	if(a.f.empty()){
		swap(a,b);
		return;
	}
	if(a.off>b.off){
		swap(a,b);
	}
	b.shift(b.off-a.off);
	a.f+=b.f;
}

void slv(){
	using A=array<int,3>;
	auto mg=[&](A a,A b,A&c){
		int amx=-inf,bmin=inf;
		rep(k,3)if(a[k]>=0)chmax(amx,k);
		rep(k,3)if(b[k]>=0)chmin(bmin,k);
		if(amx>=bmin)return false;
		rep(k,3)c[k]=max(a[k],b[k]);
		return true;
	};
	vc<A> X;
	//single
	rep(k,3){
		rep(v,5){
			A a;fila(a,-1);
			a[k]=v;
			X.pb(a);
		}
	}
	//double
	for(auto [i,j]:{pi(0,1),pi(1,2)}){
		rep(x,5)rep(y,5)if(x+y<=6){
			A a;fila(a,-1);
			a[i]=x;
			a[j]=y;
			X.pb(a);
		}
	}
	mkuni(X);
	vc<tuple<int,int,int>> X2X;
	rep(i,si(X))rep(j,si(X)){
		A c{};
		if(mg(X[i],X[j],c)){
			int k=findid(X,c);
			if(k!=-1)X2X.eb(i,j,k);
		}
	}
	auto mul=[&](vc<pair<A,mint>> f,vc<pair<A,mint>> g){
		vc<pair<A,mint>> h;
		for(auto [x,a]:f){
			for(auto [y,b]:g){
				A z{};
				rep(k,3)z[k]=x[k]+y[k];
				h.eb(z,a*b);
			}
		}
		unique_freq(h);
		return h;
	};
	vc<pair<A,mint>> Z;
	Z.eb(A{},1);
	rep(_,2)Z=mul(Z,{{A{1,0,0},-1},{A{0,1,0},1}});
	rep(_,2)Z=mul(Z,{{A{1,0,0},-1},{A{0,0,1},1}});
	rep(_,2)Z=mul(Z,{{A{0,1,0},-1},{A{0,0,1},1}});
	vc<tuple<int,int,mint>> X2Z;
	rep(i,si(X))rep(j,si(X)){
		A c{};
		if(mg(X[i],X[j],c)&&MIN(c)>=0&&SUM(c)==6){
			int k=lwb_by_first(Z,c);
			assert(k!=-1);
			X2Z.eb(i,j,Z[k].b);
		}
	}
	
	dmp(si(X));
	dmp(X);
	dmp(si(X2X));
	dmp(X2X);
	dmp(si(X2Z));
	dmp(X2Z);
	
	int power;cin>>power;
	vc<mint> lenw(N);
	rng(i,2,N)lenw[i]=mint(i-2).pow(power);
	dmp(lenw);
	
	auto getweight=[&](int k,int i,int c){
		mint w=sq(finv[i]);
		rep(_,c)w*=i;
		if(k==2)w*=lenw[i];
		return w;
	};
	
	vc<mint> ans(N);
	auto dfs=[&](auto self,int l,int r)->vc<S>{
		if(r-l==1){
			vc<S> res(si(X));
			rep(i,si(X)){
				int j=-1;
				rep(k,3)if(X[i][k]>=0){
					if(j==-1)j=k;
					else j=-2;
				}
				if(j>=0){
					vc<mint> rw{getweight(j,l,X[i][j])};
					int off=l;
					res[i]=S{rw,off};
				}
			}
			return res;
		}else{
			int mid=(l+r)/2;
			auto a=self(self,l,mid);
			auto b=self(self,mid,r);
			auto res=a;
			rep(i,si(X))addS(res[i],b[i]);
			for(auto [i,j,k]:X2X){
				addS(res[k],multS(a[i],b[j]));
			}
			for(auto [i,j,w]:X2Z){
				auto q=multS(a[i],b[j]);
				auto rw=q.f.getrw();
				rep(k,si(rw)){
					if(q.off+k<N)ans[q.off+k]+=rw[k]*w;
				}
			}
			return res;
		}
	};
	dfs(dfs,1,N);
	
	dmp(ans);
	ans.erase(ans.bg,ans.bg+3);
	rep(i,si(ans))ans[i]*=sq(fact[i]);
	ans.erase(ans.bg);
	dmp(ans);
	
	print(si(ans));
	print(ans);
}

signed main(){
	cin.tie(0);
	ios::sync_with_stdio(0);
	cout<<fixed<<setprecision(20);
	
	initfact();
	if(dbg){
		while(1)slv();
	}else{
		//int t;cin>>t;rep(_,t)
		slv();
	}
}
yukicoder

結果

ソースコード
