/** * date : 2023-05-28 17:43:54 * author : Nyaan */ #define NDEBUG using namespace std; // intrinstic #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include // utility namespace Nyaan { using ll = long long; using i64 = long long; using u64 = unsigned long long; using i128 = __int128_t; using u128 = __uint128_t; template using V = vector; template using VV = vector>; using vi = vector; using vl = vector; using vd = V; using vs = V; using vvi = vector>; using vvl = vector>; template struct P : pair { template P(Args... args) : pair(args...) {} using pair::first; using pair::second; P &operator+=(const P &r) { first += r.first; second += r.second; return *this; } P &operator-=(const P &r) { first -= r.first; second -= r.second; return *this; } P &operator*=(const P &r) { first *= r.first; second *= r.second; return *this; } template P &operator*=(const S &r) { first *= r, second *= r; return *this; } P operator+(const P &r) const { return P(*this) += r; } P operator-(const P &r) const { return P(*this) -= r; } P operator*(const P &r) const { return P(*this) *= r; } template P operator*(const S &r) const { return P(*this) *= r; } P operator-() const { return P{-first, -second}; } }; using pl = P; using pi = P; using vp = V; constexpr int inf = 1001001001; constexpr long long infLL = 4004004004004004004LL; template int sz(const T &t) { return t.size(); } template inline bool amin(T &x, U y) { return (y < x) ? (x = y, true) : false; } template inline bool amax(T &x, U y) { return (x < y) ? (x = y, true) : false; } template inline T Max(const vector &v) { return *max_element(begin(v), end(v)); } template inline T Min(const vector &v) { return *min_element(begin(v), end(v)); } template inline long long Sum(const vector &v) { return accumulate(begin(v), end(v), 0LL); } template int lb(const vector &v, const T &a) { return lower_bound(begin(v), end(v), a) - begin(v); } template int ub(const vector &v, const T &a) { return upper_bound(begin(v), end(v), a) - begin(v); } constexpr long long TEN(int n) { long long ret = 1, x = 10; for (; n; x *= x, n >>= 1) ret *= (n & 1 ? x : 1); return ret; } template pair mkp(const T &t, const U &u) { return make_pair(t, u); } template vector mkrui(const vector &v, bool rev = false) { vector ret(v.size() + 1); if (rev) { for (int i = int(v.size()) - 1; i >= 0; i--) ret[i] = v[i] + ret[i + 1]; } else { for (int i = 0; i < int(v.size()); i++) ret[i + 1] = ret[i] + v[i]; } return ret; }; template vector mkuni(const vector &v) { vector ret(v); sort(ret.begin(), ret.end()); ret.erase(unique(ret.begin(), ret.end()), ret.end()); return ret; } template vector mkord(int N,F f) { vector ord(N); iota(begin(ord), end(ord), 0); sort(begin(ord), end(ord), f); return ord; } template vector mkinv(vector &v) { int max_val = *max_element(begin(v), end(v)); vector inv(max_val + 1, -1); for (int i = 0; i < (int)v.size(); i++) inv[v[i]] = i; return inv; } vector mkiota(int n) { vector ret(n); iota(begin(ret), end(ret), 0); return ret; } template T mkrev(const T &v) { T w{v}; reverse(begin(w), end(w)); return w; } template bool nxp(vector &v) { return next_permutation(begin(v), end(v)); } template using minpq = priority_queue, greater>; } // namespace Nyaan // bit operation namespace Nyaan { __attribute__((target("popcnt"))) inline int popcnt(const u64 &a) { return _mm_popcnt_u64(a); } inline int lsb(const u64 &a) { return a ? __builtin_ctzll(a) : 64; } inline int ctz(const u64 &a) { return a ? __builtin_ctzll(a) : 64; } inline int msb(const u64 &a) { return a ? 63 - __builtin_clzll(a) : -1; } template inline int gbit(const T &a, int i) { return (a >> i) & 1; } template inline void sbit(T &a, int i, bool b) { if (gbit(a, i) != b) a ^= T(1) << i; } constexpr long long PW(int n) { return 1LL << n; } constexpr long long MSK(int n) { return (1LL << n) - 1; } } // namespace Nyaan // inout namespace Nyaan { template ostream &operator<<(ostream &os, const pair &p) { os << p.first << " " << p.second; return os; } template istream &operator>>(istream &is, pair &p) { is >> p.first >> p.second; return is; } template ostream &operator<<(ostream &os, const vector &v) { int s = (int)v.size(); for (int i = 0; i < s; i++) os << (i ? " " : "") << v[i]; return os; } template istream &operator>>(istream &is, vector &v) { for (auto &x : v) is >> x; return is; } istream &operator>>(istream &is, __int128_t &x) { string S; is >> S; x = 0; int flag = 0; for (auto &c : S) { if (c == '-') { flag = true; continue; } x *= 10; x += c - '0'; } if (flag) x = -x; return is; } istream &operator>>(istream &is, __uint128_t &x) { string S; is >> S; x = 0; for (auto &c : S) { x *= 10; x += c - '0'; } return is; } ostream &operator<<(ostream &os, __int128_t x) { if (x == 0) return os << 0; if (x < 0) os << '-', x = -x; string S; while (x) S.push_back('0' + x % 10), x /= 10; reverse(begin(S), end(S)); return os << S; } ostream &operator<<(ostream &os, __uint128_t x) { if (x == 0) return os << 0; string S; while (x) S.push_back('0' + x % 10), x /= 10; reverse(begin(S), end(S)); return os << S; } void in() {} template void in(T &t, U &...u) { cin >> t; in(u...); } void out() { cout << "\n"; } template void out(const T &t, const U &...u) { cout << t; if (sizeof...(u)) cout << sep; out(u...); } struct IoSetupNya { IoSetupNya() { cin.tie(nullptr); ios::sync_with_stdio(false); cout << fixed << setprecision(15); cerr << fixed << setprecision(7); } } iosetupnya; } // namespace Nyaan // debug #ifdef NyaanDebug #define trc(...) (void(0)) #else #define trc(...) (void(0)) #endif #ifdef NyaanLocal #define trc2(...) (void(0)) #else #define trc2(...) (void(0)) #endif // macro #define each(x, v) for (auto&& x : v) #define each2(x, y, v) for (auto&& [x, y] : v) #define all(v) (v).begin(), (v).end() #define rep(i, N) for (long long i = 0; i < (long long)(N); i++) #define repr(i, N) for (long long i = (long long)(N)-1; i >= 0; i--) #define rep1(i, N) for (long long i = 1; i <= (long long)(N); i++) #define repr1(i, N) for (long long i = (N); (long long)(i) > 0; i--) #define reg(i, a, b) for (long long i = (a); i < (b); i++) #define regr(i, a, b) for (long long i = (b)-1; i >= (a); i--) #define fi first #define se second #define ini(...) \ int __VA_ARGS__; \ in(__VA_ARGS__) #define inl(...) \ long long __VA_ARGS__; \ in(__VA_ARGS__) #define ins(...) \ string __VA_ARGS__; \ in(__VA_ARGS__) #define in2(s, t) \ for (int i = 0; i < (int)s.size(); i++) { \ in(s[i], t[i]); \ } #define in3(s, t, u) \ for (int i = 0; i < (int)s.size(); i++) { \ in(s[i], t[i], u[i]); \ } #define in4(s, t, u, v) \ for (int i = 0; i < (int)s.size(); i++) { \ in(s[i], t[i], u[i], v[i]); \ } #define die(...) \ do { \ Nyaan::out(__VA_ARGS__); \ return; \ } while (0) namespace Nyaan { void solve(); } int main() { Nyaan::solve(); } // using namespace std; using namespace std; // コンストラクタの MAX に 「C(n, r) や fac(n) でクエリを投げる最大の n 」 // を入れると倍速くらいになる // mod を超えて前計算して 0 割りを踏むバグは対策済み template struct Binomial { vector f, g, h; Binomial(int MAX = 0) { assert(T::get_mod() != 0 && "Binomial()"); f.resize(1, T{1}); g.resize(1, T{1}); h.resize(1, T{1}); if (MAX > 0) extend(MAX + 1); } void extend(int m = -1) { int n = f.size(); if (m == -1) m = n * 2; m = min(m, T::get_mod()); if (n >= m) return; f.resize(m); g.resize(m); h.resize(m); for (int i = n; i < m; i++) f[i] = f[i - 1] * T(i); g[m - 1] = f[m - 1].inverse(); h[m - 1] = g[m - 1] * f[m - 2]; for (int i = m - 2; i >= n; i--) { g[i] = g[i + 1] * T(i + 1); h[i] = g[i] * f[i - 1]; } } T fac(int i) { if (i < 0) return T(0); while (i >= (int)f.size()) extend(); return f[i]; } T finv(int i) { if (i < 0) return T(0); while (i >= (int)g.size()) extend(); return g[i]; } T inv(int i) { if (i < 0) return -inv(-i); while (i >= (int)h.size()) extend(); return h[i]; } T C(int n, int r) { if (n < 0 || n < r || r < 0) return T(0); return fac(n) * finv(n - r) * finv(r); } inline T operator()(int n, int r) { return C(n, r); } template T multinomial(const vector& r) { static_assert(is_integral::value == true); int n = 0; for (auto& x : r) { if (x < 0) return T(0); n += x; } T res = fac(n); for (auto& x : r) res *= finv(x); return res; } template T operator()(const vector& r) { return multinomial(r); } T C_naive(int n, int r) { if (n < 0 || n < r || r < 0) return T(0); T ret = T(1); r = min(r, n - r); for (int i = 1; i <= r; ++i) ret *= inv(i) * (n--); return ret; } T P(int n, int r) { if (n < 0 || n < r || r < 0) return T(0); return fac(n) * finv(n - r); } // [x^r] 1 / (1-x)^n T H(int n, int r) { if (n < 0 || r < 0) return T(0); return r == 0 ? 1 : C(n + r - 1, r); } }; // 長さが等しい列同士の畳み込みしかしない template fps multivariate_multiplication(const fps& f, const fps& g, const vector& base) { assert(f.size() == g.size()); int n = f.size(), s = base.size(), W = 1; if (s == 0) return fps{f[0] * g[0]}; while (W < 2 * n) W *= 2; vector chi(n); for (int i = 0; i < n; i++) { int x = i; for (int j = 0; j < s - 1; j++) chi[i] += (x /= base[j]); chi[i] %= s; } vector F(s, fps(W)), G(s, fps(W)); for (int i = 0; i < n; i++) F[chi[i]][i] = f[i], G[chi[i]][i] = g[i]; for (auto& x : F) x.ntt(); for (auto& x : G) x.ntt(); fps a(s); for (int k = 0; k < W; k++) { fill(begin(a), end(a), typename fps::value_type()); for (int i = 0; i < s; i++) for (int j = 0; j < s; j++) { a[i + j - (i + j >= s ? s : 0)] += F[i][k] * G[j][k]; } for (int i = 0; i < s; i++) F[i][k] = a[i]; } for (auto& x : F) x.intt(); fps h(n); for (int i = 0; i < n; i++) h[i] = F[chi[i]][i]; return h; } /** * @brief Multivariate Multiplication * @docs docs/ntt/multivariate-multiplication.md */ template struct FormalPowerSeries : vector { using vector::vector; using FPS = FormalPowerSeries; FPS &operator+=(const FPS &r) { if (r.size() > this->size()) this->resize(r.size()); for (int i = 0; i < (int)r.size(); i++) (*this)[i] += r[i]; return *this; } FPS &operator+=(const mint &r) { if (this->empty()) this->resize(1); (*this)[0] += r; return *this; } FPS &operator-=(const FPS &r) { if (r.size() > this->size()) this->resize(r.size()); for (int i = 0; i < (int)r.size(); i++) (*this)[i] -= r[i]; return *this; } FPS &operator-=(const mint &r) { if (this->empty()) this->resize(1); (*this)[0] -= r; return *this; } FPS &operator*=(const mint &v) { for (int k = 0; k < (int)this->size(); k++) (*this)[k] *= v; return *this; } FPS &operator/=(const FPS &r) { if (this->size() < r.size()) { this->clear(); return *this; } int n = this->size() - r.size() + 1; if ((int)r.size() <= 64) { FPS f(*this), g(r); g.shrink(); mint coeff = g.back().inverse(); for (auto &x : g) x *= coeff; int deg = (int)f.size() - (int)g.size() + 1; int gs = g.size(); FPS quo(deg); for (int i = deg - 1; i >= 0; i--) { quo[i] = f[i + gs - 1]; for (int j = 0; j < gs; j++) f[i + j] -= quo[i] * g[j]; } *this = quo * coeff; this->resize(n, mint(0)); return *this; } return *this = ((*this).rev().pre(n) * r.rev().inv(n)).pre(n).rev(); } FPS &operator%=(const FPS &r) { *this -= *this / r * r; shrink(); return *this; } FPS operator+(const FPS &r) const { return FPS(*this) += r; } FPS operator+(const mint &v) const { return FPS(*this) += v; } FPS operator-(const FPS &r) const { return FPS(*this) -= r; } FPS operator-(const mint &v) const { return FPS(*this) -= v; } FPS operator*(const FPS &r) const { return FPS(*this) *= r; } FPS operator*(const mint &v) const { return FPS(*this) *= v; } FPS operator/(const FPS &r) const { return FPS(*this) /= r; } FPS operator%(const FPS &r) const { return FPS(*this) %= r; } FPS operator-() const { FPS ret(this->size()); for (int i = 0; i < (int)this->size(); i++) ret[i] = -(*this)[i]; return ret; } void shrink() { while (this->size() && this->back() == mint(0)) this->pop_back(); } FPS rev() const { FPS ret(*this); reverse(begin(ret), end(ret)); return ret; } FPS dot(FPS r) const { FPS ret(min(this->size(), r.size())); for (int i = 0; i < (int)ret.size(); i++) ret[i] = (*this)[i] * r[i]; return ret; } FPS pre(int sz) const { return FPS(begin(*this), begin(*this) + min((int)this->size(), sz)); } FPS operator>>(int sz) const { if ((int)this->size() <= sz) return {}; FPS ret(*this); ret.erase(ret.begin(), ret.begin() + sz); return ret; } FPS operator<<(int sz) const { FPS ret(*this); ret.insert(ret.begin(), sz, mint(0)); return ret; } FPS diff() const { const int n = (int)this->size(); FPS ret(max(0, n - 1)); mint one(1), coeff(1); for (int i = 1; i < n; i++) { ret[i - 1] = (*this)[i] * coeff; coeff += one; } return ret; } FPS integral() const { const int n = (int)this->size(); FPS ret(n + 1); ret[0] = mint(0); if (n > 0) ret[1] = mint(1); auto mod = mint::get_mod(); for (int i = 2; i <= n; i++) ret[i] = (-ret[mod % i]) * (mod / i); for (int i = 0; i < n; i++) ret[i + 1] *= (*this)[i]; return ret; } mint eval(mint x) const { mint r = 0, w = 1; for (auto &v : *this) r += w * v, w *= x; return r; } FPS log(int deg = -1) const { assert((*this)[0] == mint(1)); if (deg == -1) deg = (int)this->size(); return (this->diff() * this->inv(deg)).pre(deg - 1).integral(); } FPS pow(int64_t k, int deg = -1) const { const int n = (int)this->size(); if (deg == -1) deg = n; if (k == 0) { FPS ret(deg); if (deg) ret[0] = 1; return ret; } for (int i = 0; i < n; i++) { if ((*this)[i] != mint(0)) { mint rev = mint(1) / (*this)[i]; FPS ret = (((*this * rev) >> i).log(deg) * k).exp(deg); ret *= (*this)[i].pow(k); ret = (ret << (i * k)).pre(deg); if ((int)ret.size() < deg) ret.resize(deg, mint(0)); return ret; } if (__int128_t(i + 1) * k >= deg) return FPS(deg, mint(0)); } return FPS(deg, mint(0)); } static void *ntt_ptr; static void set_fft(); FPS &operator*=(const FPS &r); void ntt(); void intt(); void ntt_doubling(); static int ntt_pr(); FPS inv(int deg = -1) const; FPS exp(int deg = -1) const; }; template void *FormalPowerSeries::ntt_ptr = nullptr; /** * @brief 多項式/形式的冪級数ライブラリ * @docs docs/fps/formal-power-series.md */ // FFT mod でないと使えない // f.size() != (base の総積) でも動くが, 二項演算は長さが等しい列同士しかダメ template struct MultivariateFormalPowerSeries { using mfps = MultivariateFormalPowerSeries; using fps = FormalPowerSeries; fps f; vector base; MultivariateFormalPowerSeries() = default; MultivariateFormalPowerSeries(const vector& _base) : base(_base) { int n = 1; for (auto& x : base) n *= x; f.resize(n); } MultivariateFormalPowerSeries(const fps& _f, const vector& _base) : f(_f), base(_base) {} friend ostream& operator<<(ostream& os, const mfps& rhs) { os << "[ "; for (int i = 0; i < (int)rhs.f.size(); i++) { os << rhs.f[i] << (i + 1 == (int)rhs.f.size() ? "" : ", "); } return os << " ]"; } mfps& operator+=(const mfps& rhs) { assert(base == rhs.base && f.size() == rhs.f.size()); for (int i = 0; i < (int)f.size(); i++) f[i] += rhs.f[i]; return *this; } mfps& operator-=(const mfps& rhs) { assert(base == rhs.base && f.size() == rhs.f.size()); for (int i = 0; i < (int)f.size(); i++) f[i] -= rhs.f[i]; return *this; } mfps& operator*=(const mfps& rhs) { assert(base == rhs.base && f.size() == rhs.f.size()); f = multivariate_multiplication(f, rhs.f, base); return *this; } mfps& operator/=(const mfps& rhs) { assert(base == rhs.base); assert(f.size() == rhs.f.size()); return (*this) *= rhs.inv(); } mfps& operator+=(const mint& rhs) { assert(!f.empty()); f[0] += rhs; return *this; } mfps& operator-=(const mint& rhs) { assert(!f.empty()); f[0] -= rhs; return *this; } mfps& operator*=(const mint& rhs) { for (auto& x : f) x *= rhs; return *this; } mfps& operator/=(const mint& rhs) { for (auto& x : f) x /= rhs; return *this; } mfps operator+(const mfps& rhs) const { return mfps{*this} += rhs; } mfps operator-(const mfps& rhs) const { return mfps{*this} -= rhs; } mfps operator*(const mfps& rhs) const { return mfps{*this} *= rhs; } mfps operator/(const mfps& rhs) const { return mfps{*this} /= rhs; } mfps operator+(const mint& rhs) const { return mfps{*this} += rhs; } mfps operator-(const mint& rhs) const { return mfps{*this} -= rhs; } mfps operator*(const mint& rhs) const { return mfps{*this} *= rhs; } mfps operator/(const mint& rhs) const { return mfps{*this} /= rhs; } mfps operator+() const { return mfps{*this}; } mfps operator-() const { return mfps{-f, base}; } friend bool operator==(const mfps& lhs, const mfps& rhs) { return lhs.f == rhs.f && lhs.base == rhs.base; } friend bool operator!=(const mfps& lhs, const mfps& rhs) { return lhs.f != rhs.f || lhs.base != rhs.base; } mfps diff() const { mfps g{*this}; for (int i = 0; i < (int)g.f.size(); i++) g.f[i] *= i; return g; } mfps integral() const { static Binomial binom; mfps g{*this}; for (int i = 1; i < (int)g.f.size(); i++) g.f[i] *= binom.inv(i); return g; } mfps inv() const { assert(f[0] != 0); if (base.empty()) return mfps{fps{f[0].inverse()}, base}; int n = f.size(), s = base.size(), W = 1; while (W < 2 * n) W *= 2; vector chi(W); for (int i = 0; i < W; i++) { int x = i; for (int j = 0; j < s - 1; j++) chi[i] += (x /= base[j]); chi[i] %= s; } auto hadamard_prod = [&s](vector& F, vector& G, vector& H) { fps a(s); for (int k = 0; k < (int)F[0].size(); k++) { fill(begin(a), end(a), typename fps::value_type()); for (int i = 0; i < s; i++) for (int j = 0; j < s; j++) { a[i + j - (i + j >= s ? s : 0)] += F[i][k] * G[j][k]; } for (int i = 0; i < s; i++) H[i][k] = a[i]; } }; fps g(W); g[0] = f[0].inverse(); for (int d = 1; d < n; d *= 2) { vector F(s, fps(2 * d)), G(s, fps(2 * d)), H(s, fps(2 * d)); for (int j = 0; j < min((int)f.size(), 2 * d); j++) F[chi[j]][j] = f[j]; for (int j = 0; j < d; j++) G[chi[j]][j] = g[j]; for (auto& x : F) x.ntt(); for (auto& x : G) x.ntt(); hadamard_prod(F, G, H); for (auto& x : H) x.intt(); for (auto& x : F) fill(begin(x), end(x), typename fps::value_type()); for (int j = d; j < 2 * d; j++) F[chi[j]][j] = H[chi[j]][j]; for (auto& x : F) x.ntt(); hadamard_prod(F, G, H); for (auto& x : H) x.intt(); for (int j = d; j < 2 * d; j++) g[j] = -H[chi[j]][j]; } mfps res{*this}; res.f = fps{begin(g), begin(g) + n}; return res; } mfps log() const { assert(!f.empty() && f[0] == 1); return ((*this).diff() / *this).integral(); } mfps exp() const { assert(!f.empty() && f[0] == mint{0}); int n = f.size(); mfps g{fps{1}, base}; for (int d = 1; d < n; d *= 2) { int s = min(n, d * 2); g.f.resize(s, mint{0}); g *= mfps{fps{begin(f), begin(f) + s}, base} - g.log() + 1; } return g; } mfps pow(long long e) const { assert(!f.empty()); if (f[0] != 0) { mint f0inv = f[0].inverse(), coe = f[0].pow(e); return (((*this) * f0inv).log() * e).exp() * coe; } int n = f.size(); long long base_sum = 0; for (auto& b : base) base_sum += b - 1; if (e > base_sum) return mfps{fps(n), base}; mfps res{fps(n), base}, a{*this}; res.f[0] = 1; for (; e; a *= a, e >>= 1) { if (e & 1) res *= a; } return res; } }; /** * @brief 多変数形式的冪級数ライブラリ */ // // f(vector& a, bool rev) : 1 次元 DFT (rev は逆変換かどうか) template struct MultidimensionalFourierTransform { vector base; function&, bool)> dft1d; MultidimensionalFourierTransform(const vector& bs, const function&, bool)>& f) : base(bs), dft1d(f) {} bool ascend(vector& v) { int i = 0; v[i] += 1; while (v[i] == base[i]) { if (i == (int)v.size() - 1) return false; v[i] = 0; v[++i] += 1; } return true; } int operator()(vector& a) { int res = a[0], coeff = 1; for (int i = 1; i < (int)a.size(); i++) coeff *= base[i - 1], res += coeff * a[i]; return res; } void inner(vector& a, int dim, bool rev = false) { int i = 0, shift = 1, n = base[dim]; vector f(n); vector id(base.size()); for (int j = 0; j < dim; j++) shift *= base[j]; do { if (id[dim] != 0) continue; for (int j = 0, t = i; j < n; j++, t += shift) f[j] = a[t]; dft1d(f, rev); for (int j = 0, t = i; j < n; j++, t += shift) a[t] = f[j]; id[dim] = 0; } while (++i && ascend(id)); } void fft(vector& a, bool rev = false) { if (!rev) for (int i = 0; i < (int)base.size(); i++) inner(a, i); else for (int i = (int)base.size(); i--;) inner(a, i, true); } }; /** * @brief 多次元FFT */ // constexpr uint32_t PrimitiveRoot(uint32_t mod) { using u64 = uint64_t; if (mod == 2) return 1; u64 ds[32] = {}; int idx = 0; u64 m = mod - 1; for (u64 i = 2; i * i <= m; ++i) { if (m % i == 0) { ds[idx++] = i; while (m % i == 0) m /= i; } } if (m != 1) ds[idx++] = m; uint32_t pr = 2; while (1) { int flg = 1; for (int i = 0; i < idx; ++i) { u64 a = pr, b = (mod - 1) / ds[i], r = 1; while (b) { if (b & 1) r = r * a % mod; a = a * a % mod; b >>= 1; } if (r == 1) { flg = 0; break; } } if (flg == 1) break; ++pr; } return pr; } /** * @brief 原始根 */ // template struct NTT { static constexpr uint32_t get_pr() { uint32_t _mod = mint::get_mod(); using u64 = uint64_t; u64 ds[32] = {}; int idx = 0; u64 m = _mod - 1; for (u64 i = 2; i * i <= m; ++i) { if (m % i == 0) { ds[idx++] = i; while (m % i == 0) m /= i; } } if (m != 1) ds[idx++] = m; uint32_t _pr = 2; while (1) { int flg = 1; for (int i = 0; i < idx; ++i) { u64 a = _pr, b = (_mod - 1) / ds[i], r = 1; while (b) { if (b & 1) r = r * a % _mod; a = a * a % _mod; b >>= 1; } if (r == 1) { flg = 0; break; } } if (flg == 1) break; ++_pr; } return _pr; }; static constexpr uint32_t mod = mint::get_mod(); static constexpr uint32_t pr = get_pr(); static constexpr int level = __builtin_ctzll(mod - 1); mint dw[level], dy[level]; void setwy(int k) { mint w[level], y[level]; w[k - 1] = mint(pr).pow((mod - 1) / (1 << k)); y[k - 1] = w[k - 1].inverse(); for (int i = k - 2; i > 0; --i) w[i] = w[i + 1] * w[i + 1], y[i] = y[i + 1] * y[i + 1]; dw[1] = w[1], dy[1] = y[1], dw[2] = w[2], dy[2] = y[2]; for (int i = 3; i < k; ++i) { dw[i] = dw[i - 1] * y[i - 2] * w[i]; dy[i] = dy[i - 1] * w[i - 2] * y[i]; } } NTT() { setwy(level); } void fft4(vector &a, int k) { if ((int)a.size() <= 1) return; if (k == 1) { mint a1 = a[1]; a[1] = a[0] - a[1]; a[0] = a[0] + a1; return; } if (k & 1) { int v = 1 << (k - 1); for (int j = 0; j < v; ++j) { mint ajv = a[j + v]; a[j + v] = a[j] - ajv; a[j] += ajv; } } int u = 1 << (2 + (k & 1)); int v = 1 << (k - 2 - (k & 1)); mint one = mint(1); mint imag = dw[1]; while (v) { // jh = 0 { int j0 = 0; int j1 = v; int j2 = j1 + v; int j3 = j2 + v; for (; j0 < v; ++j0, ++j1, ++j2, ++j3) { mint t0 = a[j0], t1 = a[j1], t2 = a[j2], t3 = a[j3]; mint t0p2 = t0 + t2, t1p3 = t1 + t3; mint t0m2 = t0 - t2, t1m3 = (t1 - t3) * imag; a[j0] = t0p2 + t1p3, a[j1] = t0p2 - t1p3; a[j2] = t0m2 + t1m3, a[j3] = t0m2 - t1m3; } } // jh >= 1 mint ww = one, xx = one * dw[2], wx = one; for (int jh = 4; jh < u;) { ww = xx * xx, wx = ww * xx; int j0 = jh * v; int je = j0 + v; int j2 = je + v; for (; j0 < je; ++j0, ++j2) { mint t0 = a[j0], t1 = a[j0 + v] * xx, t2 = a[j2] * ww, t3 = a[j2 + v] * wx; mint t0p2 = t0 + t2, t1p3 = t1 + t3; mint t0m2 = t0 - t2, t1m3 = (t1 - t3) * imag; a[j0] = t0p2 + t1p3, a[j0 + v] = t0p2 - t1p3; a[j2] = t0m2 + t1m3, a[j2 + v] = t0m2 - t1m3; } xx *= dw[__builtin_ctzll((jh += 4))]; } u <<= 2; v >>= 2; } } void ifft4(vector &a, int k) { if ((int)a.size() <= 1) return; if (k == 1) { mint a1 = a[1]; a[1] = a[0] - a[1]; a[0] = a[0] + a1; return; } int u = 1 << (k - 2); int v = 1; mint one = mint(1); mint imag = dy[1]; while (u) { // jh = 0 { int j0 = 0; int j1 = v; int j2 = v + v; int j3 = j2 + v; for (; j0 < v; ++j0, ++j1, ++j2, ++j3) { mint t0 = a[j0], t1 = a[j1], t2 = a[j2], t3 = a[j3]; mint t0p1 = t0 + t1, t2p3 = t2 + t3; mint t0m1 = t0 - t1, t2m3 = (t2 - t3) * imag; a[j0] = t0p1 + t2p3, a[j2] = t0p1 - t2p3; a[j1] = t0m1 + t2m3, a[j3] = t0m1 - t2m3; } } // jh >= 1 mint ww = one, xx = one * dy[2], yy = one; u <<= 2; for (int jh = 4; jh < u;) { ww = xx * xx, yy = xx * imag; int j0 = jh * v; int je = j0 + v; int j2 = je + v; for (; j0 < je; ++j0, ++j2) { mint t0 = a[j0], t1 = a[j0 + v], t2 = a[j2], t3 = a[j2 + v]; mint t0p1 = t0 + t1, t2p3 = t2 + t3; mint t0m1 = (t0 - t1) * xx, t2m3 = (t2 - t3) * yy; a[j0] = t0p1 + t2p3, a[j2] = (t0p1 - t2p3) * ww; a[j0 + v] = t0m1 + t2m3, a[j2 + v] = (t0m1 - t2m3) * ww; } xx *= dy[__builtin_ctzll(jh += 4)]; } u >>= 4; v <<= 2; } if (k & 1) { u = 1 << (k - 1); for (int j = 0; j < u; ++j) { mint ajv = a[j] - a[j + u]; a[j] += a[j + u]; a[j + u] = ajv; } } } void ntt(vector &a) { if ((int)a.size() <= 1) return; fft4(a, __builtin_ctz(a.size())); } void intt(vector &a) { if ((int)a.size() <= 1) return; ifft4(a, __builtin_ctz(a.size())); mint iv = mint(a.size()).inverse(); for (auto &x : a) x *= iv; } vector multiply(const vector &a, const vector &b) { int l = a.size() + b.size() - 1; if (min(a.size(), b.size()) <= 40) { vector s(l); for (int i = 0; i < (int)a.size(); ++i) for (int j = 0; j < (int)b.size(); ++j) s[i + j] += a[i] * b[j]; return s; } int k = 2, M = 4; while (M < l) M <<= 1, ++k; setwy(k); vector s(M); for (int i = 0; i < (int)a.size(); ++i) s[i] = a[i]; fft4(s, k); if (a.size() == b.size() && a == b) { for (int i = 0; i < M; ++i) s[i] *= s[i]; } else { vector t(M); for (int i = 0; i < (int)b.size(); ++i) t[i] = b[i]; fft4(t, k); for (int i = 0; i < M; ++i) s[i] *= t[i]; } ifft4(s, k); s.resize(l); mint invm = mint(M).inverse(); for (int i = 0; i < l; ++i) s[i] *= invm; return s; } void ntt_doubling(vector &a) { int M = (int)a.size(); auto b = a; intt(b); mint r = 1, zeta = mint(pr).pow((mint::get_mod() - 1) / (M << 1)); for (int i = 0; i < M; i++) b[i] *= r, r *= zeta; ntt(b); copy(begin(b), end(b), back_inserter(a)); } }; template void FormalPowerSeries::set_fft() { if (!ntt_ptr) ntt_ptr = new NTT; } template FormalPowerSeries& FormalPowerSeries::operator*=( const FormalPowerSeries& r) { if (this->empty() || r.empty()) { this->clear(); return *this; } set_fft(); auto ret = static_cast*>(ntt_ptr)->multiply(*this, r); return *this = FormalPowerSeries(ret.begin(), ret.end()); } template void FormalPowerSeries::ntt() { set_fft(); static_cast*>(ntt_ptr)->ntt(*this); } template void FormalPowerSeries::intt() { set_fft(); static_cast*>(ntt_ptr)->intt(*this); } template void FormalPowerSeries::ntt_doubling() { set_fft(); static_cast*>(ntt_ptr)->ntt_doubling(*this); } template int FormalPowerSeries::ntt_pr() { set_fft(); return static_cast*>(ntt_ptr)->pr; } template FormalPowerSeries FormalPowerSeries::inv(int deg) const { assert((*this)[0] != mint(0)); if (deg == -1) deg = (int)this->size(); FormalPowerSeries res(deg); res[0] = {mint(1) / (*this)[0]}; for (int d = 1; d < deg; d <<= 1) { FormalPowerSeries f(2 * d), g(2 * d); for (int j = 0; j < min((int)this->size(), 2 * d); j++) f[j] = (*this)[j]; for (int j = 0; j < d; j++) g[j] = res[j]; f.ntt(); g.ntt(); for (int j = 0; j < 2 * d; j++) f[j] *= g[j]; f.intt(); for (int j = 0; j < d; j++) f[j] = 0; f.ntt(); for (int j = 0; j < 2 * d; j++) f[j] *= g[j]; f.intt(); for (int j = d; j < min(2 * d, deg); j++) res[j] = -f[j]; } return res.pre(deg); } template FormalPowerSeries FormalPowerSeries::exp(int deg) const { using fps = FormalPowerSeries; assert((*this).size() == 0 || (*this)[0] == mint(0)); if (deg == -1) deg = this->size(); fps inv; inv.reserve(deg + 1); inv.push_back(mint(0)); inv.push_back(mint(1)); auto inplace_integral = [&](fps& F) -> void { const int n = (int)F.size(); auto mod = mint::get_mod(); while ((int)inv.size() <= n) { int i = inv.size(); inv.push_back((-inv[mod % i]) * (mod / i)); } F.insert(begin(F), mint(0)); for (int i = 1; i <= n; i++) F[i] *= inv[i]; }; auto inplace_diff = [](fps& F) -> void { if (F.empty()) return; F.erase(begin(F)); mint coeff = 1, one = 1; for (int i = 0; i < (int)F.size(); i++) { F[i] *= coeff; coeff += one; } }; fps b{1, 1 < (int)this->size() ? (*this)[1] : 0}, c{1}, z1, z2{1, 1}; for (int m = 2; m < deg; m *= 2) { auto y = b; y.resize(2 * m); y.ntt(); z1 = z2; fps z(m); for (int i = 0; i < m; ++i) z[i] = y[i] * z1[i]; z.intt(); fill(begin(z), begin(z) + m / 2, mint(0)); z.ntt(); for (int i = 0; i < m; ++i) z[i] *= -z1[i]; z.intt(); c.insert(end(c), begin(z) + m / 2, end(z)); z2 = c; z2.resize(2 * m); z2.ntt(); fps x(begin(*this), begin(*this) + min(this->size(), m)); x.resize(m); inplace_diff(x); x.push_back(mint(0)); x.ntt(); for (int i = 0; i < m; ++i) x[i] *= y[i]; x.intt(); x -= b.diff(); x.resize(2 * m); for (int i = 0; i < m - 1; ++i) x[m + i] = x[i], x[i] = mint(0); x.ntt(); for (int i = 0; i < 2 * m; ++i) x[i] *= z2[i]; x.intt(); x.pop_back(); inplace_integral(x); for (int i = m; i < min(this->size(), 2 * m); ++i) x[i] += (*this)[i]; fill(begin(x), begin(x) + m, mint(0)); x.ntt(); for (int i = 0; i < 2 * m; ++i) x[i] *= y[i]; x.intt(); b.insert(end(b), begin(x) + m, end(x)); } return fps{begin(b), begin(b) + deg}; } /** * @brief NTT mod用FPSライブラリ * @docs docs/fps/ntt-friendly-fps.md */ template struct LazyMontgomeryModInt { using mint = LazyMontgomeryModInt; using i32 = int32_t; using u32 = uint32_t; using u64 = uint64_t; static constexpr u32 get_r() { u32 ret = mod; for (i32 i = 0; i < 4; ++i) ret *= 2 - mod * ret; return ret; } static constexpr u32 r = get_r(); static constexpr u32 n2 = -u64(mod) % mod; static_assert(r * mod == 1, "invalid, r * mod != 1"); static_assert(mod < (1 << 30), "invalid, mod >= 2 ^ 30"); static_assert((mod & 1) == 1, "invalid, mod % 2 == 0"); u32 a; constexpr LazyMontgomeryModInt() : a(0) {} constexpr LazyMontgomeryModInt(const int64_t &b) : a(reduce(u64(b % mod + mod) * n2)){}; static constexpr u32 reduce(const u64 &b) { return (b + u64(u32(b) * u32(-r)) * mod) >> 32; } constexpr mint &operator+=(const mint &b) { if (i32(a += b.a - 2 * mod) < 0) a += 2 * mod; return *this; } constexpr mint &operator-=(const mint &b) { if (i32(a -= b.a) < 0) a += 2 * mod; return *this; } constexpr mint &operator*=(const mint &b) { a = reduce(u64(a) * b.a); return *this; } constexpr mint &operator/=(const mint &b) { *this *= b.inverse(); return *this; } constexpr mint operator+(const mint &b) const { return mint(*this) += b; } constexpr mint operator-(const mint &b) const { return mint(*this) -= b; } constexpr mint operator*(const mint &b) const { return mint(*this) *= b; } constexpr mint operator/(const mint &b) const { return mint(*this) /= b; } constexpr bool operator==(const mint &b) const { return (a >= mod ? a - mod : a) == (b.a >= mod ? b.a - mod : b.a); } constexpr bool operator!=(const mint &b) const { return (a >= mod ? a - mod : a) != (b.a >= mod ? b.a - mod : b.a); } constexpr mint operator-() const { return mint() - mint(*this); } constexpr mint pow(u64 n) const { mint ret(1), mul(*this); while (n > 0) { if (n & 1) ret *= mul; mul *= mul; n >>= 1; } return ret; } constexpr mint inverse() const { return pow(mod - 2); } friend ostream &operator<<(ostream &os, const mint &b) { return os << b.get(); } friend istream &operator>>(istream &is, mint &b) { int64_t t; is >> t; b = LazyMontgomeryModInt(t); return (is); } constexpr u32 get() const { u32 ret = reduce(a); return ret >= mod ? ret - mod : ret; } static constexpr u32 get_mod() { return mod; } }; // using namespace Nyaan; constexpr unsigned _mod = 120586241; using mint = LazyMontgomeryModInt<_mod>; using vm = vector; using vvm = vector; Binomial C; using fps = FormalPowerSeries; using mfps = MultivariateFormalPowerSeries; using namespace Nyaan; void q() { int pr = PrimitiveRoot(_mod); vm pr10(110); pr10[0] = 1; pr10[1] = mint(pr).pow((_mod - 1) / 10); rep1(i, 102) pr10[i + 1] = pr10[i] * pr10[1]; auto dft = [&](vector& f, bool rev) { int n = sz(f); vector F(n); rep(i, n) rep(j, n) F[j] += f[i] * pr10[i * j]; if (rev) reverse(begin(F) + 1, end(F)); f.swap(F); }; long long n, k, m, t; cin >> n >> k >> m >> t; vector a(n); each(x, a) { cin >> x; assert(0 <= x and x < TEN(k)); } vector base(k, 10); vector f(TEN(k)); each(x, a) f[x] += 1; MultidimensionalFourierTransform mntt(base, dft); for (int i = k - 1; i >= t; i--) mntt.inner(f, i); for (int i = 0, di = TEN(t); i < TEN(k); i += di) { mfps g{{begin(f) + i, begin(f) + i + di}, vi(t, 10)}; trc(g, m); g = g.pow(m); trc(g); for (int j = 0; j < di; j++) f[i + j] = g.f[j]; } for (int i = t; i < k; i++) mntt.inner(f, i, true); mint coe = mint(10).inverse().pow(k - t); each(x, f) cout << x * coe << "\n"; } void Nyaan::solve() { int t = 1; // in(t); while (t--) q(); }