QOJ.ac
QOJ
ID | Problem | Submitter | Result | Time | Memory | Language | File size | Submit time | Judge time |
---|---|---|---|---|---|---|---|---|---|
#576888 | #8173. T Tile Placement Counting | maspy | AC ✓ | 2121ms | 29780kb | C++20 | 50.8kb | 2024-09-19 23:13:56 | 2024-09-19 23:14:02 |
Judging History
answer
#line 1 "/home/maspy/compro/library/my_template.hpp"
#if defined(LOCAL)
#include <my_template_compiled.hpp>
#else
// https://codeforces.com/blog/entry/96344
#pragma GCC optimize("Ofast,unroll-loops")
// いまの CF だとこれ入れると動かない?
// #pragma GCC target("avx2,popcnt")
#include <bits/stdc++.h>
using namespace std;
using ll = long long;
using u32 = unsigned int;
using u64 = unsigned long long;
using i128 = __int128;
using u128 = unsigned __int128;
using f128 = __float128;
template <class T>
constexpr T infty = 0;
template <>
constexpr int infty<int> = 1'010'000'000;
template <>
constexpr ll infty<ll> = 2'020'000'000'000'000'000;
template <>
constexpr u32 infty<u32> = infty<int>;
template <>
constexpr u64 infty<u64> = infty<ll>;
template <>
constexpr i128 infty<i128> = i128(infty<ll>) * 2'000'000'000'000'000'000;
template <>
constexpr double infty<double> = infty<ll>;
template <>
constexpr long double infty<long double> = infty<ll>;
using pi = pair<ll, ll>;
using vi = vector<ll>;
template <class T>
using vc = vector<T>;
template <class T>
using vvc = vector<vc<T>>;
template <class T>
using vvvc = vector<vvc<T>>;
template <class T>
using vvvvc = vector<vvvc<T>>;
template <class T>
using vvvvvc = vector<vvvvc<T>>;
template <class T>
using pq = priority_queue<T>;
template <class T>
using pqg = priority_queue<T, vector<T>, greater<T>>;
#define vv(type, name, h, ...) vector<vector<type>> name(h, vector<type>(__VA_ARGS__))
#define vvv(type, name, h, w, ...) vector<vector<vector<type>>> name(h, vector<vector<type>>(w, vector<type>(__VA_ARGS__)))
#define vvvv(type, name, a, b, c, ...) \
vector<vector<vector<vector<type>>>> name(a, vector<vector<vector<type>>>(b, vector<vector<type>>(c, vector<type>(__VA_ARGS__))))
// https://trap.jp/post/1224/
#define FOR1(a) for (ll _ = 0; _ < ll(a); ++_)
#define FOR2(i, a) for (ll i = 0; i < ll(a); ++i)
#define FOR3(i, a, b) for (ll i = a; i < ll(b); ++i)
#define FOR4(i, a, b, c) for (ll i = a; i < ll(b); i += (c))
#define FOR1_R(a) for (ll i = (a)-1; i >= ll(0); --i)
#define FOR2_R(i, a) for (ll i = (a)-1; i >= ll(0); --i)
#define FOR3_R(i, a, b) for (ll i = (b)-1; i >= ll(a); --i)
#define overload4(a, b, c, d, e, ...) e
#define overload3(a, b, c, d, ...) d
#define FOR(...) overload4(__VA_ARGS__, FOR4, FOR3, FOR2, FOR1)(__VA_ARGS__)
#define FOR_R(...) overload3(__VA_ARGS__, FOR3_R, FOR2_R, FOR1_R)(__VA_ARGS__)
#define FOR_subset(t, s) for (ll t = (s); t >= 0; t = (t == 0 ? -1 : (t - 1) & (s)))
#define all(x) x.begin(), x.end()
#define len(x) ll(x.size())
#define elif else if
#define eb emplace_back
#define mp make_pair
#define mt make_tuple
#define fi first
#define se second
#define stoi stoll
int popcnt(int x) { return __builtin_popcount(x); }
int popcnt(u32 x) { return __builtin_popcount(x); }
int popcnt(ll x) { return __builtin_popcountll(x); }
int popcnt(u64 x) { return __builtin_popcountll(x); }
int popcnt_mod_2(int x) { return __builtin_parity(x); }
int popcnt_mod_2(u32 x) { return __builtin_parity(x); }
int popcnt_mod_2(ll x) { return __builtin_parityll(x); }
int popcnt_mod_2(u64 x) { return __builtin_parityll(x); }
// (0, 1, 2, 3, 4) -> (-1, 0, 1, 1, 2)
int topbit(int x) { return (x == 0 ? -1 : 31 - __builtin_clz(x)); }
int topbit(u32 x) { return (x == 0 ? -1 : 31 - __builtin_clz(x)); }
int topbit(ll x) { return (x == 0 ? -1 : 63 - __builtin_clzll(x)); }
int topbit(u64 x) { return (x == 0 ? -1 : 63 - __builtin_clzll(x)); }
// (0, 1, 2, 3, 4) -> (-1, 0, 1, 0, 2)
int lowbit(int x) { return (x == 0 ? -1 : __builtin_ctz(x)); }
int lowbit(u32 x) { return (x == 0 ? -1 : __builtin_ctz(x)); }
int lowbit(ll x) { return (x == 0 ? -1 : __builtin_ctzll(x)); }
int lowbit(u64 x) { return (x == 0 ? -1 : __builtin_ctzll(x)); }
template <typename T>
T floor(T a, T b) {
return a / b - (a % b && (a ^ b) < 0);
}
template <typename T>
T ceil(T x, T y) {
return floor(x + y - 1, y);
}
template <typename T>
T bmod(T x, T y) {
return x - y * floor(x, y);
}
template <typename T>
pair<T, T> divmod(T x, T y) {
T q = floor(x, y);
return {q, x - q * y};
}
template <typename T, typename U>
T SUM(const vector<U> &A) {
T sm = 0;
for (auto &&a: A) sm += a;
return sm;
}
#define MIN(v) *min_element(all(v))
#define MAX(v) *max_element(all(v))
#define LB(c, x) distance((c).begin(), lower_bound(all(c), (x)))
#define UB(c, x) distance((c).begin(), upper_bound(all(c), (x)))
#define UNIQUE(x) sort(all(x)), x.erase(unique(all(x)), x.end()), x.shrink_to_fit()
template <typename T>
T POP(deque<T> &que) {
T a = que.front();
que.pop_front();
return a;
}
template <typename T>
T POP(pq<T> &que) {
T a = que.top();
que.pop();
return a;
}
template <typename T>
T POP(pqg<T> &que) {
T a = que.top();
que.pop();
return a;
}
template <typename T>
T POP(vc<T> &que) {
T a = que.back();
que.pop_back();
return a;
}
template <typename F>
ll binary_search(F check, ll ok, ll ng, bool check_ok = true) {
if (check_ok) assert(check(ok));
while (abs(ok - ng) > 1) {
auto x = (ng + ok) / 2;
(check(x) ? ok : ng) = x;
}
return ok;
}
template <typename F>
double binary_search_real(F check, double ok, double ng, int iter = 100) {
FOR(iter) {
double x = (ok + ng) / 2;
(check(x) ? ok : ng) = x;
}
return (ok + ng) / 2;
}
template <class T, class S>
inline bool chmax(T &a, const S &b) {
return (a < b ? a = b, 1 : 0);
}
template <class T, class S>
inline bool chmin(T &a, const S &b) {
return (a > b ? a = b, 1 : 0);
}
// ? は -1
vc<int> s_to_vi(const string &S, char first_char) {
vc<int> A(S.size());
FOR(i, S.size()) { A[i] = (S[i] != '?' ? S[i] - first_char : -1); }
return A;
}
template <typename T, typename U>
vector<T> cumsum(vector<U> &A, int off = 1) {
int N = A.size();
vector<T> B(N + 1);
FOR(i, N) { B[i + 1] = B[i] + A[i]; }
if (off == 0) B.erase(B.begin());
return B;
}
// stable sort
template <typename T>
vector<int> argsort(const vector<T> &A) {
vector<int> ids(len(A));
iota(all(ids), 0);
sort(all(ids), [&](int i, int j) { return (A[i] == A[j] ? i < j : A[i] < A[j]); });
return ids;
}
// A[I[0]], A[I[1]], ...
template <typename T>
vc<T> rearrange(const vc<T> &A, const vc<int> &I) {
vc<T> B(len(I));
FOR(i, len(I)) B[i] = A[I[i]];
return B;
}
template <typename T, typename... Vectors>
void concat(vc<T> &first, const Vectors &... others) {
vc<T> &res = first;
(res.insert(res.end(), others.begin(), others.end()), ...);
}
#endif
#line 1 "/home/maspy/compro/library/other/io.hpp"
#define FASTIO
#include <unistd.h>
// https://judge.yosupo.jp/submission/21623
namespace fastio {
static constexpr uint32_t SZ = 1 << 17;
char ibuf[SZ];
char obuf[SZ];
char out[100];
// pointer of ibuf, obuf
uint32_t pil = 0, pir = 0, por = 0;
struct Pre {
char num[10000][4];
constexpr Pre() : num() {
for (int i = 0; i < 10000; i++) {
int n = i;
for (int j = 3; j >= 0; j--) {
num[i][j] = n % 10 | '0';
n /= 10;
}
}
}
} constexpr pre;
inline void load() {
memcpy(ibuf, ibuf + pil, pir - pil);
pir = pir - pil + fread(ibuf + pir - pil, 1, SZ - pir + pil, stdin);
pil = 0;
if (pir < SZ) ibuf[pir++] = '\n';
}
inline void flush() {
fwrite(obuf, 1, por, stdout);
por = 0;
}
void rd(char &c) {
do {
if (pil + 1 > pir) load();
c = ibuf[pil++];
} while (isspace(c));
}
void rd(string &x) {
x.clear();
char c;
do {
if (pil + 1 > pir) load();
c = ibuf[pil++];
} while (isspace(c));
do {
x += c;
if (pil == pir) load();
c = ibuf[pil++];
} while (!isspace(c));
}
template <typename T>
void rd_real(T &x) {
string s;
rd(s);
x = stod(s);
}
template <typename T>
void rd_integer(T &x) {
if (pil + 100 > pir) load();
char c;
do
c = ibuf[pil++];
while (c < '-');
bool minus = 0;
if constexpr (is_signed<T>::value || is_same_v<T, i128>) {
if (c == '-') { minus = 1, c = ibuf[pil++]; }
}
x = 0;
while ('0' <= c) { x = x * 10 + (c & 15), c = ibuf[pil++]; }
if constexpr (is_signed<T>::value || is_same_v<T, i128>) {
if (minus) x = -x;
}
}
void rd(int &x) { rd_integer(x); }
void rd(ll &x) { rd_integer(x); }
void rd(i128 &x) { rd_integer(x); }
void rd(u32 &x) { rd_integer(x); }
void rd(u64 &x) { rd_integer(x); }
void rd(u128 &x) { rd_integer(x); }
void rd(double &x) { rd_real(x); }
void rd(long double &x) { rd_real(x); }
void rd(f128 &x) { rd_real(x); }
template <class T, class U>
void rd(pair<T, U> &p) {
return rd(p.first), rd(p.second);
}
template <size_t N = 0, typename T>
void rd_tuple(T &t) {
if constexpr (N < std::tuple_size<T>::value) {
auto &x = std::get<N>(t);
rd(x);
rd_tuple<N + 1>(t);
}
}
template <class... T>
void rd(tuple<T...> &tpl) {
rd_tuple(tpl);
}
template <size_t N = 0, typename T>
void rd(array<T, N> &x) {
for (auto &d: x) rd(d);
}
template <class T>
void rd(vc<T> &x) {
for (auto &d: x) rd(d);
}
void read() {}
template <class H, class... T>
void read(H &h, T &... t) {
rd(h), read(t...);
}
void wt(const char c) {
if (por == SZ) flush();
obuf[por++] = c;
}
void wt(const string s) {
for (char c: s) wt(c);
}
void wt(const char *s) {
size_t len = strlen(s);
for (size_t i = 0; i < len; i++) wt(s[i]);
}
template <typename T>
void wt_integer(T x) {
if (por > SZ - 100) flush();
if (x < 0) { obuf[por++] = '-', x = -x; }
int outi;
for (outi = 96; x >= 10000; outi -= 4) {
memcpy(out + outi, pre.num[x % 10000], 4);
x /= 10000;
}
if (x >= 1000) {
memcpy(obuf + por, pre.num[x], 4);
por += 4;
} else if (x >= 100) {
memcpy(obuf + por, pre.num[x] + 1, 3);
por += 3;
} else if (x >= 10) {
int q = (x * 103) >> 10;
obuf[por] = q | '0';
obuf[por + 1] = (x - q * 10) | '0';
por += 2;
} else
obuf[por++] = x | '0';
memcpy(obuf + por, out + outi + 4, 96 - outi);
por += 96 - outi;
}
template <typename T>
void wt_real(T x) {
ostringstream oss;
oss << fixed << setprecision(15) << double(x);
string s = oss.str();
wt(s);
}
void wt(int x) { wt_integer(x); }
void wt(ll x) { wt_integer(x); }
void wt(i128 x) { wt_integer(x); }
void wt(u32 x) { wt_integer(x); }
void wt(u64 x) { wt_integer(x); }
void wt(u128 x) { wt_integer(x); }
void wt(double x) { wt_real(x); }
void wt(long double x) { wt_real(x); }
void wt(f128 x) { wt_real(x); }
template <class T, class U>
void wt(const pair<T, U> val) {
wt(val.first);
wt(' ');
wt(val.second);
}
template <size_t N = 0, typename T>
void wt_tuple(const T t) {
if constexpr (N < std::tuple_size<T>::value) {
if constexpr (N > 0) { wt(' '); }
const auto x = std::get<N>(t);
wt(x);
wt_tuple<N + 1>(t);
}
}
template <class... T>
void wt(tuple<T...> tpl) {
wt_tuple(tpl);
}
template <class T, size_t S>
void wt(const array<T, S> val) {
auto n = val.size();
for (size_t i = 0; i < n; i++) {
if (i) wt(' ');
wt(val[i]);
}
}
template <class T>
void wt(const vector<T> val) {
auto n = val.size();
for (size_t i = 0; i < n; i++) {
if (i) wt(' ');
wt(val[i]);
}
}
void print() { wt('\n'); }
template <class Head, class... Tail>
void print(Head &&head, Tail &&... tail) {
wt(head);
if (sizeof...(Tail)) wt(' ');
print(forward<Tail>(tail)...);
}
// gcc expansion. called automaticall after main.
void __attribute__((destructor)) _d() { flush(); }
} // namespace fastio
using fastio::read;
using fastio::print;
using fastio::flush;
#if defined(LOCAL)
#define SHOW(...) SHOW_IMPL(__VA_ARGS__, SHOW6, SHOW5, SHOW4, SHOW3, SHOW2, SHOW1)(__VA_ARGS__)
#define SHOW_IMPL(_1, _2, _3, _4, _5, _6, NAME, ...) NAME
#define SHOW1(x) print(#x, "=", (x)), flush()
#define SHOW2(x, y) print(#x, "=", (x), #y, "=", (y)), flush()
#define SHOW3(x, y, z) print(#x, "=", (x), #y, "=", (y), #z, "=", (z)), flush()
#define SHOW4(x, y, z, w) print(#x, "=", (x), #y, "=", (y), #z, "=", (z), #w, "=", (w)), flush()
#define SHOW5(x, y, z, w, v) print(#x, "=", (x), #y, "=", (y), #z, "=", (z), #w, "=", (w), #v, "=", (v)), flush()
#define SHOW6(x, y, z, w, v, u) print(#x, "=", (x), #y, "=", (y), #z, "=", (z), #w, "=", (w), #v, "=", (v), #u, "=", (u)), flush()
#else
#define SHOW(...)
#endif
#define INT(...) \
int __VA_ARGS__; \
read(__VA_ARGS__)
#define LL(...) \
ll __VA_ARGS__; \
read(__VA_ARGS__)
#define U32(...) \
u32 __VA_ARGS__; \
read(__VA_ARGS__)
#define U64(...) \
u64 __VA_ARGS__; \
read(__VA_ARGS__)
#define STR(...) \
string __VA_ARGS__; \
read(__VA_ARGS__)
#define CHAR(...) \
char __VA_ARGS__; \
read(__VA_ARGS__)
#define DBL(...) \
double __VA_ARGS__; \
read(__VA_ARGS__)
#define VEC(type, name, size) \
vector<type> name(size); \
read(name)
#define VV(type, name, h, w) \
vector<vector<type>> name(h, vector<type>(w)); \
read(name)
void YES(bool t = 1) { print(t ? "YES" : "NO"); }
void NO(bool t = 1) { YES(!t); }
void Yes(bool t = 1) { print(t ? "Yes" : "No"); }
void No(bool t = 1) { Yes(!t); }
void yes(bool t = 1) { print(t ? "yes" : "no"); }
void no(bool t = 1) { yes(!t); }
#line 3 "main.cpp"
#line 2 "/home/maspy/compro/library/mod/modint_common.hpp"
struct has_mod_impl {
template <class T>
static auto check(T &&x) -> decltype(x.get_mod(), std::true_type{});
template <class T>
static auto check(...) -> std::false_type;
};
template <class T>
class has_mod : public decltype(has_mod_impl::check<T>(std::declval<T>())) {};
template <typename mint>
mint inv(int n) {
static const int mod = mint::get_mod();
static vector<mint> dat = {0, 1};
assert(0 <= n);
if (n >= mod) n %= mod;
while (len(dat) <= n) {
int k = len(dat);
int q = (mod + k - 1) / k;
dat.eb(dat[k * q - mod] * mint::raw(q));
}
return dat[n];
}
template <typename mint>
mint fact(int n) {
static const int mod = mint::get_mod();
assert(0 <= n && n < mod);
static vector<mint> dat = {1, 1};
while (len(dat) <= n) dat.eb(dat[len(dat) - 1] * mint::raw(len(dat)));
return dat[n];
}
template <typename mint>
mint fact_inv(int n) {
static vector<mint> dat = {1, 1};
if (n < 0) return mint(0);
while (len(dat) <= n) dat.eb(dat[len(dat) - 1] * inv<mint>(len(dat)));
return dat[n];
}
template <class mint, class... Ts>
mint fact_invs(Ts... xs) {
return (mint(1) * ... * fact_inv<mint>(xs));
}
template <typename mint, class Head, class... Tail>
mint multinomial(Head &&head, Tail &&... tail) {
return fact<mint>(head) * fact_invs<mint>(std::forward<Tail>(tail)...);
}
template <typename mint>
mint C_dense(int n, int k) {
static vvc<mint> C;
static int H = 0, W = 0;
auto calc = [&](int i, int j) -> mint {
if (i == 0) return (j == 0 ? mint(1) : mint(0));
return C[i - 1][j] + (j ? C[i - 1][j - 1] : 0);
};
if (W <= k) {
FOR(i, H) {
C[i].resize(k + 1);
FOR(j, W, k + 1) { C[i][j] = calc(i, j); }
}
W = k + 1;
}
if (H <= n) {
C.resize(n + 1);
FOR(i, H, n + 1) {
C[i].resize(W);
FOR(j, W) { C[i][j] = calc(i, j); }
}
H = n + 1;
}
return C[n][k];
}
template <typename mint, bool large = false, bool dense = false>
mint C(ll n, ll k) {
assert(n >= 0);
if (k < 0 || n < k) return 0;
if constexpr (dense) return C_dense<mint>(n, k);
if constexpr (!large) return multinomial<mint>(n, k, n - k);
k = min(k, n - k);
mint x(1);
FOR(i, k) x *= mint(n - i);
return x * fact_inv<mint>(k);
}
template <typename mint, bool large = false>
mint C_inv(ll n, ll k) {
assert(n >= 0);
assert(0 <= k && k <= n);
if (!large) return fact_inv<mint>(n) * fact<mint>(k) * fact<mint>(n - k);
return mint(1) / C<mint, 1>(n, k);
}
// [x^d](1-x)^{-n}
template <typename mint, bool large = false, bool dense = false>
mint C_negative(ll n, ll d) {
assert(n >= 0);
if (d < 0) return mint(0);
if (n == 0) { return (d == 0 ? mint(1) : mint(0)); }
return C<mint, large, dense>(n + d - 1, d);
}
#line 3 "/home/maspy/compro/library/mod/modint.hpp"
template <int mod>
struct modint {
static constexpr u32 umod = u32(mod);
static_assert(umod < u32(1) << 31);
u32 val;
static modint raw(u32 v) {
modint x;
x.val = v;
return x;
}
constexpr modint() : val(0) {}
constexpr modint(u32 x) : val(x % umod) {}
constexpr modint(u64 x) : val(x % umod) {}
constexpr modint(u128 x) : val(x % umod) {}
constexpr modint(int x) : val((x %= mod) < 0 ? x + mod : x){};
constexpr modint(ll x) : val((x %= mod) < 0 ? x + mod : x){};
constexpr modint(i128 x) : val((x %= mod) < 0 ? x + mod : x){};
bool operator<(const modint &other) const { return val < other.val; }
modint &operator+=(const modint &p) {
if ((val += p.val) >= umod) val -= umod;
return *this;
}
modint &operator-=(const modint &p) {
if ((val += umod - p.val) >= umod) val -= umod;
return *this;
}
modint &operator*=(const modint &p) {
val = u64(val) * p.val % umod;
return *this;
}
modint &operator/=(const modint &p) {
*this *= p.inverse();
return *this;
}
modint operator-() const { return modint::raw(val ? mod - val : u32(0)); }
modint operator+(const modint &p) const { return modint(*this) += p; }
modint operator-(const modint &p) const { return modint(*this) -= p; }
modint operator*(const modint &p) const { return modint(*this) *= p; }
modint operator/(const modint &p) const { return modint(*this) /= p; }
bool operator==(const modint &p) const { return val == p.val; }
bool operator!=(const modint &p) const { return val != p.val; }
modint inverse() const {
int a = val, b = mod, u = 1, v = 0, t;
while (b > 0) {
t = a / b;
swap(a -= t * b, b), swap(u -= t * v, v);
}
return modint(u);
}
modint pow(ll n) const {
assert(n >= 0);
modint ret(1), mul(val);
while (n > 0) {
if (n & 1) ret *= mul;
mul *= mul;
n >>= 1;
}
return ret;
}
static constexpr int get_mod() { return mod; }
// (n, r), r は 1 の 2^n 乗根
static constexpr pair<int, int> ntt_info() {
if (mod == 120586241) return {20, 74066978};
if (mod == 167772161) return {25, 17};
if (mod == 469762049) return {26, 30};
if (mod == 754974721) return {24, 362};
if (mod == 880803841) return {23, 211};
if (mod == 943718401) return {22, 663003469};
if (mod == 998244353) return {23, 31};
if (mod == 1004535809) return {21, 836905998};
if (mod == 1045430273) return {20, 363};
if (mod == 1051721729) return {20, 330};
if (mod == 1053818881) return {20, 2789};
return {-1, -1};
}
static constexpr bool can_ntt() { return ntt_info().fi != -1; }
};
#ifdef FASTIO
template <int mod>
void rd(modint<mod> &x) {
fastio::rd(x.val);
x.val %= mod;
// assert(0 <= x.val && x.val < mod);
}
template <int mod>
void wt(modint<mod> x) {
fastio::wt(x.val);
}
#endif
using modint107 = modint<1000000007>;
using modint998 = modint<998244353>;
#line 2 "/home/maspy/compro/library/ds/hashmap.hpp"
// u64 -> Val
template <typename Val>
struct HashMap {
// n は入れたいものの個数で ok
HashMap(u32 n = 0) { build(n); }
void build(u32 n) {
u32 k = 8;
while (k < n * 2) k *= 2;
cap = k / 2, mask = k - 1;
key.resize(k), val.resize(k), used.assign(k, 0);
}
// size を保ったまま. size=0 にするときは build すること.
void clear() {
used.assign(len(used), 0);
cap = (mask + 1) / 2;
}
int size() { return len(used) / 2 - cap; }
int index(const u64& k) {
int i = 0;
for (i = hash(k); used[i] && key[i] != k; i = (i + 1) & mask) {}
return i;
}
Val& operator[](const u64& k) {
if (cap == 0) extend();
int i = index(k);
if (!used[i]) { used[i] = 1, key[i] = k, val[i] = Val{}, --cap; }
return val[i];
}
Val get(const u64& k, Val default_value) {
int i = index(k);
return (used[i] ? val[i] : default_value);
}
bool count(const u64& k) {
int i = index(k);
return used[i] && key[i] == k;
}
// f(key, val)
template <typename F>
void enumerate_all(F f) {
FOR(i, len(used)) if (used[i]) f(key[i], val[i]);
}
private:
u32 cap, mask;
vc<u64> key;
vc<Val> val;
vc<bool> used;
u64 hash(u64 x) {
static const u64 FIXED_RANDOM = std::chrono::steady_clock::now().time_since_epoch().count();
x += FIXED_RANDOM;
x = (x ^ (x >> 30)) * 0xbf58476d1ce4e5b9;
x = (x ^ (x >> 27)) * 0x94d049bb133111eb;
return (x ^ (x >> 31)) & mask;
}
void extend() {
vc<pair<u64, Val>> dat;
dat.reserve(len(used) / 2 - cap);
FOR(i, len(used)) {
if (used[i]) dat.eb(key[i], val[i]);
}
build(2 * len(dat));
for (auto& [a, b]: dat) (*this)[a] = b;
}
};
#line 2 "/home/maspy/compro/library/random/hash_vector.hpp"
#line 2 "/home/maspy/compro/library/random/base.hpp"
u64 RNG_64() {
static uint64_t x_
= uint64_t(chrono::duration_cast<chrono::nanoseconds>(chrono::high_resolution_clock::now().time_since_epoch()).count()) * 10150724397891781847ULL;
x_ ^= x_ << 7;
return x_ ^= x_ >> 9;
}
u64 RNG(u64 lim) { return RNG_64() % lim; }
ll RNG(ll l, ll r) { return l + RNG_64() % (r - l); }
#line 2 "/home/maspy/compro/library/mod/modint61.hpp"
struct modint61 {
static constexpr u64 mod = (1ULL << 61) - 1;
u64 val;
constexpr modint61() : val(0ULL) {}
constexpr modint61(u32 x) : val(x) {}
constexpr modint61(u64 x) : val(x % mod) {}
constexpr modint61(int x) : val((x < 0) ? (x + static_cast<ll>(mod)) : x) {}
constexpr modint61(ll x) : val(((x %= static_cast<ll>(mod)) < 0) ? (x + static_cast<ll>(mod)) : x) {}
static constexpr u64 get_mod() { return mod; }
modint61 &operator+=(const modint61 &a) {
val = ((val += a.val) >= mod) ? (val - mod) : val;
return *this;
}
modint61 &operator-=(const modint61 &a) {
val = ((val -= a.val) >= mod) ? (val + mod) : val;
return *this;
}
modint61 &operator*=(const modint61 &a) {
const unsigned __int128 y = static_cast<unsigned __int128>(val) * a.val;
val = (y >> 61) + (y & mod);
val = (val >= mod) ? (val - mod) : val;
return *this;
}
modint61 operator-() const { return modint61(val ? mod - val : u64(0)); }
modint61 &operator/=(const modint61 &a) { return (*this *= a.inverse()); }
modint61 operator+(const modint61 &p) const { return modint61(*this) += p; }
modint61 operator-(const modint61 &p) const { return modint61(*this) -= p; }
modint61 operator*(const modint61 &p) const { return modint61(*this) *= p; }
modint61 operator/(const modint61 &p) const { return modint61(*this) /= p; }
bool operator<(const modint61 &other) const { return val < other.val; }
bool operator==(const modint61 &p) const { return val == p.val; }
bool operator!=(const modint61 &p) const { return val != p.val; }
modint61 inverse() const {
ll a = val, b = mod, u = 1, v = 0, t;
while (b > 0) {
t = a / b;
swap(a -= t * b, b), swap(u -= t * v, v);
}
return modint61(u);
}
modint61 pow(ll n) const {
assert(n >= 0);
modint61 ret(1), mul(val);
while (n > 0) {
if (n & 1) ret *= mul;
mul *= mul, n >>= 1;
}
return ret;
}
};
#ifdef FASTIO
void rd(modint61 &x) {
fastio::rd(x.val);
assert(0 <= x.val && x.val < modint61::mod);
}
void wt(modint61 x) { fastio::wt(x.val); }
#endif
#line 5 "/home/maspy/compro/library/random/hash_vector.hpp"
template <typename T>
u64 hash_vector(vc<T> X) {
using mint = modint61;
static vc<mint> hash_base;
int n = len(X);
while (len(hash_base) <= n) { hash_base.eb(RNG(mint::get_mod())); }
mint H = 0;
FOR(i, n) H += hash_base[i] * mint(X[i]);
H += hash_base[n];
return H.val;
}
#line 2 "/home/maspy/compro/library/ds/unionfind/unionfind.hpp"
struct UnionFind {
int n, n_comp;
vc<int> dat; // par or (-size)
UnionFind(int n = 0) { build(n); }
void build(int m) {
n = m, n_comp = m;
dat.assign(n, -1);
}
void reset() { build(n); }
int operator[](int x) {
while (dat[x] >= 0) {
int pp = dat[dat[x]];
if (pp < 0) { return dat[x]; }
x = dat[x] = pp;
}
return x;
}
ll size(int x) {
x = (*this)[x];
return -dat[x];
}
bool merge(int x, int y) {
x = (*this)[x], y = (*this)[y];
if (x == y) return false;
if (-dat[x] < -dat[y]) swap(x, y);
dat[x] += dat[y], dat[y] = x, n_comp--;
return true;
}
vc<int> get_all() {
vc<int> A(n);
FOR(i, n) A[i] = (*this)[i];
return A;
}
};
#line 2 "/home/maspy/compro/library/seq/interpolate_linear_rec.hpp"
#line 2 "/home/maspy/compro/library/seq/find_linear_rec.hpp"
template <typename mint>
vector<mint> find_linear_rec(vector<mint>& A) {
int N = len(A);
vc<mint> B = {1}, C = {1};
int l = 0, m = 1;
mint p = 1;
FOR(i, N) {
mint d = A[i];
FOR3(j, 1, l + 1) { d += C[j] * A[i - j]; }
if (d == 0) {
++m;
continue;
}
auto tmp = C;
mint q = d / p;
if (len(C) < len(B) + m) C.insert(C.end(), len(B) + m - len(C), 0);
FOR(j, len(B)) C[j + m] -= q * B[j];
if (l + l <= i) {
B = tmp;
l = i + 1 - l, m = 1;
p = d;
} else {
++m;
}
}
return C;
}
#line 2 "/home/maspy/compro/library/poly/fps_div.hpp"
#line 2 "/home/maspy/compro/library/poly/count_terms.hpp"
template<typename mint>
int count_terms(const vc<mint>& f){
int t = 0;
FOR(i, len(f)) if(f[i] != mint(0)) ++t;
return t;
}
#line 2 "/home/maspy/compro/library/mod/mod_inv.hpp"
// long でも大丈夫
// (val * x - 1) が mod の倍数になるようにする
// 特に mod=0 なら x=0 が満たす
ll mod_inv(ll val, ll mod) {
if (mod == 0) return 0;
mod = abs(mod);
val %= mod;
if (val < 0) val += mod;
ll a = val, b = mod, u = 1, v = 0, t;
while (b > 0) {
t = a / b;
swap(a -= t * b, b), swap(u -= t * v, v);
}
if (u < 0) u += mod;
return u;
}
#line 2 "/home/maspy/compro/library/mod/crt3.hpp"
constexpr u32 mod_pow_constexpr(u64 a, u64 n, u32 mod) {
a %= mod;
u64 res = 1;
FOR(32) {
if (n & 1) res = res * a % mod;
a = a * a % mod, n /= 2;
}
return res;
}
template <typename T, u32 p0, u32 p1>
T CRT2(u64 a0, u64 a1) {
static_assert(p0 < p1);
static constexpr u64 x0_1 = mod_pow_constexpr(p0, p1 - 2, p1);
u64 c = (a1 - a0 + p1) * x0_1 % p1;
return a0 + c * p0;
}
template <typename T, u32 p0, u32 p1, u32 p2>
T CRT3(u64 a0, u64 a1, u64 a2) {
static_assert(p0 < p1 && p1 < p2);
static constexpr u64 x1 = mod_pow_constexpr(p0, p1 - 2, p1);
static constexpr u64 x2 = mod_pow_constexpr(u64(p0) * p1 % p2, p2 - 2, p2);
static constexpr u64 p01 = u64(p0) * p1;
u64 c = (a1 - a0 + p1) * x1 % p1;
u64 ans_1 = a0 + c * p0;
c = (a2 - ans_1 % p2 + p2) * x2 % p2;
return T(ans_1) + T(c) * T(p01);
}
template <typename T, u32 p0, u32 p1, u32 p2, u32 p3, u32 p4>
T CRT5(u64 a0, u64 a1, u64 a2, u64 a3, u64 a4) {
static_assert(p0 < p1 && p1 < p2 && p2 < p3 && p3 < p4);
static constexpr u64 x1 = mod_pow_constexpr(p0, p1 - 2, p1);
static constexpr u64 x2 = mod_pow_constexpr(u64(p0) * p1 % p2, p2 - 2, p2);
static constexpr u64 x3
= mod_pow_constexpr(u64(p0) * p1 % p3 * p2 % p3, p3 - 2, p3);
static constexpr u64 x4
= mod_pow_constexpr(u64(p0) * p1 % p4 * p2 % p4 * p3 % p4, p4 - 2, p4);
static constexpr u64 p01 = u64(p0) * p1;
static constexpr u64 p23 = u64(p2) * p3;
u64 c = (a1 - a0 + p1) * x1 % p1;
u64 ans_1 = a0 + c * p0;
c = (a2 - ans_1 % p2 + p2) * x2 % p2;
u128 ans_2 = ans_1 + c * static_cast<u128>(p01);
c = static_cast<u64>(a3 - ans_2 % p3 + p3) * x3 % p3;
u128 ans_3 = ans_2 + static_cast<u128>(c * p2) * p01;
c = static_cast<u64>(a4 - ans_3 % p4 + p4) * x4 % p4;
return T(ans_3) + T(c) * T(p01) * T(p23);
}
#line 2 "/home/maspy/compro/library/poly/convolution_naive.hpp"
template <class T, typename enable_if<!has_mod<T>::value>::type* = nullptr>
vc<T> convolution_naive(const vc<T>& a, const vc<T>& b) {
int n = int(a.size()), m = int(b.size());
if (n > m) return convolution_naive<T>(b, a);
if (n == 0) return {};
vector<T> ans(n + m - 1);
FOR(i, n) FOR(j, m) ans[i + j] += a[i] * b[j];
return ans;
}
template <class T, typename enable_if<has_mod<T>::value>::type* = nullptr>
vc<T> convolution_naive(const vc<T>& a, const vc<T>& b) {
int n = int(a.size()), m = int(b.size());
if (n > m) return convolution_naive<T>(b, a);
if (n == 0) return {};
vc<T> ans(n + m - 1);
if (n <= 16 && (T::get_mod() < (1 << 30))) {
for (int k = 0; k < n + m - 1; ++k) {
int s = max(0, k - m + 1);
int t = min(n, k + 1);
u64 sm = 0;
for (int i = s; i < t; ++i) { sm += u64(a[i].val) * (b[k - i].val); }
ans[k] = sm;
}
} else {
for (int k = 0; k < n + m - 1; ++k) {
int s = max(0, k - m + 1);
int t = min(n, k + 1);
u128 sm = 0;
for (int i = s; i < t; ++i) { sm += u64(a[i].val) * (b[k - i].val); }
ans[k] = T::raw(sm % T::get_mod());
}
}
return ans;
}
#line 2 "/home/maspy/compro/library/poly/convolution_karatsuba.hpp"
// 任意の環でできる
template <typename T>
vc<T> convolution_karatsuba(const vc<T>& f, const vc<T>& g) {
const int thresh = 30;
if (min(len(f), len(g)) <= thresh) return convolution_naive(f, g);
int n = max(len(f), len(g));
int m = ceil(n, 2);
vc<T> f1, f2, g1, g2;
if (len(f) < m) f1 = f;
if (len(f) >= m) f1 = {f.begin(), f.begin() + m};
if (len(f) >= m) f2 = {f.begin() + m, f.end()};
if (len(g) < m) g1 = g;
if (len(g) >= m) g1 = {g.begin(), g.begin() + m};
if (len(g) >= m) g2 = {g.begin() + m, g.end()};
vc<T> a = convolution_karatsuba(f1, g1);
vc<T> b = convolution_karatsuba(f2, g2);
FOR(i, len(f2)) f1[i] += f2[i];
FOR(i, len(g2)) g1[i] += g2[i];
vc<T> c = convolution_karatsuba(f1, g1);
vc<T> F(len(f) + len(g) - 1);
assert(2 * m + len(b) <= len(F));
FOR(i, len(a)) F[i] += a[i], c[i] -= a[i];
FOR(i, len(b)) F[2 * m + i] += b[i], c[i] -= b[i];
if (c.back() == T(0)) c.pop_back();
FOR(i, len(c)) if (c[i] != T(0)) F[m + i] += c[i];
return F;
}
#line 2 "/home/maspy/compro/library/poly/ntt.hpp"
template <class mint>
void ntt(vector<mint>& a, bool inverse) {
assert(mint::can_ntt());
const int rank2 = mint::ntt_info().fi;
const int mod = mint::get_mod();
static array<mint, 30> root, iroot;
static array<mint, 30> rate2, irate2;
static array<mint, 30> rate3, irate3;
assert(rank2 != -1 && len(a) <= (1 << max(0, rank2)));
static bool prepared = 0;
if (!prepared) {
prepared = 1;
root[rank2] = mint::ntt_info().se;
iroot[rank2] = mint(1) / root[rank2];
FOR_R(i, rank2) {
root[i] = root[i + 1] * root[i + 1];
iroot[i] = iroot[i + 1] * iroot[i + 1];
}
mint prod = 1, iprod = 1;
for (int i = 0; i <= rank2 - 2; i++) {
rate2[i] = root[i + 2] * prod;
irate2[i] = iroot[i + 2] * iprod;
prod *= iroot[i + 2];
iprod *= root[i + 2];
}
prod = 1, iprod = 1;
for (int i = 0; i <= rank2 - 3; i++) {
rate3[i] = root[i + 3] * prod;
irate3[i] = iroot[i + 3] * iprod;
prod *= iroot[i + 3];
iprod *= root[i + 3];
}
}
int n = int(a.size());
int h = topbit(n);
assert(n == 1 << h);
if (!inverse) {
int len = 0;
while (len < h) {
if (h - len == 1) {
int p = 1 << (h - len - 1);
mint rot = 1;
FOR(s, 1 << len) {
int offset = s << (h - len);
FOR(i, p) {
auto l = a[i + offset];
auto r = a[i + offset + p] * rot;
a[i + offset] = l + r;
a[i + offset + p] = l - r;
}
rot *= rate2[topbit(~s & -~s)];
}
len++;
} else {
int p = 1 << (h - len - 2);
mint rot = 1, imag = root[2];
for (int s = 0; s < (1 << len); s++) {
mint rot2 = rot * rot;
mint rot3 = rot2 * rot;
int offset = s << (h - len);
for (int i = 0; i < p; i++) {
u64 mod2 = u64(mod) * mod;
u64 a0 = a[i + offset].val;
u64 a1 = u64(a[i + offset + p].val) * rot.val;
u64 a2 = u64(a[i + offset + 2 * p].val) * rot2.val;
u64 a3 = u64(a[i + offset + 3 * p].val) * rot3.val;
u64 a1na3imag = (a1 + mod2 - a3) % mod * imag.val;
u64 na2 = mod2 - a2;
a[i + offset] = a0 + a2 + a1 + a3;
a[i + offset + 1 * p] = a0 + a2 + (2 * mod2 - (a1 + a3));
a[i + offset + 2 * p] = a0 + na2 + a1na3imag;
a[i + offset + 3 * p] = a0 + na2 + (mod2 - a1na3imag);
}
rot *= rate3[topbit(~s & -~s)];
}
len += 2;
}
}
} else {
mint coef = mint(1) / mint(len(a));
FOR(i, len(a)) a[i] *= coef;
int len = h;
while (len) {
if (len == 1) {
int p = 1 << (h - len);
mint irot = 1;
FOR(s, 1 << (len - 1)) {
int offset = s << (h - len + 1);
FOR(i, p) {
u64 l = a[i + offset].val;
u64 r = a[i + offset + p].val;
a[i + offset] = l + r;
a[i + offset + p] = (mod + l - r) * irot.val;
}
irot *= irate2[topbit(~s & -~s)];
}
len--;
} else {
int p = 1 << (h - len);
mint irot = 1, iimag = iroot[2];
FOR(s, (1 << (len - 2))) {
mint irot2 = irot * irot;
mint irot3 = irot2 * irot;
int offset = s << (h - len + 2);
for (int i = 0; i < p; i++) {
u64 a0 = a[i + offset + 0 * p].val;
u64 a1 = a[i + offset + 1 * p].val;
u64 a2 = a[i + offset + 2 * p].val;
u64 a3 = a[i + offset + 3 * p].val;
u64 x = (mod + a2 - a3) * iimag.val % mod;
a[i + offset] = a0 + a1 + a2 + a3;
a[i + offset + 1 * p] = (a0 + mod - a1 + x) * irot.val;
a[i + offset + 2 * p] = (a0 + a1 + 2 * mod - a2 - a3) * irot2.val;
a[i + offset + 3 * p] = (a0 + 2 * mod - a1 - x) * irot3.val;
}
irot *= irate3[topbit(~s & -~s)];
}
len -= 2;
}
}
}
}
#line 1 "/home/maspy/compro/library/poly/fft.hpp"
namespace CFFT {
using real = double;
struct C {
real x, y;
C() : x(0), y(0) {}
C(real x, real y) : x(x), y(y) {}
inline C operator+(const C& c) const { return C(x + c.x, y + c.y); }
inline C operator-(const C& c) const { return C(x - c.x, y - c.y); }
inline C operator*(const C& c) const {
return C(x * c.x - y * c.y, x * c.y + y * c.x);
}
inline C conj() const { return C(x, -y); }
};
const real PI = acosl(-1);
int base = 1;
vector<C> rts = {{0, 0}, {1, 0}};
vector<int> rev = {0, 1};
void ensure_base(int nbase) {
if (nbase <= base) return;
rev.resize(1 << nbase);
rts.resize(1 << nbase);
for (int i = 0; i < (1 << nbase); i++) {
rev[i] = (rev[i >> 1] >> 1) + ((i & 1) << (nbase - 1));
}
while (base < nbase) {
real angle = PI * 2.0 / (1 << (base + 1));
for (int i = 1 << (base - 1); i < (1 << base); i++) {
rts[i << 1] = rts[i];
real angle_i = angle * (2 * i + 1 - (1 << base));
rts[(i << 1) + 1] = C(cos(angle_i), sin(angle_i));
}
++base;
}
}
void fft(vector<C>& a, int n) {
assert((n & (n - 1)) == 0);
int zeros = __builtin_ctz(n);
ensure_base(zeros);
int shift = base - zeros;
for (int i = 0; i < n; i++) {
if (i < (rev[i] >> shift)) { swap(a[i], a[rev[i] >> shift]); }
}
for (int k = 1; k < n; k <<= 1) {
for (int i = 0; i < n; i += 2 * k) {
for (int j = 0; j < k; j++) {
C z = a[i + j + k] * rts[j + k];
a[i + j + k] = a[i + j] - z;
a[i + j] = a[i + j] + z;
}
}
}
}
} // namespace CFFT
#line 9 "/home/maspy/compro/library/poly/convolution.hpp"
template <class mint>
vector<mint> convolution_ntt(vector<mint> a, vector<mint> b) {
if (a.empty() || b.empty()) return {};
int n = int(a.size()), m = int(b.size());
int sz = 1;
while (sz < n + m - 1) sz *= 2;
// sz = 2^k のときの高速化。分割統治的なやつで損しまくるので。
if ((n + m - 3) <= sz / 2) {
auto a_last = a.back(), b_last = b.back();
a.pop_back(), b.pop_back();
auto c = convolution(a, b);
c.resize(n + m - 1);
c[n + m - 2] = a_last * b_last;
FOR(i, len(a)) c[i + len(b)] += a[i] * b_last;
FOR(i, len(b)) c[i + len(a)] += b[i] * a_last;
return c;
}
a.resize(sz), b.resize(sz);
bool same = a == b;
ntt(a, 0);
if (same) {
b = a;
} else {
ntt(b, 0);
}
FOR(i, sz) a[i] *= b[i];
ntt(a, 1);
a.resize(n + m - 1);
return a;
}
template <typename mint>
vector<mint> convolution_garner(const vector<mint>& a, const vector<mint>& b) {
int n = len(a), m = len(b);
if (!n || !m) return {};
static constexpr int p0 = 167772161;
static constexpr int p1 = 469762049;
static constexpr int p2 = 754974721;
using mint0 = modint<p0>;
using mint1 = modint<p1>;
using mint2 = modint<p2>;
vc<mint0> a0(n), b0(m);
vc<mint1> a1(n), b1(m);
vc<mint2> a2(n), b2(m);
FOR(i, n) a0[i] = a[i].val, a1[i] = a[i].val, a2[i] = a[i].val;
FOR(i, m) b0[i] = b[i].val, b1[i] = b[i].val, b2[i] = b[i].val;
auto c0 = convolution_ntt<mint0>(a0, b0);
auto c1 = convolution_ntt<mint1>(a1, b1);
auto c2 = convolution_ntt<mint2>(a2, b2);
vc<mint> c(len(c0));
FOR(i, n + m - 1) { c[i] = CRT3<mint, p0, p1, p2>(c0[i].val, c1[i].val, c2[i].val); }
return c;
}
template <typename R>
vc<double> convolution_fft(const vc<R>& a, const vc<R>& b) {
using C = CFFT::C;
int need = (int)a.size() + (int)b.size() - 1;
int nbase = 1;
while ((1 << nbase) < need) nbase++;
CFFT::ensure_base(nbase);
int sz = 1 << nbase;
vector<C> fa(sz);
for (int i = 0; i < sz; i++) {
double x = (i < (int)a.size() ? a[i] : 0);
double y = (i < (int)b.size() ? b[i] : 0);
fa[i] = C(x, y);
}
CFFT::fft(fa, sz);
C r(0, -0.25 / (sz >> 1)), s(0, 1), t(0.5, 0);
for (int i = 0; i <= (sz >> 1); i++) {
int j = (sz - i) & (sz - 1);
C z = (fa[j] * fa[j] - (fa[i] * fa[i]).conj()) * r;
fa[j] = (fa[i] * fa[i] - (fa[j] * fa[j]).conj()) * r;
fa[i] = z;
}
for (int i = 0; i < (sz >> 1); i++) {
C A0 = (fa[i] + fa[i + (sz >> 1)]) * t;
C A1 = (fa[i] - fa[i + (sz >> 1)]) * t * CFFT::rts[(sz >> 1) + i];
fa[i] = A0 + A1 * s;
}
CFFT::fft(fa, sz >> 1);
vector<double> ret(need);
for (int i = 0; i < need; i++) { ret[i] = (i & 1 ? fa[i >> 1].y : fa[i >> 1].x); }
return ret;
}
vector<ll> convolution(const vector<ll>& a, const vector<ll>& b) {
int n = len(a), m = len(b);
if (!n || !m) return {};
if (min(n, m) <= 2500) return convolution_naive(a, b);
ll abs_sum_a = 0, abs_sum_b = 0;
ll LIM = 1e15;
FOR(i, n) abs_sum_a = min(LIM, abs_sum_a + abs(a[i]));
FOR(i, m) abs_sum_b = min(LIM, abs_sum_b + abs(b[i]));
if (i128(abs_sum_a) * abs_sum_b < 1e15) {
vc<double> c = convolution_fft<ll>(a, b);
vc<ll> res(len(c));
FOR(i, len(c)) res[i] = ll(floor(c[i] + .5));
return res;
}
static constexpr u32 MOD1 = 167772161; // 2^25
static constexpr u32 MOD2 = 469762049; // 2^26
static constexpr u32 MOD3 = 754974721; // 2^24
using mint1 = modint<MOD1>;
using mint2 = modint<MOD2>;
using mint3 = modint<MOD3>;
vc<mint1> a1(n), b1(m);
vc<mint2> a2(n), b2(m);
vc<mint3> a3(n), b3(m);
FOR(i, n) a1[i] = a[i], a2[i] = a[i], a3[i] = a[i];
FOR(i, m) b1[i] = b[i], b2[i] = b[i], b3[i] = b[i];
auto c1 = convolution_ntt<mint1>(a1, b1);
auto c2 = convolution_ntt<mint2>(a2, b2);
auto c3 = convolution_ntt<mint3>(a3, b3);
u128 prod = u128(MOD1) * MOD2 * MOD3;
vc<ll> res(n + m - 1);
FOR(i, n + m - 1) {
u128 x = CRT3<u128, MOD1, MOD2, MOD3>(c1[i].val, c2[i].val, c3[i].val);
res[i] = (x < prod / 2 ? ll(x) : -ll(prod - x));
}
return res;
}
template <typename mint>
vc<mint> convolution(const vc<mint>& a, const vc<mint>& b) {
int n = len(a), m = len(b);
if (!n || !m) return {};
if (mint::can_ntt()) {
if (min(n, m) <= 50) return convolution_karatsuba<mint>(a, b);
return convolution_ntt(a, b);
}
if (min(n, m) <= 200) return convolution_karatsuba<mint>(a, b);
return convolution_garner(a, b);
}
#line 4 "/home/maspy/compro/library/poly/fps_inv.hpp"
template <typename mint>
vc<mint> fps_inv_sparse(const vc<mint>& f) {
int N = len(f);
vc<pair<int, mint>> dat;
FOR(i, 1, N) if (f[i] != mint(0)) dat.eb(i, f[i]);
vc<mint> g(N);
mint g0 = mint(1) / f[0];
g[0] = g0;
FOR(n, 1, N) {
mint rhs = 0;
for (auto&& [k, fk]: dat) {
if (k > n) break;
rhs -= fk * g[n - k];
}
g[n] = rhs * g0;
}
return g;
}
template <typename mint>
vc<mint> fps_inv_dense_ntt(const vc<mint>& F) {
vc<mint> G = {mint(1) / F[0]};
ll N = len(F), n = 1;
G.reserve(N);
while (n < N) {
vc<mint> f(2 * n), g(2 * n);
FOR(i, min(N, 2 * n)) f[i] = F[i];
FOR(i, n) g[i] = G[i];
ntt(f, false), ntt(g, false);
FOR(i, 2 * n) f[i] *= g[i];
ntt(f, true);
FOR(i, n) f[i] = 0;
ntt(f, false);
FOR(i, 2 * n) f[i] *= g[i];
ntt(f, true);
FOR(i, n, min(N, 2 * n)) G.eb(-f[i]);
n *= 2;
}
return G;
}
template <typename mint>
vc<mint> fps_inv_dense(const vc<mint>& F) {
if (mint::can_ntt()) return fps_inv_dense_ntt(F);
const int N = len(F);
vc<mint> R = {mint(1) / F[0]};
vc<mint> p;
int m = 1;
while (m < N) {
p = convolution(R, R);
p.resize(m + m);
vc<mint> f = {F.begin(), F.begin() + min(m + m, N)};
p = convolution(p, f);
R.resize(m + m);
FOR(i, m + m) R[i] = R[i] + R[i] - p[i];
m += m;
}
R.resize(N);
return R;
}
template <typename mint>
vc<mint> fps_inv(const vc<mint>& f) {
assert(f[0] != mint(0));
int n = count_terms(f);
int t = (mint::can_ntt() ? 160 : 820);
return (n <= t ? fps_inv_sparse<mint>(f) : fps_inv_dense<mint>(f));
}
#line 5 "/home/maspy/compro/library/poly/fps_div.hpp"
// f/g. f の長さで出力される.
template <typename mint, bool SPARSE = false>
vc<mint> fps_div(vc<mint> f, vc<mint> g) {
if (SPARSE || count_terms(g) < 200) return fps_div_sparse(f, g);
int n = len(f);
g.resize(n);
g = fps_inv<mint>(g);
f = convolution(f, g);
f.resize(n);
return f;
}
// f/g ただし g は sparse
template <typename mint>
vc<mint> fps_div_sparse(vc<mint> f, vc<mint>& g) {
if (g[0] != mint(1)) {
mint cf = g[0].inverse();
for (auto&& x: f) x *= cf;
for (auto&& x: g) x *= cf;
}
vc<pair<int, mint>> dat;
FOR(i, 1, len(g)) if (g[i] != mint(0)) dat.eb(i, -g[i]);
FOR(i, len(f)) {
for (auto&& [j, x]: dat) {
if (i >= j) f[i] += x * f[i - j];
}
}
return f;
}
#line 2 "/home/maspy/compro/library/poly/ntt_doubling.hpp"
#line 4 "/home/maspy/compro/library/poly/ntt_doubling.hpp"
// 2^k 次多項式の長さ 2^k が与えられるので 2^k+1 にする
template <typename mint, bool transposed = false>
void ntt_doubling(vector<mint>& a) {
static array<mint, 30> root;
static bool prepared = 0;
if (!prepared) {
prepared = 1;
const int rank2 = mint::ntt_info().fi;
root[rank2] = mint::ntt_info().se;
FOR_R(i, rank2) { root[i] = root[i + 1] * root[i + 1]; }
}
if constexpr (!transposed) {
const int M = (int)a.size();
auto b = a;
ntt(b, 1);
mint r = 1, zeta = root[topbit(2 * M)];
FOR(i, M) b[i] *= r, r *= zeta;
ntt(b, 0);
copy(begin(b), end(b), back_inserter(a));
} else {
const int M = len(a) / 2;
vc<mint> tmp = {a.begin(), a.begin() + M};
a = {a.begin() + M, a.end()};
transposed_ntt(a, 0);
mint r = 1, zeta = root[topbit(2 * M)];
FOR(i, M) a[i] *= r, r *= zeta;
transposed_ntt(a, 1);
FOR(i, M) a[i] += tmp[i];
}
}
#line 2 "/home/maspy/compro/library/poly/poly_divmod.hpp"
#line 4 "/home/maspy/compro/library/poly/poly_divmod.hpp"
template <typename mint>
pair<vc<mint>, vc<mint>> poly_divmod(vc<mint> f, vc<mint> g) {
assert(g.back() != 0);
if (len(f) < len(g)) { return {{}, f}; }
auto rf = f, rg = g;
reverse(all(rf)), reverse(all(rg));
ll deg = len(rf) - len(rg) + 1;
rf.resize(deg), rg.resize(deg);
rg = fps_inv(rg);
auto q = convolution(rf, rg);
q.resize(deg);
reverse(all(q));
auto h = convolution(q, g);
FOR(i, len(f)) f[i] -= h[i];
while (len(f) > 0 && f.back() == 0) f.pop_back();
return {q, f};
}
#line 4 "/home/maspy/compro/library/poly/coef_of_rational_fps.hpp"
template <typename mint>
mint coef_of_rational_fps_small(vector<mint> P, vector<mint> Q, ll N) {
assert(0 <= len(P) && len(P) + 1 == len(Q) && len(Q) <= 16
&& Q[0] == mint(1));
if (P.empty()) return 0;
int m = len(Q) - 1;
vc<u32> Q32(m + 1);
FOR(i, m + 1) Q32[i] = (-Q[i]).val;
using poly = vc<u64>;
auto dfs = [&](auto& dfs, const ll N) -> poly {
// x^N mod G
if (N == 0) {
poly f(m);
f[0] = 1;
return f;
}
poly f = dfs(dfs, N / 2);
poly g(len(f) * 2 - 1 + (N & 1));
FOR(i, len(f)) FOR(j, len(f)) { g[i + j + (N & 1)] += f[i] * f[j]; }
FOR(i, len(g)) g[i] = mint(g[i]).val;
FOR_R(i, len(g)) {
g[i] = mint(g[i]).val;
if (i >= m) FOR(j, 1, len(Q)) g[i - j] += Q32[j] * g[i];
}
g.resize(m);
return g;
};
poly f = dfs(dfs, N);
FOR(i, m) FOR(j, 1, i + 1) { P[i] -= Q[j] * P[i - j]; }
u64 res = 0;
FOR(i, m) res += f[i] * P[i].val;
return res;
}
template <typename mint>
mint coef_of_rational_fps_ntt(vector<mint> P, vector<mint> Q, ll N) {
assert(0 <= len(P) && len(P) + 1 == len(Q) && Q[0] == mint(1));
if (P.empty()) return 0;
int n = 1;
while (n < len(Q)) n += n;
vc<mint> W(n);
{
vc<int> btr(n);
int log = topbit(n);
FOR(i, n) { btr[i] = (btr[i >> 1] >> 1) + ((i & 1) << (log - 1)); }
int t = mint::ntt_info().fi;
mint r = mint::ntt_info().se;
mint dw = r.inverse().pow((1 << t) / (2 * n));
mint w = inv<mint>(2);
for (auto& i: btr) { W[i] = w, w *= dw; }
}
P.resize(2 * n), Q.resize(2 * n);
ntt(P, 0), ntt(Q, 0);
while (N >= n) {
if (N % 2 == 0) {
FOR(i, n) {
P[i] = (P[2 * i] * Q[2 * i + 1] + P[2 * i + 1] * Q[2 * i])
* inv<mint>(2);
}
} else {
FOR(i, n) {
P[i] = (P[2 * i] * Q[2 * i + 1] - P[2 * i + 1] * Q[2 * i]) * W[i];
}
}
FOR(i, n) Q[i] = Q[2 * i] * Q[2 * i + 1];
P.resize(n), Q.resize(n);
N /= 2;
if (N < n) break;
ntt_doubling(P), ntt_doubling(Q);
}
ntt(P, 1), ntt(Q, 1);
Q = fps_inv<mint>(Q);
mint ans = 0;
FOR(i, N + 1) ans += P[i] * Q[N - i];
return ans;
}
template <typename mint>
mint coef_of_rational_fps_convolution(vector<mint> P, vector<mint> Q, ll N) {
assert(0 <= len(P) && len(P) + 1 == len(Q) && Q[0] == mint(1));
if (P.empty()) return 0;
while (N >= len(P)) {
vc<mint> Q1 = Q;
FOR(i, len(Q1)) if (i & 1) Q1[i] = -Q1[i];
P = convolution(P, Q1);
Q = convolution(Q, Q1);
FOR(i, len(Q1)) Q[i] = Q[2 * i];
FOR(i, len(Q1) - 1) P[i] = P[2 * i | (N & 1)];
P.resize(len(Q1) - 1);
Q.resize(len(Q1));
N /= 2;
}
return fps_div(P, Q)[N];
}
template <typename mint>
mint coef_of_rational_fps(vector<mint> P, vector<mint> Q, ll N) {
if (P.empty()) return 0;
assert(len(Q) > 0 && Q[0] != mint(0));
while (Q.back() == mint(0)) POP(Q);
mint c = mint(1) / Q[0];
for (auto& x: P) x *= c;
for (auto& x: Q) x *= c;
mint base = 0;
if (len(P) >= len(Q)) {
auto [f, g] = poly_divmod<mint>(P, Q);
base = (N < len(f) ? f[N] : mint(0));
P = g;
}
P.resize(len(Q) - 1);
int n = len(Q);
if (mint::ntt_info().fi != -1) {
if (n <= 10) return base + coef_of_rational_fps_small(P, Q, N);
if (n > 10) return base + coef_of_rational_fps_ntt(P, Q, N);
}
mint x = (n <= 16 ? coef_of_rational_fps_small(P, Q, N)
: coef_of_rational_fps_convolution(P, Q, N));
return base + x;
}
#line 5 "/home/maspy/compro/library/seq/interpolate_linear_rec.hpp"
template <typename mint>
mint interpolate_linear_rec(vector<mint> A, ll N, int off) {
if (N < len(A)) return A[N];
A = {A.begin() + off, A.end()};
N -= off;
auto G = find_linear_rec(A);
auto F = convolution(A, G);
F.resize(len(G) - 1);
return coef_of_rational_fps(F, G, N);
}
#line 9 "main.cpp"
using mint = modint998;
/*
辺 *2
成分 *2
*/
vc<int> reverse_state(const vc<int>& now) {
int N = len(now);
vc<int> max_i(N, -1);
FOR(i, N) if (now[i] != -1) max_i[now[i]] = i;
vc<int> rev(N, -1);
FOR(i, N) {
if (now[i] == -1) continue;
int x = max_i[now[i]];
rev[N - 1 - i] = N - 1 - x;
}
return rev;
}
void solve() {
LL(N, M);
if (N % 4 != 0 || M % 4 != 0) return print(0);
N /= 4, M /= 4;
vc<tuple<int, int, mint>> edges;
vc<pair<int, mint>> init;
vvc<int> state;
HashMap<int> MP;
auto get_idx = [&](vc<int>& A) -> int {
u64 a = hash_vector(A);
u64 b = hash_vector(reverse_state(A));
u64 k = min(a, b);
if (!MP.count(k)) {
MP[k] = len(state);
state.eb(A);
}
return MP[k];
};
FOR(s, 1 << (N - 1)) {
// 行間でのマージが s
vc<int> A(N);
FOR(i, 1, N) A[i] = (s >> (i - 1) & 1 ? A[i - 1] : i);
int idx = get_idx(A);
int m = popcnt(s);
// 2(N-m)+m
init.eb(idx, mint(2).pow(2 * N - m));
}
FOR(p, len(state)) {
auto A = state[p];
FOR(s, 1 << N) {
FOR(t, 1 << (N - 1)) {
UnionFind uf(2 * N);
FOR(i, N) uf.merge(i, A[i]);
mint wt = 1;
FOR(uf.n_comp - N) wt *= inv<mint>(4);
FOR(i, N) {
if (s >> i & 1) uf.merge(i, N + i);
}
FOR(i, N - 1) {
if (t >> i & 1) uf.merge(N + i, N + i + 1);
}
FOR(uf.n_comp) wt *= mint(4);
wt *= mint(2).pow(popcnt(s) + popcnt(t));
vc<int> label(2 * N);
FOR_R(i, N) label[uf[N + i]] = i;
vc<int> B(N);
FOR(i, N) B[i] = label[uf[N + i]];
int q = get_idx(B);
edges.eb(p, q, wt);
}
}
}
int LIM = len(state) * 2 + 10;
vc<mint> ANS(LIM);
int n = len(state);
vc<mint> dp(n);
for (auto& [a, b]: init) dp[a] += b;
FOR(k, 1, LIM) {
ANS[k] = SUM<mint>(dp) * inv<mint>(2).pow(N * k);
vc<mint> newdp(n);
for (auto& [i, j, x]: edges) newdp[j] += dp[i] * x;
swap(dp, newdp);
}
mint ans = interpolate_linear_rec<mint>(ANS, M, 1);
print(ans);
}
signed main() {
solve();
return 0;
}
Details
Tip: Click on the bar to expand more detailed information
Test #1:
score: 100
Accepted
time: 1ms
memory: 4048kb
input:
4 4
output:
2
result:
ok "2"
Test #2:
score: 0
Accepted
time: 0ms
memory: 4068kb
input:
2 8
output:
0
result:
ok "0"
Test #3:
score: 0
Accepted
time: 0ms
memory: 3888kb
input:
12 3456
output:
491051233
result:
ok "491051233"
Test #4:
score: 0
Accepted
time: 0ms
memory: 3736kb
input:
1 1
output:
0
result:
ok "0"
Test #5:
score: 0
Accepted
time: 0ms
memory: 3732kb
input:
20 999999999999999983
output:
0
result:
ok "0"
Test #6:
score: 0
Accepted
time: 0ms
memory: 3784kb
input:
24 999999999999999994
output:
0
result:
ok "0"
Test #7:
score: 0
Accepted
time: 0ms
memory: 3776kb
input:
24 999999999999999955
output:
0
result:
ok "0"
Test #8:
score: 0
Accepted
time: 1962ms
memory: 28468kb
input:
28 999999999999999928
output:
846645622
result:
ok "846645622"
Test #9:
score: 0
Accepted
time: 0ms
memory: 3848kb
input:
28 999999999999999971
output:
0
result:
ok "0"
Test #10:
score: 0
Accepted
time: 0ms
memory: 3884kb
input:
28 999999999999999901
output:
0
result:
ok "0"
Test #11:
score: 0
Accepted
time: 0ms
memory: 3928kb
input:
28 999999999999999945
output:
0
result:
ok "0"
Test #12:
score: 0
Accepted
time: 0ms
memory: 3816kb
input:
30 1000000000000000000
output:
0
result:
ok "0"
Test #13:
score: 0
Accepted
time: 0ms
memory: 3916kb
input:
4 144115188075855868
output:
479168365
result:
ok "479168365"
Test #14:
score: 0
Accepted
time: 0ms
memory: 4000kb
input:
4 288230376151711740
output:
661413713
result:
ok "661413713"
Test #15:
score: 0
Accepted
time: 0ms
memory: 4064kb
input:
4 576460752303423484
output:
854857972
result:
ok "854857972"
Test #16:
score: 0
Accepted
time: 0ms
memory: 3812kb
input:
8 144115188075855868
output:
266363233
result:
ok "266363233"
Test #17:
score: 0
Accepted
time: 0ms
memory: 4012kb
input:
8 288230376151711740
output:
862901307
result:
ok "862901307"
Test #18:
score: 0
Accepted
time: 0ms
memory: 3924kb
input:
8 576460752303423484
output:
455991900
result:
ok "455991900"
Test #19:
score: 0
Accepted
time: 0ms
memory: 3928kb
input:
12 144115188075855868
output:
226857121
result:
ok "226857121"
Test #20:
score: 0
Accepted
time: 0ms
memory: 4020kb
input:
12 288230376151711740
output:
717445399
result:
ok "717445399"
Test #21:
score: 0
Accepted
time: 0ms
memory: 3932kb
input:
12 576460752303423484
output:
935277864
result:
ok "935277864"
Test #22:
score: 0
Accepted
time: 1ms
memory: 4104kb
input:
16 144115188075855868
output:
631950327
result:
ok "631950327"
Test #23:
score: 0
Accepted
time: 1ms
memory: 4028kb
input:
16 288230376151711740
output:
73767413
result:
ok "73767413"
Test #24:
score: 0
Accepted
time: 1ms
memory: 3916kb
input:
16 576460752303423484
output:
329402693
result:
ok "329402693"
Test #25:
score: 0
Accepted
time: 2ms
memory: 4044kb
input:
20 144115188075855868
output:
125405814
result:
ok "125405814"
Test #26:
score: 0
Accepted
time: 5ms
memory: 4272kb
input:
20 288230376151711740
output:
794579293
result:
ok "794579293"
Test #27:
score: 0
Accepted
time: 5ms
memory: 4004kb
input:
20 576460752303423484
output:
726571847
result:
ok "726571847"
Test #28:
score: 0
Accepted
time: 81ms
memory: 7880kb
input:
24 144115188075855868
output:
310529292
result:
ok "310529292"
Test #29:
score: 0
Accepted
time: 75ms
memory: 6496kb
input:
24 288230376151711740
output:
493789216
result:
ok "493789216"
Test #30:
score: 0
Accepted
time: 81ms
memory: 7368kb
input:
24 576460752303423484
output:
606221157
result:
ok "606221157"
Test #31:
score: 0
Accepted
time: 1994ms
memory: 28552kb
input:
28 144115188075855868
output:
964541053
result:
ok "964541053"
Test #32:
score: 0
Accepted
time: 2121ms
memory: 29780kb
input:
28 288230376151711740
output:
42971353
result:
ok "42971353"
Test #33:
score: 0
Accepted
time: 1937ms
memory: 29744kb
input:
28 576460752303423484
output:
179569141
result:
ok "179569141"
Test #34:
score: 0
Accepted
time: 0ms
memory: 4016kb
input:
6 5
output:
0
result:
ok "0"
Test #35:
score: 0
Accepted
time: 0ms
memory: 3808kb
input:
14 28
output:
0
result:
ok "0"
Test #36:
score: 0
Accepted
time: 0ms
memory: 3816kb
input:
25 6365
output:
0
result:
ok "0"
Test #37:
score: 0
Accepted
time: 0ms
memory: 3828kb
input:
18 529543996
output:
0
result:
ok "0"
Test #38:
score: 0
Accepted
time: 0ms
memory: 3840kb
input:
10 675199829716849556
output:
0
result:
ok "0"
Test #39:
score: 0
Accepted
time: 3ms
memory: 4196kb
input:
20 425279816112802872
output:
269059955
result:
ok "269059955"
Test #40:
score: 0
Accepted
time: 1ms
memory: 4104kb
input:
8 38212554426330756
output:
207344318
result:
ok "207344318"
Test #41:
score: 0
Accepted
time: 81ms
memory: 7700kb
input:
24 666881067086698696
output:
245351821
result:
ok "245351821"
Test #42:
score: 0
Accepted
time: 0ms
memory: 4156kb
input:
4 728683474913510792
output:
466882081
result:
ok "466882081"
Test #43:
score: 0
Accepted
time: 1990ms
memory: 28168kb
input:
28 201838304882525604
output:
217184228
result:
ok "217184228"
Test #44:
score: 0
Accepted
time: 0ms
memory: 3928kb
input:
4 8560
output:
596875387
result:
ok "596875387"
Test #45:
score: 0
Accepted
time: 0ms
memory: 3864kb
input:
12 60764
output:
930662011
result:
ok "930662011"
Test #46:
score: 0
Accepted
time: 0ms
memory: 3924kb
input:
8 45272
output:
239612337
result:
ok "239612337"
Test #47:
score: 0
Accepted
time: 0ms
memory: 3928kb
input:
8 84616
output:
826857610
result:
ok "826857610"
Test #48:
score: 0
Accepted
time: 0ms
memory: 3924kb
input:
4 316408
output:
529567983
result:
ok "529567983"
Test #49:
score: 0
Accepted
time: 0ms
memory: 3924kb
input:
8 12
output:
1182
result:
ok "1182"
Test #50:
score: 0
Accepted
time: 0ms
memory: 3840kb
input:
8 16
output:
16644
result:
ok "16644"
Test #51:
score: 0
Accepted
time: 0ms
memory: 3900kb
input:
4 8
output:
6
result:
ok "6"
Test #52:
score: 0
Accepted
time: 0ms
memory: 3836kb
input:
12 16
output:
5253822
result:
ok "5253822"
Extra Test:
score: 0
Extra Test Passed