#define AUTO_GENERATED
// competitive-verifier: PROBLEM https://onlinejudge.u-aizu.ac.jp/courses/library/7/DPL/all/DPL_5_J
#include "../../../src/comb/ball_box/iia/lib.hpp"
using namespace tifa_libs;
CEXP u32 MOD = 1000000000 + 7;
#include "../../../src/fps/ds/mtt/lib.hpp"
#include "../../../src/math/ds/mint/md/lib.hpp"
using namespace tifa_libs;
using mint = mint_md<__LINE__>;
using namespace tifa_libs;
using poly = polymtt<mint>;
int main() {
mint::set_mod(MOD);
std::cin.tie(nullptr)->std::ios::sync_with_stdio(false);
u32 n, k;
std::cin >> n >> k;
std::cout << tifa_libs::ball_box_iia<poly>(n, k) << '\n';
return 0;
}
#line 1 "test/cpv/aizu-dpl/dpl_5_j.mintd-md.poly_anymod-pmtt.cpp"
#define AUTO_GENERATED
// competitive-verifier: PROBLEM https://onlinejudge.u-aizu.ac.jp/courses/library/7/DPL/all/DPL_5_J
#line 2 "src/comb/ball_box/iia/lib.hpp"
#line 2 "src/comb/ball_box/gen_ii/lib.hpp"
#line 2 "src/fps/exp/lib.hpp"
#line 2 "src/fps/ln/lib.hpp"
#line 2 "src/fps/deriv/lib.hpp"
#line 2 "src/fps/ds/poly_c/lib.hpp"
#line 2 "src/util/strip/lib.hpp"
#line 2 "src/util/util/lib.hpp"
// https://github.com/Tiphereth-A/CP-lib
#include <bits/extc++.h>
// clang-format off
namespace tifa_libs {
#define CEXP constexpr
#define CEXPE constexpr explicit
#define CR const&
#define CP const*
#define PC *const
#define CPC const*const
#define TPN typename
#define NE noexcept
#define CNE const noexcept
#define ND [[nodiscard]]
#define cT_(...) std::conditional_t<sizeof(__VA_ARGS__) <= sizeof(size_t) * 2, __VA_ARGS__, __VA_ARGS__ CR>
// NOLINTNEXTLINE(misc-const-correctness)
#define flt_(T, i, l, r, ...) for (T i = (l), i##e = (r)__VA_OPT__(, ) __VA_ARGS__; i < i##e; ++i)
#define retif_(cond, if_true, ...) if cond return if_true __VA_OPT__(; else return __VA_ARGS__)
#ifdef ONLINE_JUDGE
#undef assert
#define assert(x) 42
#endif
using namespace std::ranges;
using namespace std::literals;
template <class T>
CEXP T abs(T x) NE { retif_((x < 0), -x, x); }
} // namespace tifa_libs
// clang-format on
#line 4 "src/util/strip/lib.hpp"
namespace tifa_libs {
// pred(x) == true ==> drop
template <common_range R, class F>
CEXP auto lstrip_view(R CR range, F&& pred) NE {
auto v = range | views::drop_while(std::forward<F>(pred));
return subrange{begin(v), end(v)};
}
// pred(x) == true ==> drop
template <common_range R, class F>
CEXP auto rstrip_view(R CR range, F&& pred) NE {
auto v = range | views::reverse | views::drop_while(std::forward<F>(pred));
return subrange{end(v).base(), begin(v).base()};
}
// pred(x) == true ==> drop
template <common_range R, class F>
CEXP auto strip_view(R CR range, F&& pred) NE {
auto v = range | views::drop_while(std::forward<F>(pred)) | views::reverse | views::drop_while(std::forward<F>(pred));
return subrange{end(v).base(), begin(v).base()};
}
} // namespace tifa_libs
#line 2 "src/util/traits/others/lib.hpp"
// clang-format off
#line 2 "src/util/alias/others/lib.hpp"
#line 2 "src/util/consts/lib.hpp"
#line 2 "src/util/alias/num/lib.hpp"
#line 4 "src/util/alias/num/lib.hpp"
// clang-format off
namespace tifa_libs {
#define mk0_(w, t) using w = t; using c##w = const t
#define mk_(w, t) mk0_(w, t); CEXP w operator""_##w(unsigned long long x) NE { return (w)x; }
mk_(i8, int8_t) mk_(u8, uint8_t) mk_(i16, int16_t) mk_(u16, uint16_t) mk_(i32, int32_t) mk_(u32, uint32_t) mk_(i64, int64_t) mk_(u64, uint64_t) mk_(isz, ssize_t) mk_(usz, size_t) mk_(chr, char) mk_(schr, signed char) mk_(uchr, unsigned char) mk_(sint, signed) mk_(uint, unsigned);
mk0_(i128, __int128_t); mk0_(u128, __uint128_t); mk0_(f32, float); mk0_(f64, double); mk0_(f128, long double);
#undef mk0_
#undef mk_
} // namespace tifa_libs
// clang-format on
#line 4 "src/util/consts/lib.hpp"
// clang-format off
namespace tifa_libs {
using std::numbers::pi_v;
template <std::floating_point FP>
inline FP eps_v = std::sqrt(std::numeric_limits<FP>::epsilon());
template <std::floating_point FP>
CEXP void set_eps(FP v) NE { eps_v<FP> = v; }
CEXP u32 TIME = ((__TIME__[0] & 15) << 20) | ((__TIME__[1] & 15) << 16) | ((__TIME__[3] & 15) << 12) | ((__TIME__[4] & 15) << 8) | ((__TIME__[6] & 15) << 4) | (__TIME__[7] & 15);
CEXP auto STR2U16 = [] { std::array<u32, 65536> table{}; table.fill(-1_u32); flt_ (u32, i, 48, 58) flt_ (u32, j, 48, 58) table[i << 8 | j] = (j & 15) * 10 + (i & 15); return table; }();
inline const auto fn_0 = [](auto&&...) NE {};
inline const auto fn_is0 = [](auto x) NE { return x == 0; };
} // namespace tifa_libs
// clang-format on
#line 4 "src/util/alias/others/lib.hpp"
namespace tifa_libs {
template <class T>
struct chash {
CEXP static u64 C = u64(pi_v<f128> * 2e18) | 71;
CEXP u64 operator()(T x) CNE { return __builtin_bswap64(((u64)x ^ TIME) * C); }
};
// clang-format off
#define mk_(w, t) using w = t; using c##w = const t;
mk_(strn, std::string) mk_(strnv, std::string_view)
#undef mk_
template <class T> struct edge_t { T w; u32 u, v; CEXP auto operator<=>(edge_t CR) const = default; }; template <class T> using cedge_t = const edge_t<T>;
template <class T> struct pt3 { T _0, _1, _2; CEXP auto operator<=>(pt3 CR) const = default; }; template <class T> using cpt3 = const pt3<T>;
template <class T> struct pt4 { T _0, _1, _2, _3; CEXP auto operator<=>(pt4 CR) const = default; }; template <class T> using cpt4 = const pt4<T>;
#define mkT_(w, t, ...) template <class T> using w = t __VA_OPT__(, ) __VA_ARGS__; template <class T> using c##w = const t __VA_OPT__(, ) __VA_ARGS__;
mkT_(ptt, std::pair<T, T>) mkT_(alc, std::pmr::polymorphic_allocator<T>) mkT_(vec, std::vector<T>) mkT_(vvec, vec<vec<T>>) mkT_(v3ec, vvec<vec<T>>) mkT_(vecpt, vec<ptt<T>>) mkT_(vvecpt, vvec<ptt<T>>) mkT_(ptvec, ptt<vec<T>>) mkT_(ptvvec, ptt<vvec<T>>)
#undef mkT_
template <class T> using itl = std ::initializer_list<T>;
template <class T, usz ext = std::dynamic_extent> using spn = std::span<T const, ext>;
template <class T, usz N> using arr = std::array<T, N>; template <class T, usz N> using carr = std::array<const T, N>;
template <class U, class T> using vecp = vec<std::pair<U, T>>; template <class U, class T> using vvecp = vvec<std::pair<U, T>>;
template <class U, class T> using vvecp = vvec<std::pair<U, T>>; template <class U, class T> using vvvecp = vvec<vvec<std::pair<U, T>>>;
#ifdef PB_DS_ASSOC_CNTNR_HPP
template <class T, class C = std::less<T>> using set = __gnu_pbds::tree<T, __gnu_pbds::null_type, C>;
template <class K, class V, class C = std::less<K>> using map = __gnu_pbds::tree<K, V, C>;
// hset<u64> s({}, {}, {}, {}, {1<<16});
template <class T, class HF = chash<T>> using hset = __gnu_pbds::gp_hash_table<T, __gnu_pbds::null_type, HF>;
// hmap<u64, int> s({}, {}, {}, {}, {1<<16});
template <class K, class V, class HF = chash<K>> using hmap = __gnu_pbds::gp_hash_table<K, V, HF>;
#else
using std::set, std::map;
template <class T, class HF = chash<T>> using hset = std::unordered_set<T, HF>;
template <class K, class V, class HF = chash<K>> using hmap = std::unordered_map<K, V, HF>;
#endif
#ifdef PB_DS_PRIORITY_QUEUE_HPP
template <class T, class C = std::less<T>> using pq = __gnu_pbds::priority_queue<T, C>;
#else
template <class T, class C = std::less<T>> using pq = std::priority_queue<T, vec<T>, C>;
#endif
template <class T> using pqg = pq<T, std::greater<T>>;
// clang-format on
#define mk1_(V, A, T) using V##A = V<T>;
#define mk_(V, A, T) mk1_(V, A, T) mk1_(c##V, A, T)
#define mk(A, T) mk_(edge_t, A, T) mk_(ptt, A, T) mk_(pt3, A, T) mk_(pt4, A, T) mk_(vec, A, T) mk_(vvec, A, T) mk_(v3ec, A, T) mk_(vecpt, A, T) mk_(vvecpt, A, T) mk_(ptvec, A, T) mk_(ptvvec, A, T) mk1_(spn, A, T) mk1_(itl, A, T)
mk(b, bool) mk(c, chr) mk(i, i32) mk(u, u32) mk(ii, i64) mk(uu, u64) mk(t, isz) mk(z, usz) mk(f, f32) mk(d, f64) mk(s, strn);
#undef mk
#undef mk_
#undef mk1_
} // namespace tifa_libs
#line 4 "src/util/traits/others/lib.hpp"
namespace tifa_libs {
//! only for template without non-type argument
template <class, template <class...> class> CEXP bool specialized_from_v = false;
template <template <class...> class T, class... Args> CEXP bool specialized_from_v<T<Args...>, T> = true;
static_assert(specialized_from_v<vecu, std::vector>);
template <class T> concept container_c = common_range<T> && !std::is_array_v<std::remove_cvref_t<T>> && !std::same_as<std::remove_cvref_t<T>, strn> && !std::same_as<std::remove_cvref_t<T>, strnv>;
template <class T> concept istream_c = std::derived_from<T, std::istream> || std::derived_from<T, std::wistream> || requires(T is) { is.peek(); };
template <class T> concept ostream_c = std::derived_from<T, std::ostream> || std::derived_from<T, std::wostream> || requires(T os) { os.flush(); };
} // namespace tifa_libs
// clang-format on
#line 5 "src/fps/ds/poly_c/lib.hpp"
namespace tifa_libs {
// clang-format off
enum class CCORE : u8 { FFT_R2, NTT3, NTT };
// clang-format on
namespace poly_impl_ {
template <class ccore>
requires requires(ccore cc, vec<TPN ccore::val_t> l, vec<TPN ccore::val_t> r, u32 sz) {
{ ccore::ct_cat } -> std::same_as<CCORE CR>;
cc.conv(l, r), cc.conv(l, r, sz);
}
struct poly : vec<TPN ccore::val_t> {
using ccore_t = ccore;
using val_t = ccore_t::val_t;
using data_t = vec<val_t>;
static inline ccore_t conv_core;
CEXPE poly(u32 sz = 1, cT_(val_t) val = val_t{}) NE : data_t(sz, val) {}
CEXP poly(TPN data_t::const_iterator begin, TPN data_t::const_iterator end) NE : data_t(begin, end) {}
CEXP poly(data_t CR v) NE : data_t(v) {}
CEXP poly(data_t&& v) NE : data_t(std::move(v)) {}
CEXP poly(itl<val_t> v) NE : data_t(v) {}
CEXP poly(common_range auto CR v) NE : data_t(begin(v), end(v)) {}
friend CEXP auto& operator>>(istream_c auto& is, poly& poly) NE {
for (auto& val : poly) is >> val;
return is;
}
friend CEXP auto& operator<<(ostream_c auto& os, poly CR poly) NE {
retif_((!poly.size()) [[unlikely]], os);
flt_ (u32, i, 1, (u32)poly.size()) os << poly[i - 1] << ' ';
return os << poly.back();
}
ND CEXP bool is_zero() CNE {
for (auto&& i : *this)
if (i != 0) return false;
return true;
}
CEXP val_t operator()(val_t x) CNE {
val_t ans = 0;
for (u32 i = data_t::size() - 1; ~i; --i) ans = ans * x + data_t::data()[i];
return ans;
}
template <class F>
requires requires(F f, u32 idx, val_t& val) { f(idx, val); }
CEXP void apply_range(u32 l, u32 r, F&& f) NE {
assert(l < r && r <= data_t::size());
flt_ (u32, i, l, r) f(i, data_t::data()[i]);
}
template <class F>
CEXP void apply(F&& f) NE { apply_range(0, (u32)data_t::size(), std::forward<F>(f)); }
ND CEXP poly pre(u32 sz) CNE {
if (sz <= this->size()) return {this->begin(), this->begin() + sz};
poly _ = *this;
_.resize(sz);
return _;
}
CEXP void strip() NE {
auto [_, r] = rstrip_view(*this, [](cT_(val_t) x) NE { return x.val() == 0; });
if (data_t::erase(r, this->end()); data_t::empty()) data_t::push_back(val_t(0));
}
friend poly stripped(poly CR p) NE {
poly ret(rstrip_view(p, [](cT_(val_t) x) NE { return x.val() == 0; }));
if (ret.empty()) return {0};
return ret;
}
CEXP void reverse(u32 n = 0) NE { std::ranges::reverse(data_t::begin(), data_t::begin() + (n ? n : (u32)data_t::size())); }
CEXP void conv(poly CR r, u32 ans_size = 0) NE { conv_core.conv(*this, r, ans_size); }
CEXP poly operator-() CNE {
poly ret = *this;
ret.apply([](u32, auto& v) NE { v = -v; });
return ret;
}
friend CEXP poly operator+(poly p, val_t c) NE {
p[0] += c;
return p;
}
friend CEXP poly operator+(val_t c, poly CR p) NE { return p + c; }
friend CEXP poly operator-(poly p, val_t c) NE {
p[0] -= c;
return p;
}
friend CEXP poly operator-(val_t c, poly CR p) NE { return p - c; }
CEXP poly& operator*=(val_t c) NE {
apply([&c](u32, auto& v) NE { v *= c; });
return *this;
}
friend CEXP poly operator*(poly p, val_t c) NE { return p *= c; }
friend CEXP poly operator*(val_t c, poly p) NE { return p *= c; }
CEXP poly& operator+=(poly CR r) NE {
retif_((r.empty()) [[unlikely]], *this);
data_t::resize(max(data_t::size(), r.size())), apply_range(0, (u32)r.size(), [&r](u32 i, auto& v) NE { v += r[i]; });
return *this;
}
friend CEXP poly operator+(poly l, poly CR r) NE { return l += r; }
CEXP poly& operator-=(poly CR r) NE {
retif_((r.empty()) [[unlikely]], *this);
data_t::resize(max(data_t::size(), r.size()));
apply_range(0, (u32)r.size(), [&r](u32 i, auto& v) NE { v -= r[i]; });
return *this;
}
friend CEXP poly operator-(poly l, poly CR r) NE { return l -= r; }
CEXP poly& operator*=(poly CR r) NE {
if (r.empty()) {
data_t::resize(1), *data_t::data() = 0;
return *this;
}
conv(r);
return *this;
}
friend CEXP poly operator*(poly l, poly CR r) NE { return l *= r; }
CEXP auto operator<=>(poly CR r) CNE {
auto l_ = stripped(*this), r_ = stripped(r);
if (l_.size() != r_.size()) return l_.size() <=> r_.size();
return std::lexicographical_compare_three_way(l_.rbegin(), l_.rend(), r_.rbegin(), r_.rend());
}
CEXP bool operator==(poly CR r) CNE { return std::is_eq(*this <=> r); }
};
} // namespace poly_impl_
template <class T>
concept poly_c = std::same_as<T, poly_impl_::poly<TPN T::ccore_t>>;
} // namespace tifa_libs
#line 4 "src/fps/deriv/lib.hpp"
namespace tifa_libs {
CEXP auto deriv_fps(poly_c auto CR p) NE {
auto _ = p;
flt_ (u32, i, 1, (u32)_.size()) _[i - 1] = _[i] * i;
_.back() = 0;
return _;
}
} // namespace tifa_libs
#line 2 "src/fps/int/lib.hpp"
#line 4 "src/fps/int/lib.hpp"
namespace tifa_libs {
template <poly_c poly_t>
CEXP auto int_fps(poly_t CR p) NE {
auto _ = p;
for (u32 i = (u32)_.size() - 1; i; --i) _[i] = _[i - 1] * ((TPN poly_t::val_t)i).inv();
_[0] = 0;
return _;
}
} // namespace tifa_libs
#line 2 "src/fps/inv/lib.hpp"
#line 4 "src/fps/inv/lib.hpp"
namespace tifa_libs {
template <poly_c poly_t>
CEXP auto inv_fps(poly_t CR p, u32 n = 0) NE {
if (assert(p[0] != 0); !n) n = (u32)p.size();
poly_t a{p[0].inv()};
for (u32 i = 1; i < n; i *= 2) a = (a * 2 - (a * a * p).pre(i * 2)).pre(i * 2);
return a.pre(n);
}
} // namespace tifa_libs
#line 6 "src/fps/ln/lib.hpp"
namespace tifa_libs {
CEXP auto ln_fps(poly_c auto p, u32 n = 0) NE {
if (assert(p[0] == 1); !n) n = (u32)p.size();
auto _ = deriv_fps(p).pre(n);
_.conv(inv_fps(p, n));
return int_fps(_).pre(n);
}
} // namespace tifa_libs
#line 4 "src/fps/exp/lib.hpp"
namespace tifa_libs {
template <poly_c poly_t>
CEXP auto exp_fps(poly_t p, u32 n = 0) NE {
if (assert(p[0] == 0); !n) n = (u32)p.size();
p[0] += 1;
poly_t a{1};
for (u32 i = 1; i < n; i *= 2) a = (a * (p.pre(i * 2) - ln_fps(a, i * 2))).pre(i * 2);
return a.pre(n);
}
} // namespace tifa_libs
#line 2 "src/comb/seq/inv/lib.hpp"
#line 2 "src/math/mul_mod/lib.hpp"
#line 2 "src/math/safe_mod/lib.hpp"
#line 2 "src/util/traits/math/lib.hpp"
// clang-format off
#line 4 "src/util/traits/math/lib.hpp"
namespace tifa_libs {
template <class T> concept char_c = std::same_as<T, char> || std::same_as<T, signed char> || std::same_as<T, unsigned char>;
#pragma GCC diagnostic ignored "-Wpedantic"
template <class T> concept s128_c = std::same_as<T, __int128_t> || std::same_as<T, __int128>;
template <class T> concept u128_c = std::same_as<T, __uint128_t> || std::same_as<T, unsigned __int128>;
template <class T> concept i128_c = s128_c<T> || u128_c<T>;
#pragma GCC diagnostic warning "-Wpedantic"
template <class T> concept imost64_c = std::integral<T> && sizeof(T) * __CHAR_BIT__ <= 64;
template <class T> concept smost64_c = imost64_c<T> && std::signed_integral<T>;
template <class T> concept umost64_c = imost64_c<T> && std::unsigned_integral<T>;
template <class T> concept int_c = i128_c<T> || imost64_c<T>;
template <class T> concept sint_c = s128_c<T> || smost64_c<T>;
template <class T> concept uint_c = u128_c<T> || umost64_c<T>;
template <class T> concept arithm_c = std::is_arithmetic_v<T> || int_c<T>;
template <class T> concept mint_c = requires(T x) { {x.mod()} -> uint_c; {x.val()} -> uint_c; };
template <class T> concept dft_c = requires(T x, std::vector<TPN T::data_t> v, u32 n) { {x.size()} -> std::same_as<u32>; x.bzr(n); x.dif(v, n); x.dit(v, n); };
template <class T> concept ntt_c = dft_c<T> && requires(T x) { T::max_size; T::G; };
template <class T> struct to_sint : std::make_signed<T> {};
template <> struct to_sint<u128> { using type = i128; };
template <> struct to_sint<i128> { using type = i128; };
template <class T> using to_sint_t = TPN to_sint<T>::type;
template <class T> struct to_uint : std::make_unsigned<T> {};
template <> struct to_uint<u128> { using type = u128; };
template <> struct to_uint<i128> { using type = u128; };
template <class T> using to_uint_t = TPN to_uint<T>::type;
template <arithm_c T> struct to_bigger : std::make_unsigned<T> {};
#define _(w,ww) template <> struct to_bigger<w> { using type = ww; }
#define _2(w,ww) _(i##w,i##ww); _(u##w,u##ww);
_2(8, 16); _2(16, 32); _2(32, 64); _2(64, 128); _(f32, f64); _(f64, f128);
#undef _2
#undef _
template <class T> using to_bigger_t = TPN to_bigger<T>::type;
template <arithm_c T> CEXP T inf_v = [] {
if CEXP(sint_c<T>) return T(to_uint_t<T>(-1) / 4 - 1);
else if CEXP(uint_c<T>) return T(-1) / 2 - 1;
else return std::numeric_limits<T>::max() / 2 - 1;
}();
} // namespace tifa_libs
// clang-format on
#line 4 "src/math/safe_mod/lib.hpp"
namespace tifa_libs {
template <int_c T>
CEXP T safe_mod(T x, to_uint_t<T> mod) NE {
if CEXP (sint_c<T>) {
if (x <= -(T)mod || x >= (T)mod) x %= (T)mod;
retif_((x < 0), x + (T)mod, x);
} else {
retif_((x >= mod), x % mod, x);
}
}
} // namespace tifa_libs
#line 4 "src/math/mul_mod/lib.hpp"
namespace tifa_libs {
CEXP i64 mul_mod_s(i64 a, i64 b, u64 mod) NE {
if (std::bit_width((u64)abs(a)) + std::bit_width((u64)abs(b)) < 64) return safe_mod(a * b % (i64)mod, mod);
return safe_mod((i64)((i128)a * b % mod), mod);
}
CEXP u64 mul_mod_u(u64 a, u64 b, u64 mod) NE {
if (std::bit_width(a) + std::bit_width(b) <= 64) return a * b % mod;
return (u64)((u128)a * b % mod);
}
} // namespace tifa_libs
#line 5 "src/comb/seq/inv/lib.hpp"
namespace tifa_libs {
// i^{-1} from i=0..n-1
CEXP vecuu gen_inv(u32 n, u64 mod) NE {
retif_((n <= 1) [[unlikely]], vecuu(n, 1));
vecuu ans(n);
ans[0] = ans[1] = 1;
flt_ (u32, i, 2, n) ans[i] = mul_mod_u(mod - mod / i, ans[mod % i], mod);
return ans;
}
// i^{-1} from i=0..n-1
template <class mint>
CEXP vec<mint> gen_inv(u32 n) NE {
vec<mint> ans(n);
auto _ = gen_inv(n, mint::mod());
flt_ (u32, i, 0, n) ans[i] = _[i];
return ans;
}
} // namespace tifa_libs
#line 5 "src/comb/ball_box/gen_ii/lib.hpp"
namespace tifa_libs {
// f = \\prod_{i=1}^m 1/(1-x^i), deg(f) = n
template <class poly, class T>
CEXP poly gen_ball_box_ii(u32 m, u32 n, vec<T> CR inv) NE {
poly f(n + 1);
flt_ (u32, i, 1, m + 1)
flt_ (u32, k, 1, n / i + 1) f[i * k] += inv[k];
return exp_fps(f);
}
// f = \\prod_{i=1}^m 1/(1-x^i), deg(f) = n
template <class poly>
CEXP poly gen_ball_box_ii(u32 m, u32 n) NE { return gen_ball_box_ii<poly>(m, n, gen_inv(n + 1, poly::val_t::mod())); }
} // namespace tifa_libs
#line 4 "src/comb/ball_box/iia/lib.hpp"
namespace tifa_libs {
template <class poly>
CEXP auto ball_box_iia(u32 ball, u32, poly CR bbiif) NE { return bbiif[ball]; }
template <class poly>
CEXP auto ball_box_iia(u32 ball, u32 box) NE { return ball_box_iia<poly>(ball, box, gen_ball_box_ii<poly>(box, ball)); }
} // namespace tifa_libs
#line 4 "test/cpv/aizu-dpl/dpl_5_j.mintd-md.poly_anymod-pmtt.cpp"
using namespace tifa_libs;
CEXP u32 MOD = 1000000000 + 7;
#line 2 "src/fps/ds/mtt/lib.hpp"
#line 2 "src/conv/add/mtt/lib.hpp"
#line 2 "src/conv/trans/fft_r2/lib.hpp"
#line 4 "src/conv/trans/fft_r2/lib.hpp"
namespace tifa_libs {
template <std::floating_point FP>
class fft_r2 {
using C = std::complex<FP>;
const FP TAU = std::acos((FP)-1.) * 2;
vecu rev;
vec<C> w;
public:
using data_t = C;
ND CEXP u32 size() CNE { return (u32)rev.size(); }
CEXP void bzr(u32 len) NE {
cu32 n = max(std::bit_ceil(len), 2_u32);
if (n == size()) return;
rev.resize(n, 0);
cu32 k = (u32)(std::bit_width(n) - 1);
flt_ (u32, i, 0, n) rev[i] = (rev[i / 2] / 2) | ((i & 1) << (k - 1));
w.resize(n), w[0].real(1);
flt_ (u32, i, 1, n) w[i] = {std::cos(TAU * (FP)i / (FP)n), std::sin(TAU * (FP)i / (FP)n)};
}
CEXP void dif(vec<data_t>& f, u32 n = 0) CNE {
if (!n) n = size();
if (f.size() < n) f.resize(n);
assert(n <= size());
flt_ (u32, i, 0, n)
if (i < rev[i]) swap(f[rev[i]], f[i]);
#pragma GCC diagnostic ignored "-Wsign-conversion"
for (u32 i = 2, d = n / 2; i <= n; i *= 2, d /= 2)
for (u32 j = 0; j < n; j += i) {
auto l = begin(f) + j, r = begin(f) + j + i / 2;
auto p = begin(w);
for (u32 k = 0; k < i / 2; ++k, ++l, ++r, p += d) {
const data_t _ = *r * *p;
*r = *l - _, *l = *l + _;
}
}
#pragma GCC diagnostic warning "-Wsign-conversion"
}
CEXP void dit(vec<data_t>& f, u32 n = 0) CNE {
if (!n) n = size();
dif(f, n);
flt_ (u32, i, 0, n) f[i] /= (FP)n;
}
};
} // namespace tifa_libs
#line 2 "src/conv/add/naive/lib.hpp"
#line 4 "src/conv/add/naive/lib.hpp"
namespace tifa_libs {
CEXP inline u32 CONV_NAIVE_THRESHOLD = 16;
template <class U, class T = U>
requires(sizeof(U) <= sizeof(T))
CEXP vec<T> conv_naive(vec<U> CR l, vec<U> CR r, u32 ans_size = 0) NE {
retif_((l.empty() || r.empty()) [[unlikely]], {});
if (!ans_size) ans_size = u32(l.size() + r.size() - 1);
vec<T> ans(ans_size);
u32 n = (u32)l.size(), m = (u32)r.size();
auto &&l_ = n < m ? r : l, &&r_ = n < m ? l : r;
if (n < m) swap(n, m);
flt_ (u32, i, 0, n)
flt_ (u32, j, 0, min(m, ans_size - i)) ans[i + j] += (T)l_[i] * (T)r_[j];
return ans;
}
} // namespace tifa_libs
#line 5 "src/conv/add/mtt/lib.hpp"
namespace tifa_libs {
template <class mint, class FP>
CEXP vec<mint> conv_mtt(fft_r2<FP>& fft, vec<mint> CR l, vec<mint> CR r, u32 ans_size = 0) NE {
using C = TPN fft_r2<FP>::data_t;
if (!ans_size) ans_size = u32(l.size() + r.size() - 1);
if (min(l.size(), r.size()) < CONV_NAIVE_THRESHOLD) return conv_naive(l, r, ans_size);
if (l.size() == 1) {
vec<mint> ans = r;
for (ans.resize(ans_size); auto& i : ans) i *= l[0];
return ans;
}
if (r.size() == 1) {
vec<mint> ans = l;
for (ans.resize(ans_size); auto& i : ans) i *= r[0];
return ans;
}
fft.bzr(max({(u32)l.size(), (u32)r.size(), min(u32(l.size() + r.size() - 1), ans_size)}));
cu32 n = fft.size();
csint OFS = ((int)sizeof(decltype(mint::mod())) * 8 - std::countl_zero(mint::mod() - 1) + 1) / 2;
cu32 MSK = ((1u << OFS) - 1);
vec<mint> ans(ans_size);
vec<C> a(n), b(n);
flt_ (u32, i, 0, (u32)l.size()) a[i] = {(FP)(l[i].val() & MSK), (FP)(l[i].val() >> OFS)};
flt_ (u32, i, 0, (u32)r.size()) b[i] = {(FP)(r[i].val() & MSK), (FP)(r[i].val() >> OFS)};
fft.dif(a), fft.dif(b);
{
vec<C> p(n), q(n);
for (u32 i = 0, j; i < n; ++i) {
j = (n - i) & (n - 1);
C da = (a[i] + std::conj(a[j])) * C(.5, 0), db = (a[i] - std::conj(a[j])) * C(0, -.5),
dc = (b[i] + std::conj(b[j])) * C(.5, 0), dd = (b[i] - std::conj(b[j])) * C(.5, 0);
p[j] = da * dc + da * dd;
q[j] = db * dc + db * dd;
}
a = p, b = q;
}
fft.dif(a), fft.dif(b);
flt_ (u32, i, 0, ans_size) {
ci64 da = (i64)(a[i].real() / (FP)n + .5) % mint::smod(),
db = (i64)(a[i].imag() / (FP)n + .5) % mint::smod(),
dc = (i64)(b[i].real() / (FP)n + .5) % mint::smod(),
dd = (i64)(b[i].imag() / (FP)n + .5) % mint::smod();
ans[i] = da + ((db + dc) << OFS) % mint::smod() + (dd << (OFS * 2)) % mint::smod();
}
return ans;
}
} // namespace tifa_libs
#line 5 "src/fps/ds/mtt/lib.hpp"
namespace tifa_libs {
namespace polymtt_impl_ {
template <class mint, class FP>
struct cconv_mtt : public fft_r2<FP> {
using val_t = mint;
static CEXP auto ct_cat = CCORE::FFT_R2;
CEXP void conv(vec<val_t>& l, vec<val_t> CR r, u32 sz = 0) NE { l = conv_mtt(*this, l, r, sz); }
};
} // namespace polymtt_impl_
template <class mint, class FP = f64>
using polymtt = poly_impl_::poly<polymtt_impl_::cconv_mtt<mint, FP>>;
} // namespace tifa_libs
#line 2 "src/math/ds/mint/md/lib.hpp"
#line 2 "src/nt/mod/montgomery/lib.hpp"
#line 4 "src/nt/mod/montgomery/lib.hpp"
namespace tifa_libs {
template <u32 MOD>
struct montgomery {
static CEXP u32 MOD2 = MOD << 1, R2 = -(u64)(MOD) % MOD, R = [] {
u32 iv = MOD * (2 - MOD * MOD);
iv *= 2 - MOD * iv, iv *= 2 - MOD * iv;
return iv * (MOD * iv - 2);
}();
static_assert(MOD & 1);
static_assert(-R * MOD == 1);
static_assert((MOD >> 30) == 0);
static_assert(MOD != 1);
static CEXP u32 reduce(u64 x) NE { return u32((x + u64((u32)x * R) * MOD) >> 32); }
static CEXP u32 norm(u32 x) NE { return x - (MOD & -((MOD - 1 - x) >> 31)); }
};
template <> // dynamic
struct montgomery<0> {
u32 R, R2, MOD, MOD_ODD, OFFSET, MASK;
CEXP montgomery() NE = default;
CEXPE montgomery(u32 m) NE { reset(m); }
CEXP void reset(u32 m) NE {
for (assert(!(m == 1 || m >> 31)), MOD = MOD_ODD = m, OFFSET = 0; (MOD_ODD & 1) == 0; ++OFFSET, MOD_ODD /= 2);
MASK = (1_u32 << OFFSET) - 1_u32;
u32 iv = MOD_ODD * (2 - MOD_ODD * MOD_ODD);
iv *= 2 - MOD_ODD * iv, iv *= 2 - MOD_ODD * iv, R = iv * (MOD_ODD * iv - 2), R2 = u32(-u64(MOD_ODD) % MOD_ODD);
}
ND CEXP u32 norm(i32 x) CNE { return u32(x + (-(x < 0) & (i32)MOD)); }
ND CEXP u32 reduce(u64 x) CNE {
cu32 t = u32((x + u64((u32)x * R) * MOD_ODD) >> 32);
return t - (MOD_ODD & -((MOD_ODD - 1 - t) >> 31));
}
ND CEXP u32 tsf(u32 x) CNE { retif_((!OFFSET) [[likely]], reduce(u64(x) * R2), reduce(u64(x % MOD_ODD) * R2) << OFFSET | (x & MASK)); }
};
} // namespace tifa_libs
#line 2 "src/math/ds/mint/_base/lib.hpp"
#line 2 "src/nt/inverse/lib.hpp"
#line 2 "src/nt/gl/inv_gcd/lib.hpp"
#line 2 "src/nt/gl/exgcd/lib.hpp"
#line 4 "src/nt/gl/exgcd/lib.hpp"
namespace tifa_libs {
// Binary exgcd
template <uint_c U, bool only_x = false>
CEXP auto exgcd_b(U a, U b) NE {
using T = to_sint_t<U>;
if CEXP (only_x) {
if (!a) return std::make_tuple(b, (T)0);
if (!b) return std::make_tuple(a, (T)1);
} else {
if (!a) return std::make_tuple(b, (T)0, (T) !!b);
if (!b) return std::make_tuple(a, (T)1, (T)0);
}
auto r = std::__countr_zero(a | b);
a >>= r, b >>= r;
T x = (T)a, y = (T)b, s = 1, t = 0, u = 0, v = 1;
while (x) {
while (!(x & 1))
if (x /= 2; !((s | t) & 1)) s /= 2, t /= 2;
else s = (s + (T)b) / 2, t = (t - (T)a) / 2;
while (!(y & 1))
if (y /= 2; !((u | v) & 1)) u /= 2, v /= 2;
else u = (u + (T)b) / 2, v = (v - (T)a) / 2;
if (x >= y) x -= y, s -= u, t -= v;
else y -= x, u -= s, v -= t;
}
if (y > 1) a /= (U)y, b /= (U)y;
if (a && (U)abs(v) >= a) {
const T _ = v / (T)a;
v -= _ * (T)a, u += _ * (T)b;
}
if (b && (U)abs(u) >= b) {
const T _ = u / (T)b;
u -= _ * (T)b, v += _ * (T)a;
}
if (const T u_ = u + (T)b, v_ = v - (T)a; abs(u_) + abs(v_) <= abs(u) + abs(v)) u = u_, v = v_;
if (const T u_ = u - (T)b, v_ = v + (T)a; abs(u_) + abs(v_) <= abs(u) + abs(v)) u = u_, v = v_;
if CEXP (only_x) return std::make_tuple(U(y << r), u);
else return std::make_tuple(U(y << r), u, v);
}
// @return then return tuple(g, x[, y]) s.t. g = gcd(a, b), xa + yb = g, |x| + |y| is the minimal (primary) and x <= y (secondarily)
template <sint_c T, bool only_x = false>
CEXP auto exgcd(T a, T b) NE {
using U = to_uint_t<T>;
if (auto [x, y] = minmax(a, b); x >= 0 && y <= T(U(-1) >> sizeof(U))) return exgcd_b<U, only_x>((U)a, (U)b);
if CEXP (only_x) {
T s = 1, u = 0;
while (b) {
T c = a / b;
std::tie(s, u, a, b) = std::make_tuple(u, s - u * c, b, a - b * c);
}
return std::make_tuple((U)a, s);
} else {
T s = 1, t = 0, u = 0, v = 1;
while (b) {
T c = a / b;
std::tie(s, t, u, v, a, b) = std::make_tuple(u, v, s - u * c, t - v * c, b, a - b * c);
}
return std::make_tuple((U)a, s, t);
}
}
} // namespace tifa_libs
#line 6 "src/nt/gl/inv_gcd/lib.hpp"
namespace tifa_libs {
template <uint_c T>
CEXP ptt<T> inv_gcd(T n, T mod) NE {
using U = to_sint_t<T>;
auto [g, x] = exgcd<U, true>(U(n % mod), (U)mod);
return {g, safe_mod(x, mod)};
}
} // namespace tifa_libs
#line 4 "src/nt/inverse/lib.hpp"
namespace tifa_libs {
// simple but slower: inv(n, mod) -> 1 < n ? mod - inv(mod % n, n) * mod / n : 1;
template <uint_c T, uint_c U>
CEXP U inverse(T n, U mod) NE {
auto [g, x] = inv_gcd(U(n % mod), mod);
assert(g == 1);
return x;
}
} // namespace tifa_libs
#line 5 "src/math/ds/mint/_base/lib.hpp"
namespace tifa_libs::mint_impl_ {
struct mint_tag_base {};
template <std::derived_from<mint_tag_base> tag_t>
struct mint : tag_t {
CEXP mint() = default;
CEXP mint(int_c auto v) NE : tag_t(v) {}
using raw_t = tag_t::raw_t;
using sraw_t = to_sint_t<raw_t>;
static CEXP sraw_t smod() NE { return (sraw_t)tag_t::mod(); }
ND CEXP sraw_t sval() CNE { return (sraw_t)tag_t::val(); }
template <int_c T>
CEXPE operator T() CNE { return (T)tag_t::val(); }
CEXP mint& operator+=(mint CR r) NE {
mint::add(r);
return *this;
}
CEXP mint& operator-=(mint CR r) NE {
mint::sub(r);
return *this;
}
CEXP mint& operator*=(mint CR r) NE {
mint::mul(r);
return *this;
}
CEXP mint& operator/=(mint CR r) NE { return *this = *this * r.inv(); }
CEXP mint CR operator+() CNE { return *this; }
CEXP mint operator-() CNE { return tag_t::template neg<mint>(); }
ND CEXP mint inv() CNE { return inverse(tag_t::val(), tag_t::mod()); }
friend CEXP mint operator+(mint l, mint CR r) NE { return l += r; }
friend CEXP mint operator-(mint l, mint CR r) NE { return l -= r; }
friend CEXP mint operator*(mint l, mint CR r) NE { return l *= r; }
friend CEXP mint operator/(mint l, mint CR r) NE { return l /= r; }
friend CEXP bool operator==(mint CR l, mint CR r) NE { return l.val() == r.val(); }
friend CEXP auto operator<=>(mint CR l, mint CR r) NE { return l.sval() <=> r.sval(); }
friend auto& operator>>(istream_c auto& is, mint& x) NE {
i64 _;
is >> _, x = mint(_);
return is;
}
friend auto& operator<<(ostream_c auto& os, mint CR x) NE { return os << x.val(); }
friend CEXP auto abs(mint CR x) NE { return x.val(); }
};
} // namespace tifa_libs::mint_impl_
#line 5 "src/math/ds/mint/md/lib.hpp"
namespace tifa_libs {
template <i64 ID>
class mint_md_tag : public mint_impl_::mint_tag_base {
static inline montgomery<0> core;
public:
static CEXP bool FIXED_MOD = false;
static CEXP void set_mod(u32 m) NE { core.reset(m); }
protected:
using raw_t = u32;
raw_t v_{};
CEXP mint_md_tag() NE = default;
CEXP mint_md_tag(int_c auto v) NE : v_{mod(v)} {}
public:
static CEXP raw_t mod(sint_c auto v) NE { retif_((v >= 0) [[likely]], mod((to_uint_t<decltype(v)>)v), core.tsf(core.norm(i32(v % (i32)mod())))); }
static CEXP raw_t mod(uint_c auto v) NE {
if CEXP (umost64_c<decltype(v)>) {
retif_((cu64 x = (u64)v; x < mod()) [[likely]], core.tsf((raw_t)x), core.tsf((raw_t)(x % mod())));
} else retif_((v < mod()) [[likely]], core.tsf((raw_t)v), core.tsf((raw_t)(v % mod())));
}
static CEXP raw_t mod() NE { return core.MOD; }
ND CEXP raw_t val() CNE {
retif_((!core.OFFSET) [[likely]], core.reduce(v_));
const raw_t h = core.reduce(v_ >> core.OFFSET);
return ((h - v_) * core.R & core.MASK) * core.MOD_ODD + h;
}
CEXP raw_t& data() NE { return v_; }
protected:
template <class mint>
ND CEXP auto neg() CNE {
mint res;
if (!core.OFFSET) [[likely]] {
res.v_ = (core.MOD_ODD & -raw_t(v_ != 0)) - v_;
return res;
}
const raw_t h = v_ >> core.OFFSET;
res.v_ = (((core.MOD_ODD & -raw_t(h != 0)) - h) << core.OFFSET) | (-v_ & core.MASK);
return res;
}
CEXP void add(mint_md_tag CR r) NE {
if (!core.OFFSET) [[likely]] {
const raw_t h = v_ + r.v_ - core.MOD_ODD;
v_ = h + (core.MOD_ODD & -(h >> 31));
return;
}
const raw_t h = (v_ >> core.OFFSET) + (r.v_ >> core.OFFSET) - core.MOD_ODD;
v_ = ((h + (core.MOD_ODD & -(h >> 31))) << core.OFFSET) | ((v_ + r.v_) & core.MASK);
}
CEXP void sub(mint_md_tag CR r) NE {
if (!core.OFFSET) [[likely]] {
const raw_t h = v_ - r.v_;
v_ = h + (core.MOD_ODD & -(h >> 31));
return;
}
const raw_t h = (v_ >> core.OFFSET) - (r.v_ >> core.OFFSET);
v_ = ((h + (core.MOD_ODD & -(h >> 31))) << core.OFFSET) | ((v_ - r.v_) & core.MASK);
}
CEXP void mul(mint_md_tag CR r) NE {
if (!core.OFFSET) [[likely]] {
v_ = core.reduce((u64)v_ * r.v_);
return;
}
v_ = (core.reduce((u64)(v_ >> core.OFFSET) * (r.v_ >> core.OFFSET)) << core.OFFSET) | ((v_ * r.v_) & core.MASK);
}
};
template <i64 ID>
using mint_md = mint_impl_::mint<mint_md_tag<ID>>;
} // namespace tifa_libs
#line 10 "test/cpv/aizu-dpl/dpl_5_j.mintd-md.poly_anymod-pmtt.cpp"
using namespace tifa_libs;
using mint = mint_md<__LINE__>;
using namespace tifa_libs;
using poly = polymtt<mint>;
int main() {
mint::set_mod(MOD);
std::cin.tie(nullptr)->std::ios::sync_with_stdio(false);
u32 n, k;
std::cin >> n >> k;
std::cout << tifa_libs::ball_box_iia<poly>(n, k) << '\n';
return 0;
}
test/cpv/aizu-dpl/dpl_5_j.mintd-md.poly_anymod-pmtt.cpp