結果
| 問題 | No.1574 Swap and Repaint |
| ユーザー |
 NyaanNyaan
|
| 提出日時 | 2021-07-04 01:11:42 |
| 言語 | C++17 (gcc 12.3.0 + boost 1.83.0) |
| 結果 |
WA
|
| 実行時間 | - |
| コード長 | 22,536 bytes |
| コンパイル時間 | 3,685 ms |
| コンパイル使用メモリ | 281,740 KB |
| 実行使用メモリ | 7,368 KB |
| 最終ジャッジ日時 | 2024-04-14 13:59:08 |
| 合計ジャッジ時間 | 72,897 ms |
|
ジャッジサーバーID （参考情報） |
judge3 / judge1 |
（要ログイン）
テストケース
テストケース表示| 入力 | 結果 | 実行時間 実行使用メモリ |
|---|---|---|
| testcase_00 | WA | - |
| testcase_01 | WA | - |
| testcase_02 | WA | - |
| testcase_03 | WA | - |
| testcase_04 | WA | - |
| testcase_05 | WA | - |
| testcase_06 | WA | - |
| testcase_07 | WA | - |
| testcase_08 | WA | - |
| testcase_09 | WA | - |
| testcase_10 | WA | - |
| testcase_11 | WA | - |
| testcase_12 | WA | - |
| testcase_13 | WA | - |
| testcase_14 | WA | - |
| testcase_15 | WA | - |
| testcase_16 | WA | - |
| testcase_17 | WA | - |
| testcase_18 | WA | - |
| testcase_19 | WA | - |
| testcase_20 | WA | - |
| testcase_21 | WA | - |
| testcase_22 | WA | - |
| testcase_23 | WA | - |
| testcase_24 | WA | - |
| testcase_25 | WA | - |
| testcase_26 | WA | - |
| testcase_27 | WA | - |
| testcase_28 | WA | - |
| testcase_29 | WA | - |
| testcase_30 | WA | - |
| testcase_31 | WA | - |
| testcase_32 | WA | - |
| testcase_33 | WA | - |
| testcase_34 | WA | - |
| testcase_35 | WA | - |
ソースコード
// O(N^2)です。ゆるして
/**
* date : 2021-07-04 01:10:55
*/
#define NDEBUG
using namespace std;
// intrinstic
#include <immintrin.h>
#include <algorithm>
#include <array>
#include <bitset>
#include <cassert>
#include <cctype>
#include <cfenv>
#include <cfloat>
#include <chrono>
#include <cinttypes>
#include <climits>
#include <cmath>
#include <complex>
#include <cstdarg>
#include <cstddef>
#include <cstdint>
#include <cstdio>
#include <cstdlib>
#include <cstring>
#include <deque>
#include <fstream>
#include <functional>
#include <initializer_list>
#include <iomanip>
#include <ios>
#include <iostream>
#include <istream>
#include <iterator>
#include <limits>
#include <list>
#include <map>
#include <memory>
#include <new>
#include <numeric>
#include <ostream>
#include <queue>
#include <random>
#include <set>
#include <sstream>
#include <stack>
#include <streambuf>
#include <string>
#include <tuple>
#include <type_traits>
#include <typeinfo>
#include <unordered_map>
#include <unordered_set>
#include <utility>
#include <vector>
// 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 <typename T>
using V = vector<T>;
template <typename T>
using VV = vector<vector<T>>;
using vi = vector<int>;
using vl = vector<long long>;
using vd = V<double>;
using vs = V<string>;
using vvi = vector<vector<int>>;
using vvl = vector<vector<long long>>;
template <typename T, typename U>
struct P : pair<T, U> {
template <typename... Args>
P(Args... args) : pair<T, U>(args...) {}
using pair<T, U>::first;
using pair<T, U>::second;
T &x() { return first; }
const T &x() const { return first; }
U &y() { return second; }
const U &y() const { return 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;
}
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; }
};
using pl = P<ll, ll>;
using pi = P<int, int>;
using vp = V<pl>;
constexpr int inf = 1001001001;
constexpr long long infLL = 4004004004004004004LL;
template <typename T>
int sz(const T &t) {
return t.size();
}
template <typename T, typename U>
inline bool amin(T &x, U y) {
return (y < x) ? (x = y, true) : false;
}
template <typename T, typename U>
inline bool amax(T &x, U y) {
return (x < y) ? (x = y, true) : false;
}
template <typename T>
inline T Max(const vector<T> &v) {
return *max_element(begin(v), end(v));
}
template <typename T>
inline T Min(const vector<T> &v) {
return *min_element(begin(v), end(v));
}
template <typename T>
inline long long Sum(const vector<T> &v) {
return accumulate(begin(v), end(v), 0LL);
}
template <typename T>
int lb(const vector<T> &v, const T &a) {
return lower_bound(begin(v), end(v), a) - begin(v);
}
template <typename T>
int ub(const vector<T> &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 <typename T, typename U>
pair<T, U> mkp(const T &t, const U &u) {
return make_pair(t, u);
}
template <typename T>
vector<T> mkrui(const vector<T> &v, bool rev = false) {
vector<T> 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 <typename T>
vector<T> mkuni(const vector<T> &v) {
vector<T> ret(v);
sort(ret.begin(), ret.end());
ret.erase(unique(ret.begin(), ret.end()), ret.end());
return ret;
}
template <typename F>
vector<int> mkord(int N, F f) {
vector<int> ord(N);
iota(begin(ord), end(ord), 0);
sort(begin(ord), end(ord), f);
return ord;
}
template <typename T>
vector<int> mkinv(vector<T> &v) {
int max_val = *max_element(begin(v), end(v));
vector<int> inv(max_val + 1, -1);
for (int i = 0; i < (int)v.size(); i++) inv[v[i]] = i;
return inv;
}
} // 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 <typename T>
inline int gbit(const T &a, int i) {
return (a >> i) & 1;
}
template <typename T>
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 <typename T, typename U>
ostream &operator<<(ostream &os, const pair<T, U> &p) {
os << p.first << " " << p.second;
return os;
}
template <typename T, typename U>
istream &operator>>(istream &is, pair<T, U> &p) {
is >> p.first >> p.second;
return is;
}
template <typename T>
ostream &operator<<(ostream &os, const vector<T> &v) {
int s = (int)v.size();
for (int i = 0; i < s; i++) os << (i ? " " : "") << v[i];
return os;
}
template <typename T>
istream &operator>>(istream &is, vector<T> &v) {
for (auto &x : v) is >> x;
return is;
}
void in() {}
template <typename T, class... U>
void in(T &t, U &... u) {
cin >> t;
in(u...);
}
void out() { cout << "\n"; }
template <typename T, class... U, char sep = ' '>
void out(const T &t, const U &... u) {
cout << t;
if (sizeof...(u)) cout << sep;
out(u...);
}
void outr() {}
template <typename T, class... U, char sep = ' '>
void outr(const T &t, const U &... u) {
cout << t;
outr(u...);
}
struct IoSetupNya {
IoSetupNya() {
cin.tie(nullptr);
ios::sync_with_stdio(false);
cout << fixed << setprecision(15);
cerr << fixed << setprecision(7);
}
} iosetupnya;
} // namespace Nyaan
// debug
namespace DebugImpl {
template <typename U, typename = void>
struct is_specialize : false_type {};
template <typename U>
struct is_specialize<
U, typename conditional<false, typename U::iterator, void>::type>
: true_type {};
template <typename U>
struct is_specialize<
U, typename conditional<false, decltype(U::first), void>::type>
: true_type {};
template <typename U>
struct is_specialize<U, enable_if_t<is_integral<U>::value, void>> : true_type {
};
void dump(const char& t) { cerr << t; }
void dump(const string& t) { cerr << t; }
void dump(const bool& t) { cerr << (t ? "true" : "false"); }
template <typename U,
enable_if_t<!is_specialize<U>::value, nullptr_t> = nullptr>
void dump(const U& t) {
cerr << t;
}
template <typename T>
void dump(const T& t, enable_if_t<is_integral<T>::value>* = nullptr) {
string res;
if (t == Nyaan::inf) res = "inf";
if constexpr (is_signed<T>::value) {
if (t == -Nyaan::inf) res = "-inf";
}
if constexpr (sizeof(T) == 8) {
if (t == Nyaan::infLL) res = "inf";
if constexpr (is_signed<T>::value) {
if (t == -Nyaan::infLL) res = "-inf";
}
}
if (res.empty()) res = to_string(t);
cerr << res;
}
template <typename T, typename U>
void dump(const pair<T, U>&);
template <typename T>
void dump(const pair<T*, int>&);
template <typename T>
void dump(const T& t,
enable_if_t<!is_void<typename T::iterator>::value>* = nullptr) {
cerr << "[ ";
for (auto it = t.begin(); it != t.end();) {
dump(*it);
cerr << (++it == t.end() ? "" : ", ");
}
cerr << " ]";
}
template <typename T, typename U>
void dump(const pair<T, U>& t) {
cerr << "( ";
dump(t.first);
cerr << ", ";
dump(t.second);
cerr << " )";
}
template <typename T>
void dump(const pair<T*, int>& t) {
cerr << "[ ";
for (int i = 0; i < t.second; i++) {
dump(t.first[i]);
cerr << (i == t.second - 1 ? "" : ", ");
}
cerr << " ]";
}
void trace() { cerr << endl; }
template <typename Head, typename... Tail>
void trace(Head&& head, Tail&&... tail) {
cerr << " ";
dump(head);
if (sizeof...(tail) != 0) cerr << ",";
trace(forward<Tail>(tail)...);
}
} // namespace DebugImpl
#ifdef NyaanDebug
#define trc(...) \
do { \
cerr << "## " << #__VA_ARGS__ << " = "; \
DebugImpl::trace(__VA_ARGS__); \
} while (0)
#else
#define trc(...) (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(); }
//
struct Timer {
chrono::high_resolution_clock::time_point st;
Timer() { reset(); }
void reset() { st = chrono::high_resolution_clock::now(); }
chrono::milliseconds::rep elapsed() {
auto ed = chrono::high_resolution_clock::now();
return chrono::duration_cast<chrono::milliseconds>(ed - st).count();
}
};
template <uint32_t mod>
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<mod>(t);
return (is);
}
constexpr u32 get() const {
u32 ret = reduce(a);
return ret >= mod ? ret - mod : ret;
}
static constexpr u32 get_mod() { return mod; }
};
#pragma GCC optimize("O3,unroll-loops")
#pragma GCC target("avx2")
using m256 = __m256i;
struct alignas(32) mmint {
m256 x;
static mmint R, M0, M1, M2, N2;
mmint() : x() {}
inline mmint(const m256& _x) : x(_x) {}
inline mmint(unsigned int a) : x(_mm256_set1_epi32(a)) {}
inline mmint(unsigned int a0, unsigned int a1, unsigned int a2,
unsigned int a3, unsigned int a4, unsigned int a5,
unsigned int a6, unsigned int a7)
: x(_mm256_set_epi32(a7, a6, a5, a4, a3, a2, a1, a0)) {}
inline operator m256&() { return x; }
inline operator const m256&() const { return x; }
inline int& operator[](int i) { return *(reinterpret_cast<int*>(&x) + i); }
inline const int& operator[](int i) const {
return *(reinterpret_cast<const int*>(&x) + i);
}
friend ostream& operator<<(ostream& os, const mmint& m) {
unsigned r = R[0], mod = M1[0];
auto reduce1 = [&](const uint64_t& b) {
unsigned res = (b + uint64_t(unsigned(b) * unsigned(-r)) * mod) >> 32;
return res >= mod ? res - mod : res;
};
for (int i = 0; i < 8; i++) {
os << reduce1(m[i]) << (i == 7 ? "" : " ");
}
return os;
}
template <typename mint>
static void set_mod() {
R = _mm256_set1_epi32(mint::r);
M0 = _mm256_setzero_si256();
M1 = _mm256_set1_epi32(mint::get_mod());
M2 = _mm256_set1_epi32(mint::get_mod() * 2);
N2 = _mm256_set1_epi32(mint::n2);
}
static inline mmint reduce(const mmint& prod02, const mmint& prod13) {
m256 unpalo = _mm256_unpacklo_epi32(prod02, prod13);
m256 unpahi = _mm256_unpackhi_epi32(prod02, prod13);
m256 prodlo = _mm256_unpacklo_epi64(unpalo, unpahi);
m256 prodhi = _mm256_unpackhi_epi64(unpalo, unpahi);
m256 hiplm1 = _mm256_add_epi32(prodhi, M1);
m256 prodlohi = _mm256_shuffle_epi32(prodlo, 0xF5);
m256 lmlr02 = _mm256_mul_epu32(prodlo, R);
m256 lmlr13 = _mm256_mul_epu32(prodlohi, R);
m256 prod02_ = _mm256_mul_epu32(lmlr02, M1);
m256 prod13_ = _mm256_mul_epu32(lmlr13, M1);
m256 unpalo_ = _mm256_unpacklo_epi32(prod02_, prod13_);
m256 unpahi_ = _mm256_unpackhi_epi32(prod02_, prod13_);
m256 prod = _mm256_unpackhi_epi64(unpalo_, unpahi_);
return _mm256_sub_epi32(hiplm1, prod);
}
static inline mmint itom(const mmint& A) { return A * N2; }
static inline mmint mtoi(const mmint& A) {
m256 A13 = _mm256_shuffle_epi32(A, 0xF5);
m256 lmlr02 = _mm256_mul_epu32(A, R);
m256 lmlr13 = _mm256_mul_epu32(A13, R);
m256 prod02_ = _mm256_mul_epu32(lmlr02, M1);
m256 prod13_ = _mm256_mul_epu32(lmlr13, M1);
m256 unpalo_ = _mm256_unpacklo_epi32(prod02_, prod13_);
m256 unpahi_ = _mm256_unpackhi_epi32(prod02_, prod13_);
m256 prod = _mm256_unpackhi_epi64(unpalo_, unpahi_);
m256 cmp = _mm256_cmpgt_epi32(prod, M0);
m256 dif = _mm256_and_si256(cmp, M1);
return _mm256_sub_epi32(dif, prod);
}
friend inline mmint operator+(const mmint& A, const mmint& B) {
m256 apb = _mm256_add_epi32(A, B);
m256 ret = _mm256_sub_epi32(apb, M2);
m256 cmp = _mm256_cmpgt_epi32(M0, ret);
m256 add = _mm256_and_si256(cmp, M2);
return _mm256_add_epi32(add, ret);
}
friend inline mmint operator-(const mmint& A, const mmint& B) {
m256 ret = _mm256_sub_epi32(A, B);
m256 cmp = _mm256_cmpgt_epi32(M0, ret);
m256 add = _mm256_and_si256(cmp, M2);
return _mm256_add_epi32(add, ret);
}
friend inline mmint operator*(const mmint& A, const mmint& B) {
m256 a13 = _mm256_shuffle_epi32(A, 0xF5);
m256 b13 = _mm256_shuffle_epi32(B, 0xF5);
m256 prod02 = _mm256_mul_epu32(A, B);
m256 prod13 = _mm256_mul_epu32(a13, b13);
return reduce(prod02, prod13);
}
inline mmint& operator+=(const mmint& A) { return (*this) = (*this) + A; }
inline mmint& operator-=(const mmint& A) { return (*this) = (*this) - A; }
inline mmint& operator*=(const mmint& A) { return (*this) = (*this) * A; }
bool operator==(const mmint& A) {
m256 sub = _mm256_sub_epi32(x, A.x);
return _mm256_testz_si256(sub, sub) == 1;
}
bool operator!=(const mmint& A) { return !((*this) == A); }
};
__attribute__((aligned(32))) mmint mmint::R;
__attribute__((aligned(32))) mmint mmint::M0, mmint::M1, mmint::M2, mmint::N2;
/**
* @brief vectorize modint
*/
template <typename T>
struct Binomial {
vector<T> f, g, h;
Binomial(int MAX = 0) : f(1, T(1)), g(1, T(1)), h(1, T(1)) {
while (MAX >= (int)f.size()) extend();
}
void extend() {
int n = f.size();
int m = n * 2;
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); }
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);
}
T H(int n, int r) {
if (n < 0 || r < 0) return T(0);
return r == 0 ? 1 : C(n + r - 1, r);
}
};
using namespace Nyaan;
using mint = LazyMontgomeryModInt<998244353>;
using vm = vector<mint>;
Binomial<mint> C;
mmint F[13000];
mmint DP[13000];
mmint NX[13000];
void Nyaan::solve() {
Timer timer;
mmint::set_mod<mint>();
memset(F, 0, sizeof(F));
memset(DP, 0, sizeof(F));
memset(NX, 0, sizeof(F));
ini(N);
vi a(N);
in(a);
// calc coeff
vm f(N);
// 右端以外
f[0] = mint(1) / 2;
reg(i, 1, N - 1) {
f[i] = C.finv(i + 1) * (C.inv(2) + i) * (C.inv(2).pow(i));
f[i] += f[i - 1];
}
// 右端
{
mint buf = 1;
for (int i = 0; i < N; i++) {
f.back() += buf;
buf *= C.inv(2) * C.inv(i + 1);
}
}
mmint* dp = DP + 1;
mmint* nx = NX + 1;
for (int i = 0; i < N; i++) F[i / 8][i % 8] = f[i].a;
for (int i = 0; i < N; i++) dp[i / 8][i % 8] = mint(a[i]).a;
mint coe = mint(2).pow(1LL * 2 * (N - 1)) * C.fac(N - 1);
m256 MOD = _mm256_set1_epi32(998244353);
mmint th1, th2, zero = _mm256_setzero_si256();
th1[1] = th1[3] = th1[5] = th1[7] = mmint::M1[0];
th2[1] = th2[3] = th2[5] = th2[7] = mmint::M2[0];
auto normalize = [&MOD](const mmint& data) -> m256 {
m256 flag = _mm256_cmpgt_epi32(data, MOD);
m256 dif = _mm256_and_si256(flag, MOD);
return _mm256_sub_epi32(data, dif);
};
#define INIT_X(x) \
m256 prod02##x = _mm256_setzero_si256(); \
m256 prod13##x = _mm256_setzero_si256()
#define ADD(x) \
m256 f02##x = normalize(F[j + x]); \
m256 f13##x = _mm256_shuffle_epi32(f02##x, 0xF5); \
m256 dp02##x = normalize(dp[j + x]); \
m256 dp13##x = _mm256_shuffle_epi32(dp02##x, 0xF5); \
m256 fd02##x = _mm256_mul_epi32(f02##x, dp02##x); \
m256 fd13##x = _mm256_mul_epi32(f13##x, dp13##x); \
prod02##x = _mm256_add_epi64(prod02##x, fd02##x); \
prod13##x = _mm256_add_epi64(prod13##x, fd13##x)
#define COMP(x) \
m256 cmp02##x = _mm256_cmpgt_epi64(zero, prod02##x); \
m256 cmp13##x = _mm256_cmpgt_epi64(zero, prod13##x); \
m256 dif02##x = _mm256_and_si256(cmp02##x, th2); \
m256 dif13##x = _mm256_and_si256(cmp13##x, th2); \
prod02##x = _mm256_sub_epi64(prod02##x, dif02##x); \
prod13##x = _mm256_sub_epi64(prod13##x, dif13##x)
#define REDUCE(x) \
for (int _ = 0; _ < 2; _++) { \
m256 cmp02 = _mm256_cmpgt_epi64(prod02##x, th1); \
m256 cmp13 = _mm256_cmpgt_epi64(prod13##x, th1); \
m256 dif02 = _mm256_and_si256(cmp02, th1); \
m256 dif13 = _mm256_and_si256(cmp13, th1); \
prod02##x = _mm256_sub_epi64(prod02##x, dif02); \
prod13##x = _mm256_sub_epi64(prod13##x, dif13); \
} \
buf += mmint::reduce(prod02##x, prod13##x)
auto done = [&]() {
INIT_X(0);
INIT_X(1);
INIT_X(2);
INIT_X(3);
for (int i = 0; i < N / 8 + 32; i += 32) {
for (int j = i; j < i + 32; j += 4) {
ADD(0);
ADD(1);
ADD(2);
ADD(3);
}
COMP(0);
COMP(1);
COMP(2);
COMP(3);
}
mmint buf{0};
REDUCE(0);
REDUCE(1);
REDUCE(2);
REDUCE(3);
buf = mmint::mtoi(buf);
mint res = 0;
rep(i, 8) res += buf[i];
out(res * coe);
};
done();
mint Nm2 = N - 2;
mint Nm3 = N - 3;
rep(_, N) {
mint* dp0 = reinterpret_cast<mint*>(dp);
mint* nx0 = reinterpret_cast<mint*>(nx);
mmint NM3{Nm3.a};
for (int i = 0; i < N / 8 + 3; i++) {
__m256i p1 = _mm256_loadu_si256((__m256i*)(dp0 + i * 8 - 1));
__m256i p2 = _mm256_loadu_si256((__m256i*)(dp0 + i * 8 + 1));
nx[i] = dp[i] * NM3 + p1 + p2;
}
nx0[0] = dp0[1] + dp0[0] * Nm2;
nx0[N - 1] = dp0[N - 2] + dp0[N - 1] * Nm2;
swap(dp, nx);
trc(dp);
done();
}
cerr << timer.elapsed() << "\n";
}