NTT_any_mod.cpp

//
// 任意 mod の畳み込み
//
// verified:
//   Yosupo Library Checker - Convolution (Mod 1,000,000,007)
//     https://judge.yosupo.jp/problem/convolution_mod_1000000007
//


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


// modint
template<int MOD> struct Fp {
    // inner value
    long long val;
    
    // constructor
    constexpr Fp() : val(0) { }
    constexpr Fp(long long v) : val(v % MOD) {
        if (val < 0) val += MOD;
    }
    constexpr long long get() const { return val; }
    constexpr int get_mod() const { return MOD; }
    
    // arithmetic operators
    constexpr Fp operator + () const { return Fp(*this); }
    constexpr Fp operator - () const { return Fp(0) - Fp(*this); }
    constexpr Fp operator + (const Fp &r) const { return Fp(*this) += r; }
    constexpr Fp operator - (const Fp &r) const { return Fp(*this) -= r; }
    constexpr Fp operator * (const Fp &r) const { return Fp(*this) *= r; }
    constexpr Fp operator / (const Fp &r) const { return Fp(*this) /= r; }
    constexpr Fp& operator += (const Fp &r) {
        val += r.val;
        if (val >= MOD) val -= MOD;
        return *this;
    }
    constexpr Fp& operator -= (const Fp &r) {
        val -= r.val;
        if (val < 0) val += MOD;
        return *this;
    }
    constexpr Fp& operator *= (const Fp &r) {
        val = val * r.val % MOD;
        return *this;
    }
    constexpr Fp& operator /= (const Fp &r) {
        long long a = r.val, b = MOD, u = 1, v = 0;
        while (b) {
            long long t = a / b;
            a -= t * b, swap(a, b);
            u -= t * v, swap(u, v);
        }
        val = val * u % MOD;
        if (val < 0) val += MOD;
        return *this;
    }
    constexpr Fp pow(long long n) const {
        Fp res(1), mul(*this);
        while (n > 0) {
            if (n & 1) res *= mul;
            mul *= mul;
            n >>= 1;
        }
        return res;
    }
    constexpr Fp inv() const {
        Fp res(1), div(*this);
        return res / div;
    }

    // other operators
    constexpr bool operator == (const Fp &r) const {
        return this->val == r.val;
    }
    constexpr bool operator != (const Fp &r) const {
        return this->val != r.val;
    }
    constexpr Fp& operator ++ () {
        ++val;
        if (val >= MOD) val -= MOD;
        return *this;
    }
    constexpr Fp& operator -- () {
        if (val == 0) val += MOD;
        --val;
        return *this;
    }
    constexpr Fp operator ++ (int) const {
        Fp res = *this;
        ++*this;
        return res;
    }
    constexpr Fp operator -- (int) const {
        Fp res = *this;
        --*this;
        return res;
    }
    friend constexpr istream& operator >> (istream &is, Fp<MOD> &x) {
        is >> x.val;
        x.val %= MOD;
        if (x.val < 0) x.val += MOD;
        return is;
    }
    friend constexpr ostream& operator << (ostream &os, const Fp<MOD> &x) {
        return os << x.val;
    }
    friend constexpr Fp<MOD> pow(const Fp<MOD> &r, long long n) {
        return r.pow(n);
    }
    friend constexpr Fp<MOD> inv(const Fp<MOD> &r) {
        return r.inv();
    }
};

namespace NTT {
    long long modpow(long long a, long long n, int mod) {
        long long res = 1;
        while (n > 0) {
            if (n & 1) res = res * a % mod;
            a = a * a % mod;
            n >>= 1;
        }
        return res;
    }

    long long modinv(long long a, int mod) {
        long long b = mod, u = 1, v = 0;
        while (b) {
            long long t = a / b;
            a -= t * b, swap(a, b);
            u -= t * v, swap(u, v);
        }
        u %= mod;
        if (u < 0) u += mod;
        return u;
    }

    int calc_primitive_root(int mod) {
        if (mod == 2) return 1;
        if (mod == 167772161) return 3;
        if (mod == 469762049) return 3;
        if (mod == 754974721) return 11;
        if (mod == 998244353) return 3;
        int divs[20] = {};
        divs[0] = 2;
        int cnt = 1;
        long long x = (mod - 1) / 2;
        while (x % 2 == 0) x /= 2;
        for (long long i = 3; i * i <= x; i += 2) {
            if (x % i == 0) {
                divs[cnt++] = i;
                while (x % i == 0) x /= i;
            }
        }
        if (x > 1) divs[cnt++] = x;
        for (int g = 2;; g++) {
            bool ok = true;
            for (int i = 0; i < cnt; i++) {
                if (modpow(g, (mod - 1) / divs[i], mod) == 1) {
                    ok = false;
                    break;
                }
            }
            if (ok) return g;
        }
    }

    int get_fft_size(int N, int M) {
        int size_a = 1, size_b = 1;
        while (size_a < N) size_a <<= 1;
        while (size_b < M) size_b <<= 1;
        return max(size_a, size_b) << 1;
    }

    // number-theoretic transform
    template<class mint> void trans(vector<mint> &v, bool inv = false) {
        if (v.empty()) return;
        int N = (int)v.size();
        int MOD = v[0].get_mod();
        int PR = calc_primitive_root(MOD);
        static bool first = true;
        static vector<long long> vbw(30), vibw(30);
        if (first) {
            first = false;
            for (int k = 0; k < 30; ++k) {
                vbw[k] = modpow(PR, (MOD - 1) >> (k + 1), MOD);
                vibw[k] = modinv(vbw[k], MOD);
            }
        }
        for (int i = 0, j = 1; j < N - 1; j++) {
            for (int k = N >> 1; k > (i ^= k); k >>= 1);
            if (i > j) swap(v[i], v[j]);
        }
        for (int k = 0, t = 2; t <= N; ++k, t <<= 1) {
            long long bw = vbw[k];
            if (inv) bw = vibw[k];
            for (int i = 0; i < N; i += t) {
                mint w = 1;
                for (int j = 0; j < t/2; ++j) {
                    int j1 = i + j, j2 = i + j + t/2;
                    mint c1 = v[j1], c2 = v[j2] * w;
                    v[j1] = c1 + c2;
                    v[j2] = c1 - c2;
                    w *= bw;
                }
            }
        }
        if (inv) {
            long long invN = modinv(N, MOD);
            for (int i = 0; i < N; ++i) v[i] = v[i] * invN;
        }
    }

    // for garner
    static constexpr int MOD0 = 754974721;
    static constexpr int MOD1 = 167772161;
    static constexpr int MOD2 = 469762049;
    using mint0 = Fp<MOD0>;
    using mint1 = Fp<MOD1>;
    using mint2 = Fp<MOD2>;
    static const mint1 imod0 = 95869806; // modinv(MOD0, MOD1);
    static const mint2 imod1 = 104391568; // modinv(MOD1, MOD2);
    static const mint2 imod01 = 187290749; // imod1 / MOD0;

    // small case (T = mint, long long)
    template<class T> vector<T> naive_mul(const vector<T> &A, const vector<T> &B) {
        if (A.empty() || B.empty()) return {};
        int N = (int)A.size(), M = (int)B.size();
        vector<T> res(N + M - 1);
        for (int i = 0; i < N; ++i)
            for (int j = 0; j < M; ++j)
                res[i + j] += A[i] * B[j];
        return res;
    }

    // mul by convolution
    template<class mint> vector<mint> mul(const vector<mint> &A, const vector<mint> &B) {
        if (A.empty() || B.empty()) return {};
        int N = (int)A.size(), M = (int)B.size();
        if (min(N, M) < 30) return naive_mul(A, B);
        int MOD = A[0].get_mod();
        int size_fft = get_fft_size(N, M);
        if (MOD == 998244353) {
            vector<mint> a(size_fft), b(size_fft), c(size_fft);
            for (int i = 0; i < N; ++i) a[i] = A[i];
            for (int i = 0; i < M; ++i) b[i] = B[i];
            trans(a), trans(b);
            vector<mint> res(size_fft);
            for (int i = 0; i < size_fft; ++i) res[i] = a[i] * b[i];
            trans(res, true);
            res.resize(N + M - 1);
            return res;
        }
        vector<mint0> a0(size_fft, 0), b0(size_fft, 0), c0(size_fft, 0);
        vector<mint1> a1(size_fft, 0), b1(size_fft, 0), c1(size_fft, 0);
        vector<mint2> a2(size_fft, 0), b2(size_fft, 0), c2(size_fft, 0);
        for (int i = 0; i < N; ++i)
            a0[i] = A[i].val, a1[i] = A[i].val, a2[i] = A[i].val;
        for (int i = 0; i < M; ++i)
            b0[i] = B[i].val, b1[i] = B[i].val, b2[i] = B[i].val;
        trans(a0), trans(a1), trans(a2), trans(b0), trans(b1), trans(b2);
        for (int i = 0; i < size_fft; ++i) {
            c0[i] = a0[i] * b0[i];
            c1[i] = a1[i] * b1[i];
            c2[i] = a2[i] * b2[i];
        }
        trans(c0, true), trans(c1, true), trans(c2, true);
        mint mod0 = MOD0, mod01 = mod0 * MOD1;
        vector<mint> res(N + M - 1);
        for (int i = 0; i < N + M - 1; ++i) {
            int y0 = c0[i].val;
            int y1 = (imod0 * (c1[i] - y0)).val;
            int y2 = (imod01 * (c2[i] - y0) - imod1 * y1).val;
            res[i] = mod01 * y2 + mod0 * y1 + y0;
        }
        return res;
    }
};



//------------------------------//
// Examples
//------------------------------//

void Yosupo_Convolution_mod_1000000007() {
    const int MOD = 1000000007;
    using mint = Fp<MOD>;
    
    int N, M;
    cin >> N >> M;
    vector<mint> a(N), b(M);
    for (int i = 0; i < N; ++i) cin >> a[i];
    for (int i = 0; i < M; ++i) cin >> b[i];
    vector<mint> res = NTT::mul(a, b);
    for (auto v : res) cout << v << " ";
    cout << endl;
}


int main() {
    Yosupo_Convolution_mod_1000000007();
}