kyopro-lib

This documentation is automatically generated by online-judge-tools/verification-helper

View on GitHub

:x: Formal power series (Sqrt)
(Mylib/Math/fps_sqrt.cpp)

Operations

Requirements

Notes

Problems

References

Depends on

Verified with

Code

#pragma once
#include "Mylib/Math/formal_power_series.cpp"
#include "Mylib/Number/Mod/mod_sqrt.cpp"

namespace haar_lib {
  template <typename T, const auto &convolve>
  auto formal_power_series<T, convolve>::sqrt() const -> std::optional<formal_power_series<T, convolve>> {
    constexpr int mod = value_type::mod();

    const int n = data_.size();
    int k       = 0;
    for (; k < n; ++k)
      if (data_[k] != 0) break;

    if (k >= n) return *this;
    if (k % 2 != 0) return {};

    auto x = mod_sqrt((int64_t) data_[k], mod);

    if (not x) return {};

    const int m = n - k;

    auto it = data_.begin() + k;
    formal_power_series ret({*x});

    int t = 1;
    while (t <= m * 2) {
      formal_power_series f(std::vector<T>(it, it + std::min(t, m)));
      ret.resize(t);
      f.resize(t);
      ret = (ret + f * ret.inv()) * T(2).inv();
      t <<= 1;
    }

    ret.resize(n);
    ret = ret.shift(k / 2);

    return ret;
  }
}  // namespace haar_lib
#line 2 "Mylib/Math/formal_power_series.cpp"
#include <cassert>
#include <functional>
#include <initializer_list>
#include <vector>

namespace haar_lib {
  template <typename T, const auto &convolve>
  class formal_power_series {
  public:
    using value_type = T;

  private:
    std::vector<T> data_;

  public:
    formal_power_series() {}
    explicit formal_power_series(int N) : data_(N) {}
    formal_power_series(const std::vector<T> &data_) : data_(data_) {}
    formal_power_series(std::initializer_list<T> init) : data_(init.begin(), init.end()) {}
    formal_power_series(const formal_power_series &a) : data_(a.data_) {}
    formal_power_series(formal_power_series &&a) noexcept { *this = std::move(a); }

    size_t size() const {
      return data_.size();
    }

    const T &operator[](int i) const {
      return data_[i];
    }

    T &operator[](int i) {
      return data_[i];
    }

    auto begin() { return data_.begin(); }
    auto end() { return data_.end(); }

    const auto &data() const { return data_; }

    void resize(int n) {
      data_.resize(n);
    }

    auto &operator=(formal_power_series &&rhs) noexcept {
      if (this != &rhs) {
        data_ = std::move(rhs.data_);
      }
      return *this;
    }

    auto &operator+=(const formal_power_series &rhs) {
      if (data_.size() < rhs.size()) data_.resize(rhs.size());
      for (int i = 0; i < rhs.size(); ++i) data_[i] += rhs[i];
      return *this;
    }

    auto &operator+=(T rhs) {
      data_[0] += rhs;
      return *this;
    }

    auto operator+(T rhs) const {
      auto ret = *this;
      return ret += rhs;
    }

    auto operator+(const formal_power_series &rhs) const {
      auto ret = *this;
      return ret += rhs;
    }

    auto &operator-=(const formal_power_series &rhs) {
      if (data_.size() < rhs.size()) data_.resize(rhs.size());
      for (int i = 0; i < rhs.size(); ++i) data_[i] -= rhs[i];
      return *this;
    }

    auto &operator-=(T rhs) {
      data_[0] -= rhs;
      return *this;
    }

    auto operator-(T rhs) const {
      auto ret = *this;
      return ret -= rhs;
    }

    auto operator-(const formal_power_series &rhs) const {
      auto ret = *this;
      return ret -= rhs;
    }

    auto operator-() const {
      auto ret = *this;
      for (auto &x : ret) x = -x;
      return ret;
    }

    auto &operator*=(const formal_power_series &rhs) {
      data_ = convolve(data_, rhs.data_);
      return *this;
    }

    auto operator*(const formal_power_series &rhs) const {
      auto ret = convolve(data_, rhs.data_);
      return formal_power_series(ret);
    }

    auto &operator*=(T rhs) {
      for (auto &x : data_) x *= rhs;
      return *this;
    }

    auto operator*(T rhs) const {
      auto ret = *this;
      return ret *= rhs;
    }

    auto differentiate() const {
      const int n = data_.size();
      std::vector<T> ret(n - 1);
      for (int i = 0; i < n - 1; ++i) {
        ret[i] = data_[i + 1] * (i + 1);
      }

      return formal_power_series(ret);
    }

    auto integrate() const {
      const int n = data_.size();
      std::vector<T> ret(n + 1), invs(n + 1, 1);
      const int p = T::mod();
      for (int i = 2; i <= n; ++i) invs[i] = -invs[p % i] * (p / i);
      for (int i = 0; i < n; ++i) {
        ret[i + 1] = data_[i] * invs[i + 1];
      }

      return formal_power_series(ret);
    }

    auto inv() const {
      assert(data_[0] != 0);
      const int n = data_.size();

      int t              = 1;
      std::vector<T> ret = {data_[0].inv()};
      ret.reserve(n * 2);

      while (t <= n * 2) {
        std::vector<T> c(data_.begin(), data_.begin() + std::min(t, n));
        auto a = convolve(ret, ret, true);
        if ((int) a.size() > t) a.resize(t);

        c = convolve(c, a);

        if ((int) c.size() > t) c.resize(t);
        if ((int) ret.size() > t) ret.resize(t);

        for (int i = 0; i < (int) ret.size(); ++i) ret[i] = ret[i] * 2;

        if (ret.size() < c.size()) ret.resize(std::min<int>(c.size(), t));

        for (int i = 0; i < (int) c.size(); ++i) {
          ret[i] -= c[i];
        }

        t <<= 1;
      }

      ret.resize(n);

      return formal_power_series(ret);
    }

    auto log() const {
      assert(data_[0] == 1);
      const int n = data_.size();
      auto ret    = (differentiate() * inv()).integrate();
      ret.resize(n);
      return ret;
    }

    auto exp() const {
      const int n = data_.size();

      int t = 1;
      formal_power_series b({1});

      while (t <= n * 2) {
        t <<= 1;
        auto temp = b.log();
        temp.resize(t);

        for (int i = 0; i < t; ++i) temp[i] = -temp[i];
        temp[0] += 1;
        for (int i = 0; i < std::min(t, n); ++i) temp[i] += data_[i];

        b = b * temp;
        b.resize(t);
      }

      b.resize(n);

      return b;
    }

    auto shift(int64_t k) const {
      const int64_t n = data_.size();
      formal_power_series ret(n);

      if (k >= 0) {
        for (int64_t i = k; i < n; ++i) {
          ret[i] = data_[i - k];
        }
      } else {
        for (int64_t i = 0; i < n + k; ++i) {
          ret[i] = data_[i - k];
        }
      }

      return ret;
    }

    auto pow(int64_t M) const {
      assert(M >= 0);

      const int n = data_.size();
      int k       = 0;
      for (; k < n; ++k) {
        if (data_[k] != 0) {
          break;
        }
      }

      if (k >= n) return *this;

      T a = data_[k];

      formal_power_series ret = *this;
      ret                     = (ret.shift(-k)) * a.inv();
      ret                     = (ret.log() * (T) M).exp();
      ret                     = (ret * a.pow(M)).shift(M * k);

      return ret;
    }

    std::optional<formal_power_series> sqrt() const;
  };
}  // namespace haar_lib
#line 2 "Mylib/Number/Mod/mod_sqrt.cpp"
#include <optional>
#include <random>
#line 2 "Mylib/Number/Mod/mod_pow.cpp"
#include <cstdint>

namespace haar_lib {
  constexpr int64_t mod_pow(int64_t n, int64_t p, int64_t m) {
    int64_t ret = 1;
    while (p > 0) {
      if (p & 1) (ret *= n) %= m;
      (n *= n) %= m;
      p >>= 1;
    }
    return ret;
  }
}  // namespace haar_lib
#line 5 "Mylib/Number/Mod/mod_sqrt.cpp"

namespace haar_lib {
  std::optional<int64_t> mod_sqrt(int64_t a, int64_t p) {
    if (p == 2) return a % 2;
    if (a == 0) return 0;

    int64_t b = mod_pow(a, (p - 1) / 2, p);

    if (b == p - 1) return {};
    if (p % 4 == 3) return mod_pow(a, (p + 1) / 4, p);

    int64_t q = p - 1, s = 0;
    while (q % 2 == 0) {
      q /= 2;
      s += 1;
    }

    static std::mt19937_64 rand(time(0));
    std::uniform_int_distribution<> dist(0, p - 1);

    int64_t z;
    while (1) {
      z = dist(rand);
      if (mod_pow(z, (p - 1) / 2, p) == p - 1) break;
    }

    int64_t m = s;
    int64_t c = mod_pow(z, q, p);
    int64_t t = mod_pow(a, q, p);
    int64_t r = mod_pow(a, (q + 1) / 2, p);

    while (1) {
      if (t == 0) return 0;
      if (t == 1) return r;

      int i = 1;
      for (int64_t T = t; i < m; ++i) {
        (T *= T) %= p;
        if (T == 1) break;
      }

      int64_t b = mod_pow(c, 1LL << (m - i - 1), p);

      m = i;
      c = b * b % p;
      (t *= b * b % p) %= p;
      (r *= b) %= p;
    }
  }
}  // namespace haar_lib
#line 4 "Mylib/Math/fps_sqrt.cpp"

namespace haar_lib {
  template <typename T, const auto &convolve>
  auto formal_power_series<T, convolve>::sqrt() const -> std::optional<formal_power_series<T, convolve>> {
    constexpr int mod = value_type::mod();

    const int n = data_.size();
    int k       = 0;
    for (; k < n; ++k)
      if (data_[k] != 0) break;

    if (k >= n) return *this;
    if (k % 2 != 0) return {};

    auto x = mod_sqrt((int64_t) data_[k], mod);

    if (not x) return {};

    const int m = n - k;

    auto it = data_.begin() + k;
    formal_power_series ret({*x});

    int t = 1;
    while (t <= m * 2) {
      formal_power_series f(std::vector<T>(it, it + std::min(t, m)));
      ret.resize(t);
      f.resize(t);
      ret = (ret + f * ret.inv()) * T(2).inv();
      t <<= 1;
    }

    ret.resize(n);
    ret = ret.shift(k / 2);

    return ret;
  }
}  // namespace haar_lib
Back to top page