#include <iostream>
#include <algorithm>
#include <utility>
#include <vector>
#include <numeric>
#include <iomanip>

template <class T, class U>
inline bool chmin(T &lhs, const U &rhs) {
  if (lhs > rhs) {
    lhs = rhs;
    return true;
  }
  return false;
}

template <class T, class U>
inline bool chmax(T &lhs, const U &rhs) {
  if (lhs < rhs) {
    lhs = rhs;
    return true;
  }
  return false;
}

// [l, r) from l to r
struct range {
  struct itr {
    int i;
    constexpr itr(int i_): i(i_) { }
    constexpr void operator ++ () { ++i; }
    constexpr int operator * () const { return i; }
    constexpr bool operator != (itr x) const { return i != x.i; }
  };
  const itr l, r;
  constexpr range(int l_, int r_): l(l_), r(std::max(l_, r_)) { }
  constexpr itr begin() const { return l; }
  constexpr itr end() const { return r; }
};

// [l, r) from r to l
struct revrange {
  struct itr {
    int i;
    constexpr itr(int i_): i(i_) { }
    constexpr void operator ++ () { --i; }
    constexpr int operator * () const { return i; }
    constexpr bool operator != (itr x) const { return i != x.i; }
  };
  const itr l, r;
  constexpr revrange(int l_, int r_): l(l_ - 1), r(std::max(l_, r_) - 1) { }
  constexpr itr begin() const { return r; }
  constexpr itr end() const { return l; }
};

#include <cmath>
using dbl = long double;

int main() {
  int P, Q;
  std::cin >> P >> Q;
  dbl ok = 0, ng = 1e18L;
  for (int step: range(0, 100)) {
    dbl md = (ok + ng) / 2.0L;
    (md * md <= P + Q * md * std::log2(md) ? ok : ng) = md;
  }
  std::cout << std::fixed << std::setprecision(20);
  std::cout << ok << '\n';
  return 0;
}