CODEWITHSUNDEEP
pythonboostarbitrary-precisionboost-multiprecision
Divye Sharma
Divye Sharma

Reputation: 11

Boost multiprecision library

Boost multiprecision libraries highest value possible

Can the boost multiprecision handle values as large as googolplex , like maybe the boost multiprecision cpp dec float library?

So can someone please tell me the largest value possible in any boost multiprecision library?

Upvotes: 1

Views: 47

Answers (1)

sehe
sehe

Reputation: 393674

cpp_dec_float would have been your best bet, solely because the base-10 mantissa is always 1 for googol(plex). However, the exponent type is required to be a builtin integer type. std::int128_t doesn't even qualify if the compiler supports it. So, that's right off.

Using GMP backends appears to break down at 5e+17 digits:

#include <boost/multiprecision/gmp.hpp>
#include <iostream>

int main() {
    using namespace boost::multiprecision;

    using Real = mpf_float;
    using Int  = mpz_int;

    // create a googolplex
    Int googol = pow(Int(10), 100u);

    // print it
    std::cout << googol << std::endl;
    std::cout << "check: " << log10(Real(googol)) << std::endl;

    // no plex for you
    auto not_googolplex = pow(Real(10), Real("5e17"));
    std::cout << "check: " << log10(not_googolplex) << std::endl;

    not_googolplex = pow(Real(10), Real("5e18"));
    std::cout << "check: " << log10(not_googolplex) << std::endl;
}

Printing (on my AMD64 64GiB box using GCC14 or Clang18 and Boost 1.86):

10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
check: 100
check: 5e+17
GNU MP: Cannot allocate memory (size=4611686018427387960)
Aborted (core dumped)

Upvotes: 0

Related Questions