Large Arithmetic by MinGW fails

Question

My general task is to compute According to Wolfram,

I expected the following to return the same:

#include <iostream>
#include <vector>
#include <math.h>

unsigned long long f(const std::vector<unsigned> &coefficients)
{
    unsigned short exponent = coefficients.size() - 1;
    unsigned long long sum;
    for (unsigned i : coefficients)
    {
        sum += i * pow(36, exponent);
        --exponent;
    }
    return sum;
}

int main()
{
    std::cout << f({9,13,19,6,7,8,2});
}

but instead it returns 20416905041. The minimum capacity of an unsigned long long integer is 0 to 18446744073709551615, according to Alex B on this question, so capacity doesn't appear to be the problem.

---

Specifications:

• Compiler: x86_64-w64-mingw32 from the Windows TDM-GCC Compiler Suite
• Compiled via g++ mwe.cpp -std=c++11 -omwe
• OS: Windows 10 Home

Answer 1

From M.M's comment on the question:

pow is a floating point function and therefore will be susceptible to rounding errors. You should use integer variables (do repeated multiplication by 36) instead.