Question

I have a function that generates some large numbers, puts them in a vector, sums them, an returns the sum. I have been using std::accumulate to calculate the sum, however, after some testing, I realized it was not returning the expected sum.

typedef unsigned long long ull;
ull sum(ull kLimit)
{
    ull testSum = 0;
    vector<ull> numbers;
    for (ull n = 0; n < kLimit; ++n) {
        if (/* number I want */) {
            numbers.push_back(n);
            // directly sum for testing
            testSum += n;
        }
    }
    ull sum = accumulate(begin(numbers), end(numbers), 0);
    return sum;
}

I would expect the value of sum and testSum to be equal. However, sum is equal to 470064632 and test sum is equal to 82074443256, which is the expected value.

I tried reducing the limit to a much smaller number (500), and the values of sum and testSum were equal. This makes me think the error is overflow with accumulate, I’m not sure what the problem could be though. I am compiling with VS2012 for a x64 platform.

Answer 1

It does look like an overflow error: the numbers are 0x131C049DF8 and 0x1C049DF8.

Try casting your final parameter 0 to type ull, because the type returned by accumulate is the type of that final parameter:

T accumulate(InputIt first, InputIt last, T value) { ... }