Computation Contest #2 Results and Solution

in programming •  5 years ago 

Solution

The problem of this contest was to find the first zero on the critical line of the riemann zeta function using the following formula:
Screenshot from 2019-10-27 20-41-05.png

I implemented an example solution in java (using a self written implementation of complex numbers which isn't explicitely linked here):

public class ZeroFinder {
    public static Complex zeta(Complex in, int n) {
        Complex res = new Complex(0, 0);
        for(int j = 1; j < n; j++) {
            res = res.plus(in.negate().times(Math.log(j+1)).ePow().times(j).minus(in.negate().times(Math.log(j)).ePow().times(in.negate().plus(j))));
        }
        return res;
    }
    public static void main(String [] args) {
        // Find the first 10 zeros in the real part:
        int [] list = new int[10];
        int index = 0;
        Complex last = new Complex(0, 0);
        for(int i = 1; index < 10; i++) {
            Complex next = zeta(new Complex(0.5, i/10.0), 1000);
            //System.out.println(next);
            if(last.r*next.r < 0) { // Check if the value of r changes sign
                list[index] = i;
                index++;
            }
            last = next;
        }
        
        // Look at the first zeros in more detail and check if there is also an imaginary zero:
        double y0 = 0;
        outer:
        for(int i = 0; i < 10; i++) {
            last = new Complex(0, 0);
            for(int j = 0; j < 1000; j++) {
                double y = list[i]/10.0+j/1000.0-0.5;
                Complex next = zeta(new Complex(0.5, y), 10000);
                if(next.i*last.i < 0) { // Check if the value of i changes sign
                    y0 = y;
                    break outer;
                }
                last = next;
            }
        }
        
        // Look at the region in even more detail:
        double resI = 0, resR = 0;
        last = new Complex(0, 0);
        for(int i = 0; i < 1000; i++) {
            double y = y0+(i-500)/10000.0;
            Complex next = zeta(new Complex(0.5, y), 1000000);
            if(next.i*last.i < 0) { // Check if the value of i changes sign
                resI = y;
            }
            if(next.r*last.r < 0) { // Check if the value of r changes sign
                resR = y;
            }
            last = next;
        }
        System.out.println((resR+resI)/2); // Print the average of resR and resI which should be more accurate.
    }
}

This code prints the result 14.1332 which is not perfectly accurate even though I used 1000000 iterations in the final run.

↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓

List of participants with their entries:

NameSolution found
@tonimontana14.141171601414683A lot of digits for the fact that the fourth digit is already off.
@crokkon14.1

↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓

Winners:

As promised 2 SBI for everyone!

↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓↑↓

The next contest starts in 2 days. Don't miss it!

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
Sort Order:  

Do we even know that the zeros of the truncated sum have anything to do with the zeros of the real function? Like say you havefirefox_20191101_235635.png Then f is 0 but all the finite sums have no zeros.

There is an important difference between the sum I chose and the function you show:
Your function won't give a graph where you could read the zeros( by checking for sign changes).

If you have a sum that converges then you can conclude that your graph will be within some small range ε of the graph of the infinite sum. So the zeros you can read in the partial sum graph should also closely match those in the infinite sum graph.