relative error of machine epsilon

Hi!

I am trying to calculate the relative error of machine epsilon. I have the roundoff error but can seem to figure how to get the relative error... here is my function:

function eplison = ep(n)
 eplison = double(1);
 r = eplison/2;
 while (n+eplison) ~= n
 eplison = eplison/2;
 r = eplison/2;
 end
 eplison= eplison*2;
end

thanks in advance

Answer 1

[Rohit Pappu]

According to the formal definition, the approximation of Machine Epsilon is as follows

epsilon = 1.0;

while (1.0 + 0.5 * epsilon) ≠ 1.0:
    epsilon = 0.5 * epsilon

Extending this to n, it can be rewritten as

epsilon = 1.0; % Default datatype is double
while(n + 0.5*epsilon ~= n)
    epsilon = 0.5*epsilon;
end