Maximum errors between two vectors

Hello,

I am calculating the max(abs(X-Y)) where X is the true solution and Y is the calculated solution in some method, and obtaining the maximum difference 3.3307e-16, which is unrealistic to me. I expected up to 10^(-4) to 10^(-12). Is the obtained answer look like smaller than machine epsilon? Can any one expalin me? Thanks in advance.

Answer 1

[the cyclist]

The floating-point error of double-precision values that are of the order 1 is around 1.e-16:

eps(1)

ans = 2.2204e-16

See the documentation for the eps function.