This is a read only copy of the old FEniCS QA forum. Please visit the new QA forum to ask questions

Error in example d6_p2D in FEniCS book

+1 vote

Hello,

I have been experimenting FEniCS for a few days, and encountered an error in d6_p2D.py in FEniCS book. The error occurs right here,

Ve = FunctionSpace(mesh, 'Lagrange', degree=7)
u_e_Ve = interpolate(u_e, Ve)
error = (u - u_e_Ve)**2*dx
E3 = sqrt((assemble(error)))
         ^

It seems that when the mesh gets finer, assemble() function gives a negative value. Do you have any idea why this is happening?

closed with the note: Problem solved
asked Feb 20, 2014 by ysakamoto FEniCS Novice (260 points)
closed Mar 28, 2014 by chris_richardson

Hi, are you running that code with omega=1? If so, the exact solution is easy to represent and maybe your problems are due to errors being too small (finite precision issues). I
observed assemble(error)=1E-17 and the crash on the subsequent mesh since the error was -1E-20. Try making the solution more challenging by increasing the frequency to hundred or so.

commented Feb 20, 2014 by MiroK FEniCS Expert (80,920 points)

Yes. The finite precision arithmetic error was a source of failure. It works for a larger omega values.

Thanks!

commented Feb 21, 2014 by ysakamoto FEniCS Novice (260 points)
...