> Thank you for the feedback. I'll try to change the code with your > suggestions. With respect to your reply for step 21, I did not make any > changes to step 21. I just ran the code that is there with a cout > statement for grad_phi_i_s and noticed that it gave an output of 0 0 for > every time step (hence not contributing to rhs?) > > const Tensor<1,dim> grad_phi_i_s = fe_values[saturation].gradient > (i, q); > > std::cout << grad_phi_i_s <<endl; //this is the only line that > I added
Ah, I see now. The reason is that by default we use piecewise constant (but discontinuous) elements for the saturation in this program -- consequently, the gradient is always zero. However, this wouldn't be the case any more if you used, for example, Raviart-Thomas elements of degree 2 and piecewise linear, discontinuous elements for the saturation. Best W. > -----Original Message----- > From: Wolfgang Bangerth [mailto:[email protected]] > Sent: Fri 2/5/2010 12:04 PM > To: Tirupathi, Seshu > Subject: Re: Deal ii / Step 31 > > > Seshu, > > > 1. Change Laplacian of velocity to just velocity in the stokes > > equation. > > This leads to the mixed formulation of the Laplace equation. You should > take a look step-21 for this. In general, one thing you need to look out > for is that the finite elements that you can use for the Stokes > equations don't work for the mixed Laplace. > > > 2. I then tried to change the boundary conditions for temperature from > > Neumann to Dirichlet(as was mentioned in the website).I made a couple > > of weak attempts but I am not able to figure out if there is a > > straight forward way to do this. > > The simplest way would be to add boundary values to the ConstraintMatrix > object you have, and have the latter object eliminate boundary nodes as > you copy local contributions to the global matrix. > > > 3. As a side note, I was just checking the output of grad_phi_i_s in > > step 21 and this is always zero(at least for the first 30 time steps). > > Can someone tell me the reason for this? > > If you show me your code I may be able to tell you why. > > > PS: I tried sending the message to to dealii mailing list a couple of > > times yesterday but for some reason the mail keeps bouncing back or > > rather never delivered it till so far. > > Are you signed up for the list? You can only send to the list if you are > signed up to it with exactly the email address you use to send to the > list. > > Best > W. -- ------------------------------------------------------------------------- Wolfgang Bangerth email: [email protected] www: http://www.math.tamu.edu/~bangerth/ _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
