You don't need to change them. You can call directly mapping.compute_face_data. In the resulting InternalData object you will have access to all the data you need, including unit_tangentials. FEFaceValues calls this function internally, then it calls the fill_fe_face_values. If you know what you are doing, you can use directly the mapping functions, which give you access to all the raw data which is computed, including tangential vectors and jacobians.
Luca. -- Luca Heltai <[email protected]> http://people.sissa.it/~heltai/ Scuola Internazionale Superiore di Studi Avanzati Phone: +39 040 3787 449, Office: 255 -- There are no answers, only cross references. On 21/apr/2010, at 10.28, Jean-Paul Pelteret wrote: > Hi Luca, > > I've finally got to the point where I am implimenting the contact element > that needs the covariant base vectors and their gradients. > > So I create an arbitrary point on the boundary cell face, create a quadrature > rule at that point and use it in conjunction with a FEFaceValues object to > get the normal, shape functions etc. at that point (BTW. I'm working in space > dim 3). In all, the UpdateFlags that I use are: > update_values | update_gradients | update_quadrature_points | > update_normal_vectors | update_jacobians | update_jacobian_grads > > I'm able to get some data such as the correct face normal, however I've > noticed that the jacobian tensor (and its gradient and inverse) that you > directed me towards do not get filled with any data - all zero's on output. > After some testing, I found that this information exists only for cells, and > not for faces . I looked at the mapping classes and it appears that they do > not store the information for the tensors if called on a face. > > Whats the easiest way to solve this issue? Is the best (and only?) way to > change the fill_fe_face_values and compute_fill_face methods? If so, can one > just copy the necessary lines from the equivalent cell calculations, or do > you think the calculations would change slightly when done on a face? Or am I > just plain wrong in my implementation? > > Thanks very much in advance for the help. > Best regards, > Jean-Paul _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
