Dear Yves, Thank you for your reply!
We expressed the divergence of stress tensor with respect to the variable u. In the expression, we have something like Grad(Trace(Grad_u)), and Trace(Hess_u). When assembling these terms with GetFEM, there is an error in assembly string. I tried to assembly separately these two terms. Both did not work. For the first term Grad(Trace(Grad_u)), I think the error comes from the fact that, in GetFEM, we can only use Grad operator with a variable, i.e., Grad_u, Grad_Test_u, etc. Therefore, we can not write directly Grad(Trace(Grad_u)). Is it right? For the second term Trace(Hess_u), I think the error comes from the Trace operator, since in the GetFEM documentation, it is stated that Trace operator is used for a square matrix. In our case, since u is a vector field, Hess_u is a tensor of order 3. Can we use Trace operator in assembly string for a tensor of order 3? Thank you in advance for your reply. Best regards, Phuoc On Tue, Jan 16, 2018 at 5:32 PM, Yves Renard <[email protected]> wrote: > > Dear Phuoc, > > There is no specific function for that, but you have access to the Hessian > ('Hess_u' for the variable u), so you have to give the expression of the > divergence of the stress tensor with respect to the Hessian of the variable. > > Additionally, for the jumps on normal derivatives on element interfaces > see http://getfem.org/userdoc/gasm_high.html#evaluating- > discontinuities-across-inter-element-edges-faces > > Best regards, > > Yves > > > > Le 16/01/2018 à 06:08, Huu Phuoc BUI a écrit : > > Dear getfem users, > > I'd like to compute the divergence of stress tensor of a linear elasticity > problem which is necessary in an a posteriori error estimate. > > Do you have any idea how to compute it using generic assembly of GetFEM? > > Thank you > in advance for your support. > > > Best regards, > Phuoc > > > -- > > Yves Renard ([email protected]) tel : (33) 04.72.43.87.08 > <04%2072%2043%2087%2008> > Pole de Mathematiques, INSA-Lyon fax : (33) 04.72.43.85.29 > <04%2072%2043%2085%2029> > 20, rue Albert Einstein > 69621 Villeurbanne Cedex, FRANCE > http://math.univ-lyon1.fr/~renard > > --------- > >
