Dear All,
I still have some problem in solving my equations on a curved domain
(sphere in 3D or circle in 2D). I got error with this content:
invalid initialization of reference of type ‘dealii::Triangulation<2, 3>&’
from expression of type ‘dealii::Triangulation<3, 3>’
Does it mean that I should
Hi all,
I just made some changes in the second parameter of SolverControl in the
following way which apparently resolved the problem:
my problem is nonlinear so I solve it by newton method. It seems that in
the first Newton-Rophson iteration the CG solver can not converge if the
tolerance for
In fact, I should solve my problem on a sphere (circle). According to your
noteworthy comments, the only thing that I have to do is change the geometry
and use higher order mapping for accuracy. Another thing is that consider a
half of a circle in r-z coordinate system and the problem is axisymmet
Yes exactly. The only difference is in weak form of my equation which
should be multiplied by 2*pi*r which I define r_value =
fe_values_eta.quadrature_point(q_point)[0];
In fact, I should solve my problem on a sphere (circle). According to your
noteworthy comments, the only thing that I have to
> Yes You are right. I mean curved domain. My question is should i use codim
> instead of mapping because of the fact that there is only two tutorials in
> dealii which solved a differential equation on c curved domain according to
No, codim problems solve systems on surfaces embedded in a higher
Yes You are right. I mean curved domain. My question is should i use codim
instead of mapping because of the fact that there is only two tutorials in
dealii which solved a differential equation on c curved domain according to
the fact that my derived weak expressions are in cylindrical coordinat
> Let me put it in a such way. I want to solve my equation on a curved edge.
> According to dealii library, I should use MappingQ whenever I have nonlinear
> domain. Am I correct?
I don't think "nonlinear" is the word you are looking for. Do you mean
a curved boundary?
> In addition, Should I rep
Dear Jean-Paul,
Let me put it in a such way. I want to solve my equation on a curved edge.
According to dealii library, I should use MappingQ whenever I have
nonlinear domain. Am I correct? In addition, Should I replace all dim in
defining my tensors to spacedim? It should be noted that dim = sp
2016-10-19 11:51 GMT-04:00 Hamed Babaei :
> I need to compute the determinant of system_matrix to check if its positive
> or negative so that I know my system_matrix is positive definite or not.
That's what I thought. You don't need to know the determinant to do
that, you can check the sign of the
I need to compute the determinant of system_matrix to check if its positive
or negative so that I know my system_matrix is positive definite or not.
On Wednesday, October 19, 2016 at 10:38:09 AM UTC-5, Bruno Turcksin wrote:
>
> Hamed,
>
> On Wednesday, October 19, 2016 at 10:25:37 AM UTC-4, Hamed
Hamed,
On Wednesday, October 19, 2016 at 10:25:37 AM UTC-4, Hamed Babaei wrote:It
seems that I've been choosing wrong solver for my problem. Since my problem
has some sort of instability in its nature,
>
> although my system_matrix is symmetric, it is not necessarily positive
> definite. So I
Hi everyone,
Thank you all for your incredible guides.
It seems that I've been choosing wrong solver for my problem. Since my
problem has some sort of instability in its nature,
although my system_matrix is symmetric, it is not necessarily positive
definite. So I have two questions:
First, Is
Dear all
I am doing this
const FE_DGQArbitraryNodes fe(QGauss<1>(degree+1));
filter_matrix.reinit(fe.dofs_per_cell, fe.dofs_per_cell);
const FE_DGQArbitraryNodes fe1(QGauss<1>(degree));
FETools::get_back_interpolation_matrix(fe, fe1, filter_matrix);
but this give
Well, thinking about it, it seems that spacedim=3 is not relevant here. I
can do
const FE_DGQArbitraryNodes fe1(QGauss<1>(fe.degree+1));
const FE_DGQArbitraryNodes fe2(QGauss<1>(fe.degree));
FETools::get_back_interpolation_matrix(fe1, fe2, filter_matrix);
This should give me
Dear Josh
Many thanks for your reply. I personally have some umat written in Fortran
(i.e Von Mises, DP, ...) and because I start to learn dealii in depth, I
will write interface to use that in the code anyway. Therefore I think my
work can be useful for others. I'm glad that you and others are ta
15 matches
Mail list logo
