I understand what you meant but I have no idea how to fix it yet.(Maybe
because I lack experiences....)
How do I designate mapping method specifically?
Can I see examples of specific command lines to for mapping designation?
The original initialization of fe was follow.
template <int dim>
StokesProblem<dim>::StokesProblem (const unsigned int degree)
:
degree (degree),
*fe (FE_Q<dim>(degree+**1), dim,*
* FE_Q<dim>(degree), 1),*
dof_handler (triangulation)
{}
I tried to fix red lines into *fe(FE_Q<dim>
QGaussLobatto<1>(degree+1)),FE_Q<dim> QGaussLobatto<1>(degree))),*
template <int dim>
void StokesProblem<dim>::assemble_system ()
{
system_matrix=0;
system_rhs=0;
* QGauss<dim> quadrature_formula(degree+**2);*
FEValues<dim> fe_values (fe, quadrature_formula,
update_values |
update_quadrature_points |
update_JxW_values |
update_gradients);
........
template <int dim>
void StokesProblem<dim>::compute_drag ()
{
const long double pi = 3.141592653589793238462643;
*QGauss<dim-**1> quadrature_formula_face(degree+2);*
........
2016년 9월 20일 화요일 오전 9시 45분 0초 UTC-5, Martin Kronbichler 님의 말:
>
> Dear Jaekwang,
>
> if you are using the most recent developer version, you should
> automatically get the Gauss-Lobatto version of the node distribution. We
> made those point distributions the default for FE_Q(degree) this spring.
>
> Regarding the limits of quadrature formulas: Depending on what exactly you
> are doing, I would expect that the accuracy decreases as soon as you go
> beyond degree 10 or so. I don't think that your problem is the accuracy
> issues in terms of roundoff, because those issues would appear first as you
> go beyond 1e-10.
>
> Did you check the suggestions regarding the mapping? (I.e., you put a
> "Mapping<dim>" argument to all the constructors of FEValues, FEFaceValues,
> interpolate_boundary_values, etc.) Are the solver tolerances tight enough?
> What happens if the mesh is refined? If this still does not help, it might
> be good to share a small example where the issue is observed.
>
> Best,
> Martin
>
>
> On 09/20/2016 04:35 PM, JAEKWANG KIM wrote:
>
> thank for the reply!!
>
> my fe degree is declared as "fe (FE_Q<dim>(degree+1), dim,FE_Q<dim>(degree),
> 1)" and I used "QGauss<dim> quadrature_formula(degree+2);" to calculate
> integral over the cell.
>
> 2016년 9월 20일 화요일 오전 9시 26분 6초 UTC-5, Praveen C 님의 말:
>>
>>
>> On Tue, Sep 20, 2016 at 7:49 PM, JAEKWANG KIM <jaekw...@gmail.com> wrote:
>>
>>> Can I ask more about "the limits of our implementation of our quadrature
>>> formulas?".
>>> I wonder when it usually happens.
>>>
>>>
>>> Once I calculated drag coefficient, with Q1, my error is 10% compare to
>>> exact solution.
>>> However, I can 1.7% error when I use Q2 which is significant decrease!
>>> At Q3, I get 0.4% and At Q4 it starts to increase again 0.5%.
>>> If I go higher, then my error is more than 100%..... I really want to
>>> figure out why this happens....
>>>
>>> To summarize
>>> From Q1~Q3... it shows significant decrease in error
>>> but it is not anymore at Q4 and Q5
>>>
>>
>> Hi
>>
>> If you are using FE_Q space with uniformly spaced support points, then
>> there could be problem at higher degrees. Just to check, you should use
>> Gauss nodes, e.g.
>>
>> FE_Q(QGaussLobatto<1>(degree+1))
>>
>> Best
>> praveen
>>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see
> https://groups.google.com/d/forum/dealii?hl=en
> ---
> You received this message because you are subscribed to the Google Groups
> "deal.II User Group" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to dealii+un...@googlegroups.com <javascript:>.
> For more options, visit https://groups.google.com/d/optout.
>
>
>
--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see
https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
