In addition what Wolfgang said:
1. It would be indeed interesting to
see whether neglecting z_h really yields
the same error, the same effectivity indices,
and/or the same mesh.
2. Going back to your initial questions:
Inserting z_h is the key when classical
a posteriori bounds in terms of the mesh size h
are of interest.
The final goal is usually to obtain an error estimate
in terms of the mesh size h in order to
quantify the order of convergence of your scheme.
For this reason you need to insert z_h such that
|| z - z_h||
to apply interpolation estimates that give you
some h^{a} on the right hand side:
|| z-z_h || = O(h^a)
with the order a.
3. In practice you have indeed different choices
how to evaluate J(u) - J(u_h).
Also some people do not integrate back
into the strong form and work
with a weak form of the error estimator, which
has the advantage that no second-order operators
and partial integration needs to be applied.
From this point of view, it may be that neglecting
z_h could work in practice. But as said above,
a careful study for some model problems
would be useful.
Best Thomas W.
--
++--------------------------------------------++
Prof. Dr. Thomas Wick
Institut für Angewandte Mathematik (IfAM)
Leibniz Universität Hannover
Welfengarten 1
30167 Hannover, Germany
Tel.: +49 511 762 3360
Email: thomas.w...@ifam.uni-hannover.de
www: http://www.ifam.uni-hannover.de/wick
www: http://www.cmap.polytechnique.fr/~wick/
++--------------------------------------------++
--
On 03/10/2018 11:03 AM, Wolfgang Bangerth wrote:
On 03/06/2018 08:40 AM, 曾元圆 wrote:
Now I understand why we need to rewrite the error formula on a cell
as residual times dual weight. But I'm still a little confused with
the reason why we must introduce z_h.
Just as you mentioned, if we introduce z_h, then z-z_h is a quantity
that is only large where the dual solution is rough. But why do we
need to care about the accuracy of z here? I think the only thing we
need to care about is the value of z on that cell, because z is a
quantity that represents how important the residual on that cell is.
No. z tells you how important the *locally generated error is for the
global error functional*. (That is because z is the Green's function
associated with your error functional.) But you don't have the local
error. All you have is the local residual.
My understanding is: now the dual_weight z-z_h does not only
represent how important the residual on a certain cell is, but also
tells us some information about how good the dual solution on that
cell is. But another problem is, does z-z_h still has the same
tendency as z?
Almost. Think of it as z-phi_h where you can choose phi_h as you want.
For example, on each cell you can think of choosing phi_h so that it
cancels the constant and linear term of the Taylor expansion of z.
Then z-z_h would contain the quadratic and higher order Taylor terms,
i.e. something like z''*(x-x0)^2 where x0 can be chosen as a point on
the cell.
If not, how z-z_h can represent the importance of a certan cell as z
can?
I'm not sure if my understanding is correct. I tried to run the code
using only z as dual_weights, and I found the result almost the same
as that using z-z_h.
Nice idea to try this out. Do you get "almost the same" overall error
estimate, or "almost the same" mesh?
I think all of these are good questions to ask. Although I have worked
on this for a long time, I can not actually give you a particularly
good answer for all of this. I am sure others who are more versed in
the theory of errors, residuals, etc could tell you the precise reason
for why it is in fact necessary to subtract z_h. The best I can say is
that that's the way I've always seen it done, and while I have a vague
idea why that is so (see above), I can't say that I can describe it
well enough to explain it.
Finally, I am certainly glad to submit patches to deal.II and make my
own contribution. But I didn't fork deal.ii on my github account yet,
and this is relatively a small issue, so I will be glad if you can do
it for the moment.
OK, I will take care of this then.
Best
W.
--
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.