[deal.II] Re: MeshWorker clarifications

Just one more question about MeshWorker. I am trying to figure how I could assemble simultaneously more than one form. I've found that step 39 uses two different assemblers. My guess is that I could use a MatrixSimple assembler for bilinear forms, and ResidualSimple for linear forms. Is this t

[deal.II] Re: MeshWorker clarifications

For the record, this contribution in the original post > ((i == j) ? > > (fe_v.shape_grad(component_i, q_point) * > > fe_v.shape_grad(component_j, q_point) * > > mu_values[q_point]) : > 0) > should have read For the record, this contribution in the original post > ((component_i == compo

[deal.II] Re: MeshWorker clarifications

It was all my mistake in copy/pasting the weak form. The two matrices now are the same. Sorry for the noise. -- 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

[deal.II] Re: MeshWorker clarifications

Daniel, I've run the problem with a simple GridGenerator::hyper_rectangle and the results differ. I'm filing a bug report for this. The norm of the matrix difference is 10^11, but at least the sparsity pattern remains the same, if this could be of any consolation! Should I look at something el

[deal.II] Re: MeshWorker clarifications

Franco, on the first glance your code looks good. Can you reduce the size of your problem to a single cell? Is there still a difference then? Best, Daniel Am Mittwoch, 8. März 2017 13:04:01 UTC+1 schrieb Franco Milicchio: > > I have made available the snippet for this example here >

[deal.II] Re: MeshWorker clarifications

I have made available the snippet for this example here with the possibility of using either the MeshWorker or not (just change the #define). As far as I see, the code is copied and pasted correctly. The right hand sides *are the same*, but *mat

[deal.II] Re: MeshWorker clarifications

I have implemented, as suggested by Daniel, another MeshWorker callback for the right hand side. The code is at the bottom, but I don't know, given the status of the MeshWorker documentation, if this is correct. I cannot say this by looking at the solutions, since matices, as I said, differ gre

[deal.II] Re: MeshWorker clarifications

Thanks for the answer, Daniel. On Monday, March 6, 2017 at 7:49:53 PM UTC+1, Daniel Arndt wrote: So my first question is, should I avoid using this class and implement >> parallel loops by hand (via TBB or other means)? >> > > "amandus"[1] is in fact based on MeshWorker. If you are trying to

[deal.II] Re: MeshWorker clarifications

Franco, First of all, I have read this question > , > > and I read that MeshWorker is sort of abandoned: > > the issue with MeshWorker is that when it was implemented, we didn't write