Re: [deal.II] Preconditioners not lowering the number of GMRES iterations

2023-04-25 Thread Wolfgang Bangerth
Lucas, I'm trying to precondition my system which can be solved with GMRES (with increasing iteration number for increasing size) but the standard preconditioners are either increasing the number of iterations, or causing the solver not to converge. Preconditioner design is difficult. This

Re: [deal.II] Form a BlockSparseMatrix based on block matrices from a different BlockSparseMatrix

2023-04-25 Thread Tao Jin
Dear Wolfgang, Thank you so much for your guidance. Step-35 is the example where "SynchronousIterators" is used. "" using IteratorTuple = std::tuple::active_cell_iterator, typename DoFHandler::active_cell_iterator>; using IteratorPair = SynchronousIterators;

Re: [deal.II] CG Solver with own VectorType

2023-04-25 Thread Stefan Karch
Thanks for your quick response. I was motivated by the lecture to try to implement my own vector structure, but it wasn't at first clear for me that it requires way more effort than a vmult operation. (But of course it makes sense.) I tried this to get more used to the solver/vector/matrix stru

[deal.II] Preconditioners not lowering the number of GMRES iterations

2023-04-25 Thread Lucas Myers
Hi everyone, TL;DR: I'm trying to precondition my system which can be solved with GMRES (with increasing iteration number for increasing size) but the standard preconditioners are either increasing the number of iterations, or causing the solver not to converge. Details: I'm trying to solve a

Re: [deal.II] Integrate on local subsets of domain on distributed triangulation

2023-04-25 Thread Lucas Myers
Hi Wolfgang, This is extremely helpful, thanks so much! - Lucas On Tuesday, April 25, 2023 at 11:24:31 AM UTC-5 Wolfgang Bangerth wrote: > On 4/25/23 09:44, Lucas Myers wrote: > > > > Re: FEPointEvaluation and RemotePointEvaluation, I think I actually need > > more than that because I need th

Re: [deal.II] Integrate on local subsets of domain on distributed triangulation

2023-04-25 Thread Wolfgang Bangerth
On 4/25/23 09:44, Lucas Myers wrote: Re: FEPointEvaluation and RemotePointEvaluation, I think I actually need more than that because I need the quadrature points on each of the cells that I'm querying, so I can just do the standard procedure with FEValues and the corresponding Quadrature obje

Re: [deal.II] CG Solver with own VectorType

2023-04-25 Thread Wolfgang Bangerth
On 4/25/23 07:15, Stefan Karch wrote: I am trying to implement a CG Solver with my own VectorType. According to Lecture 21, what one needs is that the Vector and Operator are compatible, and that the operator has a vmult-function. However, the compiler misses a lot of member function of my own

Re: [deal.II] Integrate on local subsets of domain on distributed triangulation

2023-04-25 Thread Lucas Myers
Hi Wolfgang, Thanks for the help! Point taken about neighboring processes and communication latency. Re: FEPointEvaluation and RemotePointEvaluation, I think I actually need more than that because I need the quadrature points on each of the cells that I'm querying, so I can just do the standar

[deal.II] CG Solver with own VectorType

2023-04-25 Thread Stefan Karch
Hello everyone, I am trying to implement a CG Solver with my own VectorType. According to Lecture 21, what one needs is that the Vector and Operator are compatible, and that the operator has a vmult-function. However, the compiler misses a lot of member function of my own Vector class. Am I doin