On 12/11/18 12:50 AM, 'Maxi Miller' via deal.II User Group wrote:
> Still, that would require me to change every vector from
> TrilinosWrappers::MPI::Vector to LinearAlgebra::distributed::Vector.
> Is there any advantage of doing that? (And is there any advantage of using
> TrilinosWrappers comp
Still, that would require me to change every vector from
TrilinosWrappers::MPI::Vector to
LinearAlgebra::distributed::Vector. Is there any advantage of doing
that? (And is there any advantage of using TrilinosWrappers compared to
LinearAlgebra (i.e. the built-in functions from deal.II)?
Am Mon
Le lun. 10 déc. 2018 à 15:37, 'Maxi Miller' via deal.II User Group
a écrit :
>
> Do the deal.II-internal solvers work on Trilinos-MPI-Vectors? Or is there a
> way to "recreate" a trilinos-matrix in the same way as I did here?
deal.II solvers are templated on the matrix type, the vector type, and
Do the deal.II-internal solvers work on Trilinos-MPI-Vectors? Or is there a
way to "recreate" a trilinos-matrix in the same way as I did here?
Am Montag, 10. Dezember 2018 21:36:10 UTC+1 schrieb Bruno Turcksin:
>
> Le lun. 10 déc. 2018 à 15:27, 'Maxi Miller' via deal.II User Group
> > a écrit :
Le lun. 10 déc. 2018 à 15:27, 'Maxi Miller' via deal.II User Group
a écrit :
> LinearAlgebraTrilinos::SolverCG solver (solver_control);
You cannot use Trilinos solvers with your own matrix type. With
Trilinos solvers, you need to use a Trilinos matrix. You want to use
deal.II's own solvers wh
I tried to implement that (as in example 20) with a class for the matrix
class jacobian_approximation : public Subscriptor
{
public:
jacobian_approximation(std::function residual_function,
const MPI_Comm &mpi_communicator,
const IndexSet& d
Maxi,
On Tuesday, December 4, 2018 at 3:48:41 AM UTC-5, Maxi Miller wrote:
>
> What do you mean with "What I want to do with the matrix afterwards"? And
> I am not sure if I need it element-by-element, I just would like to
> implement a cheaper update method than the full recalculation.
>
If you
Maxi,
What do you mean with "What I want to do with the matrix afterwards"? And I
> am not sure if I need it element-by-element, I just would like to implement
> a cheaper update method than the full recalculation.
>
You can always obtain a matrix representation of a linear operator by
applyi
What do you mean with "What I want to do with the matrix afterwards"? And I
am not sure if I need it element-by-element, I just would like to implement
a cheaper update method than the full recalculation.
Am Dienstag, 27. November 2018 22:21:00 UTC+1 schrieb Wolfgang Bangerth:
>
> On 11/27/2018
On 11/27/2018 02:11 PM, 'Maxi Miller' via deal.II User Group wrote:
> I am trying to solve a system of nonlinear equations using the
> Newton-method as described in example 33. My problem is that the
> assembly of the newton matrix takes 71 % of the time of the calculations
> (for five iteration
I am trying to solve a system of nonlinear equations using the
Newton-method as described in example 33. My problem is that the assembly
of the newton matrix takes 71 % of the time of the calculations (for five
iterations). Thus I would like to calculate this matrix once (at the
beginning of ea
On 11/27/2018 11:22 AM, 'Maxi Miller' via deal.II User Group wrote:
> Wouldn't that give me a vector instead of a (sparse) matrix which I need
> afterwards?
Yes, of course. But as I've mentioned before, the matrix that you update
with your formula is dense. You can't store it, and you can't
eff
Wouldn't that give me a vector instead of a (sparse) matrix which I need
afterwards?
Am Mittwoch, 21. November 2018 15:57:08 UTC+1 schrieb Wolfgang Bangerth:
>
> On 11/21/18 5:39 AM, 'Maxi Miller' via deal.II User Group wrote:
> > Hmm, but in that case I have an addition (update of the sparse
>
On 11/21/18 5:39 AM, 'Maxi Miller' via deal.II User Group wrote:
> Hmm, but in that case I have an addition (update of the sparse jacobian),
> which I do not know how to handle (yet)
Well, if you also have a previous Jacobian matrix, say J, then you would
replace...
> void BroydenOperat
Hmm, but in that case I have an addition (update of the sparse jacobian),
which I do not know how to handle (yet)
Am Dienstag, 20. November 2018 23:07:17 UTC+1 schrieb Wolfgang Bangerth:
>
> On 11/20/18 2:27 PM, 'Maxi Miller' via deal.II User Group wrote:
> > how exactly can I understand that? I
On 11/20/18 2:27 PM, 'Maxi Miller' via deal.II User Group wrote:
> how exactly can I understand that? I have to generate a second matrix, which
> then will be added to the sparse matrix, but (due to the vector
> multiplication) will be a full matrix, something I would like to avoid.
No, you don'
Hei,
how exactly can I understand that? I have to generate a second matrix, which
then will be added to the sparse matrix, but (due to the vector multiplication)
will be a full matrix, something I would like to avoid.
Thanks!
--
The deal.II project is located at http://www.dealii.org/
For mail
On 11/19/18 1:18 AM, 'Maxi Miller' via deal.II User Group wrote:
> I would like to implement Broyden's method, to speed up the assembly of the
> jacobian when solving a nonlinear system, which results in an equation
> containing the difference of the residual, the difference of the solution and
I would like to implement Broyden's method, to speed up the assembly of the
jacobian when solving a nonlinear system, which results in an equation
containing the difference of the residual, the difference of the solution
and the current jacobian matrix, i.e.
J_{n+1}=J_n+\frac{\Delta f_n-J_n\Del
19 matches
Mail list logo