Define the number of iterations for the Newton-Raphson algo

You can use:

 solve(F == 0, u, bc, J=J,
       form_compiler_parameters=ffc_options,
       solver_param=dict(newton_solver=dict(maximum_iterations=20)))

to get a full view of the parameters you can set you can look at the default
parameters for a NonlinearVariationalProblem

  solver_params = NonlinearVariationalProblem.default_params()
  info(solver_params, 1)
  solver_params.newton_solver.maximum_iterations = 20
  solve(F == 0, u, bc, J=J,
       form_compiler_parameters=ffc_options,
       solver_param=solver_params)

Johan

On Tuesday August 23 2011 07:25:54 Nicolas Verdon wrote:
> New question #168953 on DOLFIN:
> https://answers.launchpad.net/dolfin/+question/168953
>
> Hi,
>
>
> I've a problem.
> I am using the nonlinear solver from dolfin from the example of
> hyperelasticity
>
> solve(F == 0, u, bc, J=J,
> form_compiler_parameters=ffc_options)
>
> When I change the parameters, I have a problem of convergence of the Newton
> algorithm which seems to be fixed at a maximum number of iterations =10
>
> I would simply know the syntax for setting up the tolerance, and the max
> number of iterations in the nonlinear solver.
>
> Thanks in advance
>
>
> Nicolas Verdon

I have now updated the docstring of solve so it should be easier to recognize
the available parameters to the underlying VariationalSolvers.

Johan

On Tuesday August 23 2011 07:25:54 Nicolas Verdon wrote:
> New question #168953 on DOLFIN:
> https://answers.launchpad.net/dolfin/+question/168953
>
> Hi,
>
>
> I've a problem.
> I am using the nonlinear solver from dolfin from the example of
> hyperelasticity
>
> solve(F == 0, u, bc, J=J,
> form_compiler_parameters=ffc_options)
>
> When I change the parameters, I have a problem of convergence of the Newton
> algorithm which seems to be fixed at a maximum number of iterations =10
>
> I would simply know the syntax for setting up the tolerance, and the max
> number of iterations in the nonlinear solver.
>
> Thanks in advance
>
>
> Nicolas Verdon

Ok thanks for all your answers !

I have an additional question concerning this example of hyperelasticity..
In fact I just would like to evaluate the function F defined in the example as the derivative of the potential energy Pi

Pi = psi*dx - dot(B, u)*dx - dot(T, u)*ds
F = derivative(Pi, u, v)

at a given vector u1..

I don't know if it's possible and once again, thank you in advance for your fast reactivity !

Nicolas Verdon

Niclas,
you do that simply by assigning the value of u1 to u before assembling F. F
is still defined in terms of u and keeps a reference to that concrete
function object.

Martin
Den 23. aug. 2011 20.05 skrev "Nicolas Verdon" <
<email address hidden>> følgende:
> Question #168953 on DOLFIN changed:
> https://answers.launchpad.net/dolfin/+question/168953
>
> Nicolas Verdon posted a new comment:
> Ok thanks for all your answers !
>
> I have an additional question concerning this example of hyperelasticity..
> In fact I just would like to evaluate the function F defined in the
example as the derivative of the potential energy Pi
>
> Pi = psi*dx - dot(B, u)*dx - dot(T, u)*ds
> F = derivative(Pi, u, v)
>
> at a given vector u1..
>
> I don't know if it's possible and once again, thank you in advance for
> your fast reactivity !
>
>
> Nicolas Verdon
>
> --
> You received this question notification because you are a member of
> DOLFIN Team, which is an answer contact for DOLFIN.

Thanks Chaffra Affouda, that solved my question.