Yury,

first, adaptive multilevel methods only work with discontinuous elements 
right now.

About your second question: step-16 in the previous versions solved the 
Neumann problem, but for the equation

-Du + u = f

The best way to make this work might be to add (a u,v) to the multilevel 
matrices and use standard components. If a is small enough, it should 
not destroy convergence.

Best,
Guido

Yury Petrov wrote:
>> Hi,
>>
>> I have two questions:
>> 1. Are the multilevel FEM method and the adaptive refinement  
>> compatible (step-6 + step-16)?
>> 2. What needs to be done to adapt tutorial step-16 for Neumann  
>> boundary conditions? Step-11 suggests using  
>> DoFTools::extract_boundary_dofs() to set the average boundary value  
>> to 0, but there is no extract_boundary_dofs() for MGTools class.  
>> Also, is the average value zeroing necessary in the  
>> assemble_multigrid() function?
>>
>> Thank you,
>> Yury
>>     
>
> _______________________________________________
>   


_______________________________________________

Reply via email to