## [Libmesh-users] How to solve this problem?

 [Libmesh-users] How to solve this problem? From: John Peterson - 2005-03-07 17:19:06 ```nelson writes: > hello all, > i have to solve a nonlinear reaction diffusion equation. To not use a > nonlinear solver I have to use a method in which every element have to > know the gradient of the FE space in his nodes (i use Tet4 elements) and > also know some parameter specific scalars. I have also to have the > volume of the element (which i have already done)... > > How can I do it? How can make the element aware of the problem i am > solving and of the value of the basis function on his nodes? Try looking at example 13. It shows how to compute the solution and solution gradients at the quadrature points using the basis functions. Note that for integration, you typically want to evaluate your "parameter specific scalars" at the quadrature points for purposes of integration, not the nodes. Cheers, John ```

### Thread view

 [Libmesh-users] How to solve this problem? From: nelson - 2005-03-05 08:48:14 ```hello all, i have to solve a nonlinear reaction diffusion equation. To not use a nonlinear solver I have to use a method in which every element have to know the gradient of the FE space in his nodes (i use Tet4 elements) and also know some parameter specific scalars. I have also to have the volume of the element (which i have already done)... How can I do it? How can make the element aware of the problem i am solving and of the value of the basis function on his nodes? Thank you, nelson ```
 [Libmesh-users] How to solve this problem? From: John Peterson - 2005-03-07 17:19:06 ```nelson writes: > hello all, > i have to solve a nonlinear reaction diffusion equation. To not use a > nonlinear solver I have to use a method in which every element have to > know the gradient of the FE space in his nodes (i use Tet4 elements) and > also know some parameter specific scalars. I have also to have the > volume of the element (which i have already done)... > > How can I do it? How can make the element aware of the problem i am > solving and of the value of the basis function on his nodes? Try looking at example 13. It shows how to compute the solution and solution gradients at the quadrature points using the basis functions. Note that for integration, you typically want to evaluate your "parameter specific scalars" at the quadrature points for purposes of integration, not the nodes. Cheers, John ```
 Re: [Libmesh-users] How to solve this problem? From: nelson - 2005-03-08 16:54:04 ```On lun, 2005-03-07 at 11:17 -0600, John Peterson wrote: > > Try looking at example 13. It shows how to compute the solution > and solution gradients at the quadrature points using the basis > functions. Note that for integration, you typically want to > evaluate your "parameter specific scalars" at the quadrature > points for purposes of integration, not the nodes. > > Cheers, > John My problem is that for every node I have to do a loop over all the elements and, for every element that shares the node, compute a quantity dependent from the volume of the element i'm looping on, on the value of the gradient of the local basis function, and on the value of the solution at the last timestep. Is example 13 the right one to look at? thank you, nelson ```