Hi, if you had a transient problem and wanted to perform several mesh adaptations based on the solution obtained at each timestep, would you need to set uold=uold (where uold is the old solution from the previous timestep) after each adaptation? This would surely resize uold such that it would fit the newly adapted mesh.
Hi, do you know if freeFEM allows mesh adaptation to multiple functions? I have a coupled-PDE problem with three unknowns and it would be good if there was a way to incorperate all three solutions into the the mesh adaptation process. Great video!
hi, as far as I know, it doesn't support such mesh adaptation. I don't expect other mesh manipulation interfaces such as MMG and TetGen support it neither. maybe you need to combine all the unknowns into one representative metric and then use it for performing the mesh adaptation.
you are absolutely right, the function is first mapped to the original mesh and then used for the adaptive refinement. that's the reason behind using a very fine mesh in example 3.
Very helpful video ! Thank you Mojtaba Can you please do a tutorial on the same topic (Adaptive mesh refinement) but in FEniCS ? I find it kinda more technical there, as we have to make good use of cell_markers etc. That would be really interesting !
![[FreeFEM 7] Visualization and input/output (IO) operations in FreeFEM](http://i.ytimg.com/vi/algfmHKNsX4/mqdefault.jpg)
![[FreeFEM 7] Visualization and input/output (IO) operations in FreeFEM](/img/tr.png)
![[FreeFEM 4] Finite element spaces and function approximation in FreeFEM](http://i.ytimg.com/vi/j4Z5Xb85dEs/mqdefault.jpg)
![[FreeFEM 8] Regions and sub-domains to define variable material properties and initial conditions](http://i.ytimg.com/vi/g43w0iZFE-c/mqdefault.jpg)
Hi, if you had a transient problem and wanted to perform several mesh adaptations based on the solution obtained at each timestep, would you need to set uold=uold (where uold is the old solution from the previous timestep) after each adaptation? This would surely resize uold such that it would fit the newly adapted mesh.
Yes, you need to do it every time you adapt the mesh so that FF can interpolate (map) the values from the old mesh to the new one.
Hi, do you know if freeFEM allows mesh adaptation to multiple functions? I have a coupled-PDE problem with three unknowns and it would be good if there was a way to incorperate all three solutions into the the mesh adaptation process. Great video!
hi, as far as I know, it doesn't support such mesh adaptation. I don't expect other mesh manipulation interfaces such as MMG and TetGen support it neither. maybe you need to combine all the unknowns into one representative metric and then use it for performing the mesh adaptation.
is it my eyes or the adaption is performed on the first rough mesh not on the geometry. The fine mesh is following the coarse mesh outlines?
you are absolutely right, the function is first mapped to the original mesh and then used for the adaptive refinement. that's the reason behind using a very fine mesh in example 3.
Very good series on FreeFEM. Is is possible to apply FreeFEM to a real world process with varying boundary conditions from measurements?
thanks!
yes, it's relatively easy to define a variable BC in FreeFEM.
@@TuxRiders will you show it in a video?
@@michaelhecht5429 yes, sure.
Thank you
You're welcome :)
Very helpful video ! Thank you Mojtaba
Can you please do a tutorial on the same topic (Adaptive mesh refinement) but in FEniCS ? I find it kinda more technical there, as we have to make good use of cell_markers etc. That would be really interesting !
glad to know it was useful. it's indeed a great suggestion. yes, doing that is more challenging in FEniCS.
cool
:)