I have a question about solving problem defined on two subdomains. Namely, I have a system of equations consisting of a Poisson equation and several drift-diffusion equations (number varies depending on a model, but its always more than 4). Poisson equation is defined on the whole domain \Omega: \…

Yes, it’s possible to create submeshes corresponding to the part of the domain your equations belong to, and to couple them in the mixed formulation. As a (very basic) non-linear example, you can take a look to this demo

Hello, Regarding the use of the mixed-dimensional branch, I would suggest building your mesh using an external software such as Gmsh. This post gives more details on how to convert a Gmsh mesh using meshio and how to import it into FEniCS using xdmf format.

Hi, Thank you for a suggestion. I already use gmsh, but for small tests I usually use mshr. However, if it is not available, it won’t be a problem. Concerning the other question, is it possible to solve the problem, where one equation is defined over a whole domain, while the others are defined on…

Thank you for the example, I checked it out and it seems that it can be easily adapted to my problem. I have couple more questions concerning the surface integrals and custom solvers in a mixed-dimensional branch. Concerning solvers, is it possible to use a custom made nonlinear solvers, such as on…

Concerning solvers, there might still be some issues with PETSc SNES. I recommend using a NewtonSolver instead. Also you could try to define a customized NewtonSolver as in the post you mention (I haven’t tried doing that). Regarding the marking of the boundaries, you can use MeshFunction as usual …

Thank you very much for your help. I think that mixed-dimensional branch solves my problem.

I am sorry for bumping the thread, however I have a problem with a solver. Namely, after implementing changes (required by mixed-dimensional branch) to my original code, the program crashes with an error stating: Unable to successfully call PETSc function 'KSPSolve. Since the original code relies on…

Hi, I have push minor fixes in the mixed-dimensional branch that might fix your issue (the Docker container has been updated accordingly, tag latest). Also, mshr should now be available in the Docker container.

Thank you for the changes, MWE now works. However, when I try to solve my original problem, I get the following PETSc error *** Error: Unable to successfully call PETSc function 'MatSetValuesLocal'. *** Reason: PETSc error code is: 63 (Argument out of range). *** Where: This error was encounte…

Can you give more details on what is different between the previous MWE and the original problem ? The geometry I suppose, is there a difference in term of variational formulation ? The error you get seems to happens during the assembly, and if so I don’t think using SNES would make any difference …

Yes, you are right, the geometry is different. In the original problem, the geometry is as in the first post. In term of variational formulation, the code is basically the same. The only thing that I can think of that could cause the error is a mistake in defining surface integrals in a weak form. I…

Coupling the equations defined on two different subdomains

variational formulation

Sabarish_Gopi April 11, 2020, 6:00pm 19

For those who might face such an issue anytime in future, please check the link below:

1 Like

Topic		Replies	Views
Solve poisson equation with three submeshes dolfinx	12	353	May 8, 2025
Solve Poisson equation with non-linear boundary conditions on a submesh variational formulation	14	2051	April 13, 2020
Mixed finite element for two fields one defined on whole mesh and the other defined on submesh General	11	366	July 2, 2024
Drift-diffusion problem involving submeshes variational formulation	0	611	July 10, 2022
Diffusion coupled with boundary equation (mixed-dimensional) variational formulation	8	1346	November 25, 2022

Coupling the equations defined on two different subdomains

Related topics