multiphysics

[asartori86]

Before start to think about the design we should have for handling different physics with different interfaces (an maybe different meshes) I would like to have clear what are the ingredients we need for handling such situations.

Let's have two comments with the list for the two different cases

[luca-heltai]

• different meshes (prototype: IFEM)
• different physics on the same mesh (prototype: step-32)
• different types (BEM-FEM)

[asartori86]

let's start with the easiest one: different physics on the same mesh (prototype: step-32)

• dof_handler?
• solution_vector?

[luca-heltai]

Case of different meshes:

• two-way access (it maybe enough to have friendship)
• a mechanism to construct coupling (this is also valid for the other case)

[luca-heltai]

I think the two are not too different. You still have two DH, two sets of explicit and implict vectors, etc. I think if we have two full pidomus interfaces, we should be good to go. Actually we could rename pi-DoMUS to SinglePhysicsProblem, and then construct the new pi-DoMUS as a collection of SinglePhysicsProblems.

[luca-heltai]

Let's discuss with Wolfgang about this...

:)