| Author | Message | |||
     Yong Wang (yw3) New member Username: yw3 Post Number: 2 Registered: 06-2005 |
hi: I am using the student version of FlexPDE and I have an iterative algorithm in my mind. At the beginning of the iterative algorithm, part of the boundary conditions is unknown and zero value is used as initial guessing to solve the PDE(Laplace equation). In each iteration, the boundary condition will be updated according to the solution of previous iteration.Finally the accurate solution can be obtained. My questiones are: 1) How to implement iterations in FlexPDE ? The geometry is not changed. The same problem is solved under different boundary conditions in each iteration.Can I use For Loop? 2) How to update the boundary conditions by using the solution of previous iteration ? Thank you very much for your help! Yong | |||
| Yong Wang (yw3) Junior Member Username: yw3 Post Number: 3 Registered: 06-2005 |
After reading FlexPDE documents, I think my problems can be solved using Stage. But I still have a problem: In my algorithm, the type of the boundary condition for a boundary is changing from one stage to next stage. For example, for boundary 1, in the first stage, I put Dirichlet boundary condition on it. in the second stage, I put Neumann boundary condition on it. In the third stage, I put Dirichlet boundary condition on it again. and so on... THe boundary condition type is alternating. I do not know how to change the boundary condition type for each stage. ? Please give some help! THanks a lot! Yong | |||
| Yong Wang (yw3) Member Username: yw3 Post Number: 4 Registered: 06-2005 |
Hi: I did not find the direct way to implement above alternating boundary type. However I designed a short code to simulate the alternating changing of boundary type by using two state variables. I attached the pde file here. Unfortunately I do not know why it is not converging. I just began to use FlexPDE yesterday, so I do not have any experence and idea about this problem. Can some firend kindly go through the short code and help me to fix the problem? Thank you very much! Yong
| |||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 373 Registered: 06-2003 |
I'm not sure what you are trying to accomplish here. The principle of FlexPDE is that you don't try to describe the solution procedure. You describe the PDE system and let FlexPDE figure out the solution procedure. In keeping with this idea, you should just describe the mathematical statement of the PDE and boundary conditions. Unless your alternation of states is supposed to reflect an alternation that is going on in the physical system? ALTERING BOUNDARY CONDITIONS: FLexPDE will not let you change the class of boundary condition (from VALUE to NATURAL) during a run (even a staged run). You can pose a NATURAL BC that is effectively a VALUE BC by the following: NATURAL(u)=large_number*(U0-u) (Note that the sign of this term depends on the sign of the Laplacian in your equation. You want a restoring flux that pulls u to U0.) CONVERGENCE FAILURE: I don't know why this fails to converge. But if you separate the stages and run the second one as an independent problem, it converges. So the fact that as a second stage it is starting from the solution of the first stage has somehow created an indeterminate equation. But given that I don't understand why you are doing this alternation in the first place, I'm not going to pursue the issue right away. |
alternating boundary condition