[petsc-users] unstructured parallel

[Hong Zhang]

> I have implemented PETSc for my unstructured FVM problem by external package
> Sundials, I want to move to a parallel version. I just read the example
> http://www.mcs.anl.gov/petsc/petsc-2/snapshots/petsc-current/src/dm/ao/examples/tutorials/ex2.c.html
>
> I want know if my understanding is right:
>
> DA is only for structured grid;
Yes, only for logically structured grids.

>
> There is not too much difference between
> http://www.mcs.anl.gov/petsc/petsc-2/snapshots/petsc-current/src/ts/examples/tutorials/ex3.c.html
> and
> http://www.mcs.anl.gov/petsc/petsc-2/snapshots/petsc-current/src/ts/examples/tutorials/ex4.c.html,
> because DA can manage most parallel related staff.

They solve same problem, but illustrate different TS solvers and approaches.
Although ex4.c uses DA in main() to create vectors and is intended to
illustrate a non-linear TS solver,
its function evaluation does use DA. ex4.c is not a good example for using DA.

Almost all petsc objects and solvers are parallel, which manage
parallel data distribution and implementation.
DA is one of objects.

Hong

>
> At last is there any example I can learn to get a parallel version of my
> problem.
>
>
>
> Thanks!
>
>