1: program main
2: !
3: ! This example intends to show how DMDA is used to solve a PDE on a decomposed
4: ! domain. The equation we are solving is not a PDE, but a toy example: van der
5: ! Pol's 2-variable ODE duplicated onto a 3D grid:
6: ! dx/dt = y
7: ! dy/dt = mu(1-x**2)y - x
8: !
9: ! So we are solving the same equation on all grid points, with no spatial
10: ! dependencies. Still we tell PETSc to communicate (stencil width >0) so we
11: ! have communication between different parts of the domain.
12: !
13: ! The example is structured so that one can replace the RHS function and
14: ! the forw_euler routine with a suitable RHS and a suitable time-integration
15: ! scheme and make little or no modifications to the DMDA parts. In particular,
16: ! the "inner" parts of the RHS and time-integration do not "know about" the
17: ! decomposed domain.
18: !
19: ! See: http://dx.doi.org/10.6084/m9.figshare.1368581
20: !
Binary file (standard input) matches