Examples

The code reported in Figure 4 shows how to set and apply the default multi-level preconditioner available in the real double precision version of MLD2P4 (see Table 1; we assume here that UMFPACK has been installed). This preconditioner is chosen by simply specifying 'ML' as second argument of mld_precinit (a call to mld_precset is not needed) and is applied with the BiCGSTAB solver provided by PSBLAS. As previously observed, the modules psb_base_mod, mld_prec_mod and psb_krylov_mod must be used by the example program.

The part of the code concerning the reading and assembling of the sparse matrix and the right-hand side vector, performed through the PSBLAS routines for sparse matrix and vector management, is not reported here for brevity; the statements concerning the deallocation of the PSBLAS data structure are neglected too. The complete code can be found in the example program file mld_dexample_ml.f90, in the directory examples/fileread of the MLD2P4 tree (see Section 3.5). For details on the use of the PSBLAS routines, see the PSBLAS User's Guide [17].

The setup and application of the default multi-level preconditioners for the real single precision and the complex, single and double precision, versions are obtained with straightforward modifications of the previous example (see Section 6 for details). If these versions are installed, the corresponding Fortran 95 codes are available in examples/fileread/.

Figure 4: Setup and application of the default multi-level Schwarz preconditioner.

  use psb_base_mod
  use mld_prec_mod
  use psb_krylov_mod
... ...
!
! sparse matrix
  type(psb_dspmat_type) :: A
! sparse matrix descriptor
  type(psb_desc_type)   :: desc_A
! preconditioner
  type(mld_dprec_type)  :: P
! right-hand side and solution vectors
  real(kind(1.d0))      :: b(:), x(:)
... ...
!
! initialize the parallel environment
  call psb_init(ictxt)
  call psb_info(ictxt,iam,np)
... ...
!
! read and assemble the matrix A and the right-hand side b 
! using PSBLAS routines for sparse matrix / vector management 
... ...
!
! initialize the default multi-level preconditioner, i.e. two-level
! symmetrized multiplicative Schwarz, using 1 sweep of RAS (with overlap 1
! and ILU(0) on the blocks) as smoother and 4 block-Jacobi sweeps (with 
! UMFPACK LU on the blocks) as distributed coarse-level solver
  call mld_precinit(P,'ML',info)
!
! build the preconditioner
  call mld_precbld(A,desc_A,P,info)
!
! set the solver parameters and the initial guess
  ... ...
!
! solve Ax=b with preconditioned BiCGSTAB
  call psb_krylov('BICGSTAB',A,P,b,x,tol,desc_A,info)
  ... ...
!
! deallocate the preconditioner
  call mld_precfree(P,info)
!
! deallocate other data structures
  ... ...
!
! exit the parallel environment
  call psb_exit(ictxt)
  stop

Different versions of multi-level preconditioners can be obtained by changing the default values of the preconditioner parameters. The code reported in Figure 5 shows how to set a three-level hybrid Schwarz preconditioner with post-smoothing only, which uses as smoother block-Jacobi with ILU(0) on the local blocks, has a coarsest matrix replicated on the processors, and solves the coarsest-level system with the LU factorization from UMFPACK [11]. The number of levels is specified by using mld_precinit; the other preconditioner parameters are set by calling mld_precset. Note that the type of multilevel framework (i.e. multiplicative among the levels with post-smoothing only) is not specified since it is the default set by mld_precinit.

Figure 6 shows how to set a three-level additive Schwarz preconditioner, which uses as smoother RAS, with overlap 1 and ILU(0) on the blocks, and applies five block-Jacobi sweeps, with LU from UMFPACK on the blocks, as distributed coarsest-level solver. Again, mld_precset is used only to set non-default values of the parameters (see Tables 2-5). In both cases, the construction and the application of the preconditioner are carried out as for the default multi-level preconditioner. The code fragments shown in in Figures 5-6 are included in the example program file mld_dexample_ml.f90 too.

Finally, Figure 7 shows the setup of a one-level additive Schwarz preconditioner, i.e. RAS with overlap 2. The corresponding example program is available in mld_dexample_ 1lev.f90.

For all the previous preconditioners, example programs where the sparse matrix and the right-hand side are generated by discretizing a PDE with Dirichlet boundary conditions are also available in the directory examples/pdegen.

Remark 3. Any PSBLAS-based program using the basic preconditioners implemented in PSBLAS, i.e. the point-Jacobi (diagonal) and block-Jacobi ones, can use the point-Jacobi and block-Jacobi preconditioners implemented in MLD2P4 without any change in the code. The PSBLAS-based program must be only recompiled and linked to the MLD2P4 library.

Figure 5: Setup of a hybrid three-level Schwarz preconditioner.

... ...
! set a three-level hybrid Schwarz preconditioner with post-smoothing
! only, which uses 1 sweep of block-Jacobi (with ILU(0) on the blocks)
! as smoother, a coarsest matrix replicated on the processors, and the 
! LU factorization from UMFPACK as coarsest-level solver
  call mld_precinit(P,'ML',info,nlev=3)
  call mld_precset(P,mld_smoother_pos_,'POST',info)
  call mld_precset(P,mld_smoother_type_,'BJAC',info)
  call mld_precset(P,mld_coarse_mat_,'REPL',info)
  call mld_precset(P,mld_coarse_solve_,'UMF',info)
... ...

Figure 6: Setup of an additive three-level Schwarz preconditioner.

... ...
! set a three-level additive Schwarz preconditioner, which uses
! 1 sweep of RAS (with overlap 1 and ILU(0) on the blocks) as
! smoother and 5 block-Jacobi sweeps (with UMFPACK LU on the
! blocks) as distributed coarsest-level solver
  call mld_precinit(P,'ML',info,nlev=3)
  call mld_precset(P,mld_ml_type_,'ADD',info)
  call mld_precset(P,mld_coarse_sweeps_,5,info)
... ...

Figure 7: Setup of a one-level Schwarz preconditioner.

... ...
! set RAS with overlap 2 and ILU(0) on the local blocks
  call mld_precinit(P,'AS',info)
  call mld_precset(P,mld_sub_ovr_,2,info)
... ...