Two data sets of indefinite matrix:
  afiro: contributed by S. Benson, M=N=59, nz=225, ill-conditioned.
  golan: contributed by T. Munson, M=N=4303, nz=33157, cond. approx 1.e8.  

xxx_A.dat: matrix in ascii format
xxx_b.dat: rhs in ascii format
xxx_u.dat: true solution in ascii format (greated by Matlab dense solver)
xxx:       converted Petsc binary sparse format 
           (see `~petsc/matrices/readme' for details) 

/mat/examples/tests/ex78.c: 
     example on how to convert Matlab ascii format to Petsc binary form.
/sles/examples/tests/ex25.c: 
     example on loading binary data and solve the indefinite system

sequential output of ex25.c 
==============================
1. on data set afiro: 
   ("-pc_type jacobi" fails for all the three ksp methods 
     It is an "indefinite preconditioner")
=========
-ksp_max_it 150 -ksp_rtol 1.e-10 -pc_type none:
----------------------------------------------
 CG:
Number of iterations = 150
Residual norm: 0.616078;  Error norm: 1.78961.

 MINRES:
Number of iterations = 150
Residual norm: 0.0969619;  Error norm: 1.81172.

 SYMMLQ:
Number of iterations = 150
Residual norm: 0.216997;  Error norm: 1.66926. 

-ksp_max_it 70 -ksp_rtol 1.e-10 -pc_type jacobi -pc_jacobi_rowmax
----------------------------------------------------------------
 CG:
Number of iterations =  70
Residual norm: 0.0011258;  Error norm: 0.000230106.

 MINRES:
Number of iterations =  70
Residual norm: 3.39516e-05;  Error norm: 0.470816.

 SYMMLQ:
Number of iterations =  70
Residual norm: 0.00461956;  Error norm: 0.00202006.  
===================
2. on data set golan:
   (-pc_type jacobi fails as an "Indefinite preconditioner")
=====================
-np 1:
---------
-pc_type none:
---------------
  CG:
 Number of iterations = 10000
 Residual norm 7.69948;  Error norm 397.12.

  MINRES:
 Number of iterations = 10000
 Residual norm 0.0212476;  Error norm 390.082.

  SYMMLQ:
 Number of iterations = 10000
 Residual norm 2.94641;  Error norm 388.242. 

-pc_type jacobi -pc_jacobi_rowmax:
-----------------------------------
  CG:
 Number of iterations = 10000
 Residual norm: 1.24166;  Error norm: 3.0973.

  MINRES:
 Number of iterations = 10000
 Residual norm: 0.00976014;  Error norm: 236.398.

  SYMMLQ:
 Number of iterations = 10000
 Residual norm: 0.132509;  Error norm: 0.497156.