[petsc-users] Understanding "indices" in ISLocalToGlobalMappingGetInfo

[Jed Brown]

On Sat, 13 Mar 2010 13:56:11 -0300, Pedro Torres <torres.pedrozpk at gmail.com> wrote:
> I call ISLocalToGlobalMappingGetInfo and I get the following information:
> 
> *Process 0:*
> 
> nproc = 2  procs [0 1]  numprocs = [13 13]
> indices = [0 1 2 3 5 6 10 11 17 18 19 20 28;
>                   0 1 2 3 5 6 10 11 17 18 19 20 28]
> 
> *Process 1:*
> nproc = 2  procs [1 0]  numprocs = [13 13]
> indices = [77 79 87 88 94 95 99 100 101 102 103 104 105;
>                    77 79 87 88 94 95 99 100 101 102 103 104 105]
> It seems that the indices in process 1 are correct, but not in the process
> 0.  Are these results correct?

The indices are with respect to the local numbering.  So rank 0's last
shared entry is (global index) 121 which has local index 28.  Maybe it
would be clearer if the docs read

  local indices of nodes shared with neighbor (sorted by global numbering)

instead of the present

  indices of local nodes shared with neighbor (sorted by global numbering)

Can you think of a better way to word it?

Jed