Miguel,
  In this example the DM is created initially only on proc. 0. Once the DM gets 
partitioned (DMNetworkDistribute()), another DM is created that has the 
appropriate node,edge,component info on each processor.

Shri

From: Miguel Angel Salazar de Troya 
<[email protected]<mailto:[email protected]>>
Date: Fri, 20 Feb 2015 13:10:40 -0600
To: "[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>
Subject: [petsc-users] DMNetworkSetSizes called by all processors

Hi

I was checking this example:

http://www.mcs.anl.gov/petsc/petsc-current/src/snes/examples/tutorials/network/pflow/pf.c.html

In line 443, the data is read by processor 0 and DMNetworkSetSizes is called in 
461 by all processors, since it is collective. However, the data read by 
processor 0 has not been broadcasted and numEdges and numVertices and have 
still a value of zero for the other processors. Does this mean that the other 
processors create a dummy DMNetwork and wait to receive their partition once 
the partitioning is called in line 508? Is this a standard procedure to create 
meshes in parallel?

Thanks
Miguel

--
Miguel Angel Salazar de Troya
Graduate Research Assistant
Department of Mechanical Science and Engineering
University of Illinois at Urbana-Champaign
(217) 550-2360
[email protected]<mailto:[email protected]>

Reply via email to