Dear All, For example, if I set up as below :
TrilinosWrappers::MPI::BlockVector a; TrilinosWrappers::MPI::BlockVector b; TrilinosWrappers::MPI::BlockVector c; and in the part of usage : a.reinit (partitioner); b.reinit (partitioner); c.reinit (partitioner); Solving a & b...... and then c.add (1,a,-1,b); double d = c.l2_norm(); ------------------------------------------- Dose the value of "d" keeps the "l2 norm" of the whole of size of vector from being distributed on each nodes? or just having the "l2 norm" in which each node has? I am trying to evaluate the l2 norm with mpi trilinos version. but, get the different value when single processor and 2 or more processors. Simply, How could we get the l2 norm with trilinos mpi vector/block_vector? _______________________________________________
