Some problems by using Mrbayes with MPICH2

Question

[Ginsea Chen]

Hello everyone

I tried to use Mrbayes with MPI (MPICH2) on ubuntu, as the describe of web: https://sourceforge.net/p/mrbayes/bugs/1599/, I added propset ExtTBR$prob=0 before mcmc. The results were not consisted with using Mrabyes on SUSE (MPICH2), although tree topologies were not unresolved.

By comparing two trees which from different linux server, we found that tree topologies of SUSE were better than Ubuntu, e.g some polytomy topologies were existed in tree of Ubuntu but not in SUSE.

Ubuntu server is more expedient than SUSE for us, so we want to know if there are some ways to solve it.

Thanks.

My command on Ubuntu:

begin mrbayes;
set autoclose=yes nowarn=yes;
execute cl0044_3.nex;
lset rates=invgamma;
prset aamodelpr=fixed(wag) statefreqpr=fixed(empirical);
mcmcp nruns=1 nchains=8 ngen=1000000 checkpoint=yes checkfreq=1000 file=cl0044_3.nex;
propset ExtTBR$prob=0;
mcmc append=yes;
sumt;
end;

My command on SUSE:

begin mrbayes;
set autoclose=yes nowarn=yes;
execute cl0044_3.nex;
lset rates=invgamma;
prset aamodelpr=fixed(wag) statefreqpr=fixed(empirical);
mcmc nruns=1 nchains=8 ngen=1000000 checkpoint=yes checkfreq=1000 file=cl0044_3.nex;
sumt;
end;

The version of MPICH2 is 1.5.