hi all,
I'm trying to track down some changes to the energy decomp code in sander.
It looks to me like it got updated at some point (long ago, according to
Taisung) to support decomposition in TI. I'm trying to use it for energy
postprocessing, not TI, and am running into trouble.
When decpair is called in the energy routines, sometimes it uses
nstlim/ntpr in the decpair call (to average the data?), but sometimes not,
depending on icfe (this example in ene.F90):
if(icfe /= 0 .and. decpr) then
if(ifsc /= 0 .and. nsc(ii) /= 1 .and. nsc(jj) /= 1) then
call decpair(4,ii,jj,eaw(jn)/(nstlim/ntpr))
else if(ifsc == 0) then
call decpair(4,ii,jj,eaw(jn)/(nstlim/ntpr))
end if
else if(icfe == 0) then
call decpair(4,ii,jj,eaw(jn))
end if
but in other places it doesn't appear to check icfe at all (this from
ew_force.F90)
if(decpr .and. idecomp > 0) call
decpair(2,i,k,cgi*cgk*d0/(nstlim/ntpr))
Furthermore, sometimes the behavior is different depending on ifdef MPI...
So I'm wondering if anyone knows this code, or who made the changes, that
might be able to help me understand what it's doing and give me some
guidance as I try to make some changes that hopefully don't break any
existing intent.
thanks
Carlos
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Wed Nov 30 2022 - 14:30:02 PST
