On Wed, Nov 28, 2012 at 12:46 AM, colvin <colvin4367.gmail.com> wrote:
> Thank you Jason. I already changed the value accordingly but i'm not quite
> sure what do you mean about "change the
> value of BC_PARMR from 24560 to the sum of all array lengths in the
> rparms common block". Im sorry, could you pls explain more? Many thanks.
>
The relevant part of parms.F90 is shown below:
_REAL_ rk(MAX_BOND_TYPE),req(MAX_BOND_TYPE),tk(900),teq(900),pk(1200), &
pn(1200),phase(1200),cn1(1830),cn2(1830), one_scnb(1200), &
one_scee(1200), &
solty(MAX_ATOM_TYPE), &
gamc(1200),gams(1200), &
asol(200),bsol(200),hbcut(200), &
cn3(1830),cn4(1830),cn5(1830)
common/rparms/rk,req,tk,teq,pk, &
pn,phase,cn1,cn2,one_scnb, one_scee, solty, &
gamc,gams,asol,bsol,hbcut, &
cn3,cn4,cn5 ! mjhsieh: PLEASE, there is no need for common block here.
where MAX_BOND_TYPE is set to 5000 above and MAX_ATOM_TYPE is set to 100
above the quoted text. You need to look at the lengths of every array
inside the rparms common block and add them up. That is, the size of rk,
req, tk, teq, pk, etc. BC_PARMR should be set to that sum. So in this
case the sum is:
5000 + 5000 + 900 + 900 + 1200 + 1200 + 1200 + 1830 + 1830 + 1200 + 1200 +
100 + 1200 + 1200 + 200 + 200 + 200 + 200 + 1830 + 1830 + 1830, which
should add up to 30050 (hence why I said that number is wrong). If you
take out the last 3 arrays (CN3, CN4, and CN5), then the number adds up to
the 24560 that is present there now. If you change 1830 to another number,
you will have to add up this new sum.
Note this only *really* matters when running sander.MPI, since BC_PARMR is
only used in MPI_Bcast calls.
Good luck,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 28 2012 - 07:30:02 PST
