On Thu, Jan 22, 2015 at 7:58 AM, David A Case <case.biomaps.rutgers.edu>
wrote:
> On Thu, Jan 22, 2015, Jason Swails wrote:
> >
> > While sander and pmemd have a character limit on
> > the length of the mask,
>
> Why don't we just change this?
>
We can increase the limit, but as long as we use Fortran namelists to
supply sander/pmemd input, we will *always* need some sort of limit. [1]
At the time, 256 characters probably seemed like a
"too-large-to-ever-be-exceeded" safe limit :).
The main thing I'm concerned about is MPI -- Amber likes to broadcast
entire common blocks via the first element (which is perfectly fine), but
changing the size of the individual items *in* the common blocks is a
recipe for subtle bugs if you're not careful. (That said, I don't think the
mds common block is broadcast from master to anywhere else, so this is
probably an easy change).
>
> Larger update would be to allow multiple masks, with different restraint
>
> weights. We already allow multiple masks (through GROUP), so it's just a
>
> matter of coming up with a user interface that supports it.
This would also be possible, but again, there will need to be arbitrary
compile-time limits imposed on how *many* masks we allow. [1] I'm pretty
sure I know where to make those changes, so I can draft a
proof-of-principle change that implements the above, if you want.
All the best,
Jason
[1] Fortran namelists do not allow dynamic *anything* in input -- all
array/collection sizes must be fixed at compile time
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Jan 22 2015 - 05:30:03 PST
