On Thu, Nov 13, 2003, Robert Duke wrote:
> Any update on how you guys are feeling about PMEMD? Want it to remain
part
> of the product in the 8 timeframe? Any specific requests?
Sorry. PMEMD was not on the June, 2003 "to-do" list, so I did not include
anything about it in my latest e-mail. That was an oversight on my part.
My intent is to include PMEMD officially in the amber8 release, so that
everyone will have it. I'm assuming (from earlier e-mails) that this will
be OK with you and with Lee, and that we/you can adapt your release notes
to become a chapter in the Users' Manual (I'm happy to work on that.)
PMEMD is clearly useful to amber users, whether or not NAMD might be
faster
on some machine or other. There is a whole new learning curve for NAMD,
which can be avoided by using PMEMD, which has minimal differences with
the
sander code people are already familiar with. And there appear to be lots
of (pretty common) machines where PMEMD is much faster than NAMD.
As for "specific requests": the modifications that seem to be to have the
greatest potential benefit are these:
(1) The integrators in runmd.f have been updated (yet again!). The new
code
is actually simpler than the old, but has better temperature regulation
options. It would be nice if PMEMD could incorporate this, so that
trajectories with sander8 and PMEMD could continue to be compeltely
equivalent. (I'm willing to work on this, either by directly modifying
code
or by providing advice.)
(2) The thermodynamic integration capability of sander is another key
place
where users could benefit from faster simulations that have better
parallel
scaling. I know that you thought about this a few months ago, and decided
that there were some complications (which is correct), but I don't think
they
are really major ones.
(3) There is a new "DUMPFREQ" (dump frequency) parameter in sander8 that
provides an avenue for users to create potentials of mean force from
umbrella
sampling (and similar calculations). Implementing this in PMEMD should
only
involve a few lines of code, and would allow users to transfer the PMF
instructions from sander (which will be in the new users' manual) directly
into PMEMD.
Of course, the above depends on your time and interests, and are only
intended
as ideas to think about. I know you have mentioned other things (new
electrostatics, merging the cit and non-cit code, etc.), and it's hard to
make
plans when funding is uncertain. I'm willing to help with certain
projects,
and certainly to put in my two cents, but it's clearly up to you and Lee
to
make the "real" decisions.
Hope this helps clarify things. Apologies again for leaving out the
subject
of PMEMD in the previous e-mail.
.dac
--
==================================================================
David A. Case | e-mail: case.scripps.edu
Dept. of Molecular Biology, TPC15 | fax: +1-858-784-8896
The Scripps Research Institute | phone: +1-858-784-9768
10550 N. Torrey Pines Rd. | home page:
La Jolla CA 92037 USA | http://www.scripps.edu/case
==================================================================
Received on Wed Apr 05 2006 - 23:50:07 PDT
