> Bob Duke is out of e-mail contact for a while, but the above strategy sounds
> like a good one: basically, you want to assure yourself that pmemd and sander
> are giving equivalent results with ntp=2. Step 3 is probably not really
> necessary; test pmemd with the paths (presumably cit) that you really plan to
> use.
thanks for "approval" :) - but it still doesn't fully answer my question
- i'll run the simulation for let's say 10000 steps (~hour). will it be
enough? i suspect that someone played with the benchmark directory and
set nstlim for dhfr to 100 (it is too small, i think).
and then - what (and how often) should i carefully inspect? iirc, bob
mentioned that there could be some problems with different virials
(connected with their different handling in sander6 and sander7). so
i'll look at energies, virials, pressure fluctuations - anything else?
how often should i check them? every step? or every 100 steps (current
ntpr setup for the benchmark)? i don't have any idea how much could
different virials affect the simulation and on what timescale would it
occur - so that's why i am asking. and the last question - what are
reasonable differences?
thanks for you help,
--
Lubos
_._"
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Apr 14 2004 - 17:53:01 PDT
