[gmx-users] Re:velocities

[Prasad Gajula]

Thanks for your answer. I hope I did not catch your points well.

So, you mean I can  use 'gen_vel = no'  for the md runs (MD without position
restraints, after the completion of PR run with position restarints)
my procedure as follows......

EM ---> PR(300ps @ 300K) ----> MD (30ps @ 80K. no postion restraints)---> MD
(30 ps @ 100K, no postion restraints ) ---> MD (30 ps @ 200K, no postion
restraints) ----> relaxation MD(250ps @ 300K, no postion restraints) ---->
final MD (300K, no postion restraints)

After Energy minimization, I apply position restarints and generate new
velocities at 300K.
I take the output of PR run as input to the next MD run which started at 80K
( no position restraints from now). And also from here onwards i dont
generate any new velocities, but with the velocities  which are written in
the previous runs.
Looking forward for your reply.

Please let me know
Thanks again
Gajula

> Dear Gromacs Users,
>
> I am getting different results for simulations with " gen_vel=yes"  and
> "gen_vel= no".

This is expected. No two simulations will be the same unless you run
then on the same hardware with exactly the same inputs. gen_vel requires
the use of a random number seed. What you want to do, is after
equilibration, be sampling from the correct ensemble.

> I have generated velocities only for PR MD at 300K. after that I set '
> gen_vel= no ' in mdp file.  After the position restrained MD , I slowly
> increased the temparature from 80K , 100K , 200K and last 300K. (with
> 'gen_vel=no' for all MD)
> Is it correct? Because I read both 'yes' and 'no' as possibilites in some
> references. Iam bit confused which to use for exact simulations.

Personally, I don't think the practice of velocity reassignment in
incremental heating while using position restraints is demonstrably
better than not using reassignment in incremental heating.

Mark