Re: restarting umbrella sampling simulation.

From: Aron Broom (broomsday_at_gmail.com)
Date: Tue Sep 11 2012 - 08:54:27 CDT

If you are having a lot of trouble with this "automated Umbrella sampling"
it might be useful to just do a quick analysis the old-fashioned way by
running a separate simulation with just a standard harmonic restraint that
you vary for each window. Then once all of that makes sense to you, and
you have run WHAM on it and have a feel for all the potential problems,
move onto the method where the centers are changing over time.

~Aron

On Tue, Sep 11, 2012 at 8:02 AM, Jérôme Hénin <jhenin_at_ifr88.cnrs-mrs.fr>wrote:

> Your config file defines only one group of atoms, "group2" is only a
> fixed position in space. If group1 moves towards that fixed position,
> then the code is doing what you asked from it.
>
> If you can describe what you want to do clearly in terms of a
> collective variable, then it should be simple to write the config
> file.
>
> Jerome
>
>
> On 11 September 2012 13:12, ipsita basu <ibasu788_at_gmail.com> wrote:
> > I could not solve the issue also when I set the variable as distance.
> > I just want to put harmonic restraints on one segment of the protein
> > whose centre of mass has to move to the reference point to decrease
> > the distance between the above mentioned point.
> > What I have to set in group1 and group2 selection?
> > Although the manual regarding those selection is very well written,
> > but I can't figure out what actually I've to set in that selection.
> > Really sorry for repeated disturbance but I really need your help.
> >
> > On 9/5/12, Jérôme Hénin <jhenin_at_ifr88.cnrs-mrs.fr> wrote:
> >> On 5 September 2012 09:34, ipsita basu <ibasu788_at_gmail.com> wrote:
> >>> Sorry to bother you again. But I have again some doubts which I want
> >>> to make clear.
> >>> One point is:
> >>> If I set centers at 70.0, the traj file shows that the distance is
> >>> increasing from 68.3 and if I set centers at 60.0 then the distance is
> >>> decreasing from 68.3, from this observation I think that the initial
> >>> distance is 68.3 and it tries to reach to the value set at centers.
> >>
> >> Yes, this sounds right. Remember that the current value of the
> >> restraint centers are always in the NAMD output.
> >>
> >>> Then what is the importance of targetcenters,
> >>> targetnumsteps,targetnumstages etc.?
> >>
> >> If everything is set up right, they should matter.
> >>
> >>> Second point is:
> >>> From namd list I get an instruction “give a fixed center for the
> >>> harmonic restraint, and run the simulation for some time, then you
> >>> give a different center and repeat.” and according to your suggestion
> >>> not to change the configuration file when restarting the job.
> >>
> >> That is only if you want all your windows to be run sequentially, each
> >> one starting from the end of the previous one.
> >>
> >>> So can I
> >>> run the whole umbrella sampling simulation for just one time setting
> >>> centers at 20.0 (my desire is to change the distance from 68.3 to 20)
> >>> ?
> >>
> >> My guess is, you should keep centers at 70 and set targetCenters at 20.
> >>
> >>> Can't I analyze it with wham with only one traj file?
> >>
> >> Everything is possible.
> >>
> >>> And if the
> >>> above things are possible then what will be the value of targetcenters
> >>> in my configuration file when centers at 20.0?
> >>
> >> I don't think the value of centers should be 20 here.
> >>
> >> Jerome
> >>
> >>
> >>
> >>> On 9/4/12, Jérôme Hénin <jhenin_at_ifr88.cnrs-mrs.fr> wrote:
> >>>> Ipsita,
> >>>> Please keep the list in CC so that others can follow this
> conversation.
> >>>>
> >>>> The config file should not be changed for restarting (i.e. "centers
> >>>> 70" should stay that way), as NAMD should get from the restart file
> >>>> information about the current stage of the transformation.
> >>>>
> >>>> The very first stage should be numbered zero. Please look for output
> >>>> lines mentioning changes in restraint centers in the log file of the
> >>>> first NAMD run.
> >>>>
> >>>> Finally, be aware that targetNumSteps gives the number of steps per
> >>>> stage, not the total. The total simulation time should be
> >>>> (targetNumStages + 1) * targetNumSteps.
> >>>> The +1 is for sampling the last point: for 100 stages, there are 101
> >>>> points to be sampled.
> >>>>
> >>>> Sorry if the documentation was not explicit enough about this. We try
> >>>> to improve it as we realize what's needed, thanks to user feedback
> >>>> like yours.
> >>>>
> >>>> Jerome
> >>>>
> >>>>
> >>>> On 4 September 2012 10:10, ipsita basu <ibasu788_at_gmail.com> wrote:
> >>>>> Thank you sir for your reply.
> >>>>> Whatever I choose the value of centers, keeping fixed targetCenters
> >>>>> 50.00 , the NAMD output is :
> >>>>> colvars: Moving restraint stage 1 : setting centers to { 59.802 }--e89a8ff25144a0334004c96d694c--

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:22:33 CST