Re: Scaled forces, colvars and harmonic con/restraints

From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Wed Nov 23 2011 - 09:28:50 CST

Hi Michelle. What you describe seems to me that the angle gets stuck into
a local minimum. This is pretty typical when you minimize. Then when you
try the next window, you don't get the results you want because your angle
is stuck.

I would start all windows from the same initial structure and change the
center of the static restraint, rather than changing it on the fly while
you minimize. That would be most appropriate for dynamics.

Giacomo

On Wed, Nov 23, 2011 at 9:16 AM, Michelle Kuttel <mkuttel_at_cs.uct.ac.za>wrote:

> The plot thickens....
>
> After much experimenting, I have finally determined that, for this
> particular dihedral, the optimal "harmonic" configuration is:
> harmonic {
> name scanOmega
> colvars Omega
> centers -180
> targetCenters 180
> targetNumSteps 500 #these
> targetNumStages 72 #are the values
> forceConstant 0.005 # in question
> }
>
> So ,it seems that my previous values were far too "high". However, I
> still don't understand why, if I increase targetNumStages and
> targetNumSteps, I need to increase the force constant as well.
> "targetNumSteps" should just be number of minimization steps at a
> particular force, surely, and not affect the applied force, right? But it
> does!
>
> The use of "harmonic" to create relaxed energy maps is not documented in
> the NAMD User Guide, other than the comment:
> "The harmonic biasing method may be used to enforce fixed or moving
> restraints, including variants of Steered and Targeted MD. Within energy
> minimization runs, it allows for restrained minimization, e.g. to calculate
> relaxed potential energy surfaces.".
> So, I would really appreciate it if someone can explain this to me.
> Pretty please?
> Michelle
>
>
> On 21 Nov 2011, at 12:45 PM, Michelle Kuttel wrote:
>
> Hello
>
> I am trying to calculate a simple relaxed energy profile for a dihedral
> angle (this is a simple sanity check - I can compare the profile to one I
> calculated some time ago in CHARMM).
> I am using colvars and harmonic restraints to do this. My colvarsConfig
> file looks like this:
>
> ------------------------------------
> colvarsTrajFrequency 10000
>
> colvar {
> name Omega
> width 0.5
> dihedral {
> group1 {
> atomnumbers { 7 }
> }
> group2 {
> atomnumbers { 5 }
> }
> group3 {
> atomnumbers { 20 }
> }
> group4 {
> atomnumbers { 21 }
> }
> }
>
> }
>
> harmonic {
> name scanOmega
> colvars Omega
> centers -180
> targetCenters 180
> targetNumSteps 10000
> targetNumStages 36
> forceConstant 100
> }
> ----------------------------
>
> In my main file, I then have the line:
> minimize 370000
>
> This works OK (though the energies for the first few stages are a bit
> high). HOWEVER, when I change "targetNumStages 36" to "targetNumStages
> 72" and "minimize 370000" to "minimize 730000", suddenly I need a much
> higher force of 200 for it to work. If I don't increase the force, the
> omega angle (recorded in the .traj file) "stalls" at -145 degrees. The
> angle does not rotate further than this on subsequent stages.
>
> Why would increasing the number of stages require a higher force? I
> thought that the width of the omega colvar would still be the same (0.5
> degrees) and therefore the scaling should be the same?
> I think that I do not understand the way forces are applied under harmonic
> restraints and I would really appreciate some insight!
>
> Many thanks
> Michelle
>
> ---------------------------------------------
> Dr Michelle Kuttel
> Department of Computer Science
> University of Cape Town
> Cape Town
> South Africa
> mkuttel_at_cs.uct.ac.za
> PH: +27 21 6505107
> ---------------------------------------------
>
>
>
>
>

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:01 CST