[GMX-USERS] Pull Code

[Justin MacCallum]

Hi David,

I will send you the code in a seperate email to avoid cluttering up the
mailing list.


> That would be great!  Could you answer a question or two about things as
> well?  I guess my concern is more about knowing what is going on when I use
> the pull code.  I know what parameters I input for umbrella sampling, but
> how do you validate that the restraint potential is working?. 

I'm pretty sure that the umbrella sampling in 2.1/3.0/3.1 is broken.  To
test the restraint potential is working, I restrained a single water
molecule in a box of water and compared the histogram of the position of
the water molecule with exp(-U(x)/RT) and found that the restraint was
producing the expected output.

> Is there
> some output besides the .pdo file?  I tried -debug for mdrun, but that was
> cryptic.

Not really.  The pdo file is really the only output besides the debugging
stuff.


>  Does the direction set of values also influence which components
> of direction are restrained between group_1 and reference_group?.. For
> example does direction = 0.0 0.0 0.1 mean that the spring acts only in the
> z dimension?  .. or is that only the case for afm pulling or the direction
> of position scaling for starting structure generation?

Yes, the direction entry specifies which components are used. I assume
that you meant to put direction = 0.0 0.0 1.0 for the z-axis.


> Since I couldn't determine whether umbrella sampling was working, I tried
> using afm pulling with pullrate = 0.0 , which should be the same thing as
> doing one run with umbrella sampling.  I checked the output forces with a
> manually calculated force using my input spring constant and center of mass
> data from g_traj, but the forces didn't jive.

I'm not sure what the problem is.


The options for specifying the umbrella potential have changed slightly.
There are now four K entries and four Pos entries, one for each pulled
group.  K is the force constant in kJ/(mol nm) and Pos is the position of
the center of the restraining potential with respect to the refernce
group.  For my simulations of hexane in DOPC, I used either K=500 and
windows seperated by 0.25nm or K=3200 and windows seperated by 0.1nm.

Justin