From: Rajan Vatassery (rajan_at_umn.edu)
Date: Mon Dec 10 2012 - 13:33:08 CST
Eddie,
I usually start with a PDB/PSF that comes from VMD, and using the
default setting:
binaryoutput yes
I receive a .coor, .vel, and a .xsc file from the first (NpT)
simulation. This simulation minimizes and equilibrates the system at a
variable volume.
The next simulation, NVT, receives the following input from the NpT
simluation:
binCoordinates NpT.coor
binVelocities NpT.vel
I find the average volume that was produced from the NpT simulation
after the volume stops changing drastically, and this number is set as
the fixed volume with the lines:
cellBasisVector1 xvector 0 0
cellBasisVector2 0 yvector 0
cellBasisVector3 0 0 zvector
cellOrigin 0 0 0
where xvector*yvector*zvector is your total volume. NVT runs and outputs
the same as the NpT simulation, and this time I use all three
(.vel, .coor, and .xsc) files as inputs to the NVE simulation.
binCoordinates NVT.coor
binVelocities NVT.vel
extendedSystem NVT.xsc
Because in the NVE simulation I am using the .xsc file to determine the
PBC boundaries, I comment out the cellBasisVector lines. Note that going
from NpT to NVT sometimes produces erroneously high energies. Basically
the source of this problem is that you are taking a volume of your
simulation (the average over the NpT) that doesn't necessarily
correspond to the .vel and .coor files (which are just the snapshots of
the last point in the NpT). If the volume of your last point is much
different than the average, some strange things can happen as a result
of the compression/expansion of your system. The way to detect if your
system has suffered from this is to watch the energies of your system
from NpT to NVT to NVE. If the system energy has changed drastically
from NpT to NVE, and that change isn't just a result of a small T
difference from removing the thermostat, then you'll have to find the
coordinates and velocities from the .dcd file of a configuration where
the volume is closer to your average volume. This is a complicated
process, and even more complicated to describe without
figures/equations, but the solution is simple.
I can also send over my conf files if you would like.
rajan
On Mon, 2012-12-10 at 12:20 -0600, Dr. Eddie wrote:
> Rajan,
> May I ask how you did the start and restart exactly? Did you build a
> new pdb file in vmd and start with that after the Langevin dynamics
> were done?
> Thanks,
> Eddie
>
>
> On Mon, Dec 10, 2012 at 11:00 AM, Rajan Vatassery <rajan_at_umn.edu>
> wrote:
> Hi Eddie,
> Unfortunately, as Norman Geist suggested, the best way
> to accomplish
> what I was trying to do was to set up multiple conf files. I
> looked into
> the "checkpoint" keyword as well, but I couldn't find enough
> details to
> figure out how to use it properly. It sounds like you're
> encountering a
> similar situation.
> My obsessive-compulsive side really wanted to keep my
> directories
> simple by using only one conf file and one set of output
> files, but
> separating things was just easier. If you find out how to use
> the
> checkpoint/revert keywords, I'm sure there are a lot of people
> out there
> that could use a few pointers.
>
> Good luck,
>
> Rajan
>
> On Mon, 2012-12-10 at 09:08 -0600, Dr. Eddie wrote:
> > Thanks Rajan. So how did you "solve" turning the langevin
> dynamics
> > off? Restart the simulation somehow? Were you able to
> restart the
> > simulation in 1 config file?
> >
> >
> > I know there are commands 'checkpoint' and 'revert', from
> here, but
> > the documentation on these is scarce at best! It sounds like
> I should
> > be able to make a checkpoint and then revert back changing
> some system
> > parameters, but I have no idea how!
> > Thanks!
> > Eddie
> >
> >
> > On Sun, Dec 9, 2012 at 5:02 PM, Rajan Vatassery
> <rajan_at_umn.edu> wrote:
> > Eddie,
> > I ran into a similar problem as your last
> error
> > message a while back:
> >
> >
> http://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l/14609.html
> >
> > It's not possible to change all of the conf file
> > keywords/variables
> > after the first "run" command in the conf file.
> Here's the
> > page to read:
> >
> >
> http://www.ks.uiuc.edu/Research/namd/2.8/ug/node10.html
> >
> > I haven't seen your first error before so I can't
> comment on
> > that.
> >
> >
> > Good luck,
> >
> >
> > rajan
> >
> > On Sun, 2012-12-09 at 10:16 -0600, Dr. Eddie wrote:
> > > Hi all,
> > > I've been trying to setup a single config file
> that would do
> > the
> > > minimization, then slow heating with pressure
> equilibration
> > and then
> > > turn off the Langevin dynamics once the
> structure/solvent
> > has reached
> > > the desired temperature/pressure.
> > >
> > > I have tried many things and would like some
> advice to make
> > this work
> > > (if it is possible) and ensure I'm not making some
> gross
> > error.
> > > The relevant part of the config file is (my number
> of
> > timesteps is
> > > small so I can see the crashes quickly)
> > >
> > > langevin on
> > > langevinDamping 1.
> > > langevinTemp 100 # <---- 1st problem
> is if this
> > (and all
> > > the temperatures below) is lower than 100
> > >
> #REINITIALIZING
> > VELOCITIES AT
> > > STEP 16 TO 25 KELVIN.
> > > #TCL:
> Running for 48
> > steps
> > > #FATAL
> ERROR:
> > Periodic cell has
> > > become too small for original patch grid!
> > > langevinHydrogen no
> > > useGroupPressure yes
> > > useFlexibleCell no
> > > useConstantArea no
> > > langevinPiston on
> > > langevinPistonTarget 1.01325
> > > langevinPistonPeriod 100.
> > > langevinPistonDecay 50.
> > > langevinPistonTemp 100
> > > temperature 100
> > > for { set MIN 0 } { $MIN < 3 } { incr MIN 1 } {
> > > minimize 16
> > > reinitvels 100
> > > run 48
> > > }
> > >
> > > # I need to start my heating at 100 k (set TEMP
> 100 below)
> > otherwise I
> > > also get the error:
> > > #TCL: Setting parameter langevinPistonTemp to 25
> > > #TCL: Setting parameter langevinTemp to 25
> > > #REINITIALIZING VELOCITIES AT STEP 192 TO 25
> KELVIN.
> > > #TCL: Running for 48 steps
> > > #FATAL ERROR: Periodic cell has become too small
> for
> > original patch
> > > grid!
> > > for { set TEMP 100} { $TEMP < 285 } { incr TEMP
> 25 } {
> > > langevinPistonTemp $TEMP
> > > langevinTemp $TEMP
> > > reinitvels $TEMP
> > > run 48
> > > }
> > > temperature 310 #<------- If I use 100
> K for
> > minimization
> > > etc This command fails with:
> > > #TCL: Setting
> parameter
> > > temperature to 310
> > > #FATAL ERROR:
> Setting
> > parameter
> > > temperature from script failed!
> > > # DO I need
> it?
> > >
> > > langevinPistonTemp 310
> > > langevinTemp 310
> > > reinitvels 310
> > > run 50000
> > >
> > > langevinPiston off #<---- doing all of the
> above
> > (which seems
> > > dubious) this fails with:
> > > #TCL: Setting
> parameter
> > > langevinPiston to off
> > > #TCL: Setting
> parameter
> > langevin to
> > > off
> > > #FATAL ERROR:
> Setting
> > parameter
> > > langevin from script failed!
> > > langevin off
> > > run 20000000
> > >
> > >
> > > I'd like to end up with a minimized, slowly heated
> system at
> > the right
> > > temperature/pressure. Then I'd like to run that
> system
> > without the
> > > Langevin forces on to see how much for the
> protein's
> > sidechains alone
> > > move (rmsd, distances to other CA's etc) without
> any random
> > forces
> > > increasing the distance.
> > >
> > > Thank you very much for any help!
> > > Eddie
> > >
> > >
> > > --
> > > Eddie
> > >
> >
> >
> >
> >
> >
> >
> > --
> > Eddie
> >
> >
>
>
>
>
>
>
> --
> Eddie
>
>
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:22:21 CST