[gmx-users] RE: Is non-linear data output/storage possible?

[Thomas Schlesier]

I see the same problem as Justin.
The real problem lies here:

mdrun -nt 4 -v -s 10step.tpr -c nonlinear.gro -e nonlinear.edr -x 
nonlinear.xtc -o nonlinear.trr -g nonlinear.log -cpi nonlinear.cpt -cpo 
nonlinear.cpt -rdd 2.5

Due to the '-cpi nonlinear.cpt' mdrun thinks the the time now the 
end-time of the longer simulation. But in the grompp step, you tell GMX 
thaat the time starts at 0 and should go till the shorter end-time.

You could use Justin's approach. Or skip the '-cpi nonlinear.cpt' in the 
mdrun step. Problem with the later is, that you'll end with many 
trajectories which all start at t=0.


One thing which is strange:
Why do you use in the second grompp/mdrun step, the *.tpr file of the 
previous run as input for the coordinates? Would think it will be better 
to use the final coordinates of the previous run and not the initial 
coordinates. (Being confused)

Greetings
Thomas



Am 02.07.2013 20:51, schrieb gmx-users-request at gromacs.org:
>> Hi,
>> >
>> >So the commands are:
>> >
>> >grompp -f 10step.mdp -p dppc_bilayer.top -o 10step.tpr -c dppc_bilayer.gro
>> >
>> >mdrun -nt 4 -s 10step.tpr -c nonlinear.gro -e nonlinear.edr -x
>> >nonlinear.xtc
>> >-o nonlinear.trr -g nonlinear.log -cpo nonlinear.cpt -rdd 2.5
>> >
>> >grompp -f 100step.mdp -c 10step.tpr -p dppc_bilayer.top -o 100step.tpr -t
>> >nonlinear.cpt
>> >
>> >mdrun -nt 4 -s 100step.tpr -c nonlinear.gro -e nonlinear.edr -x
>> >nonlinear.xtc -o nonlinear.trr -g nonlinear.log -cpi nonlinear.cpt -cpo
>> >nonlinear.cpt -rdd 2.5
>> >
>> >Up till this point the output is what I would expect. However now I try to
>> >use the 10step mdp again in order to save it every ten time steps -
>> >
>> >  grompp -f 10step.mdp -c 100step.tpr -p dppc_bilayer.top -o 10step.tpr -t
>> >nonlinear.cpt
>> >
>> >mdrun -nt 4 -v -s 10step.tpr -c nonlinear.gro -e nonlinear.edr -x
>> >nonlinear.xtc -o nonlinear.trr -g nonlinear.log -cpi nonlinear.cpt -cpo
>> >nonlinear.cpt -rdd 2.5
>> >
>> >And at this mdrun I get WARNING: This run will generate roughly
>> >186617909559164928 Mb of data
>> >
>> >starting mdrun 'DPPC BILAYER'
>> >100 steps, infinite ps (continuing from step 1000,     20.0 ps).
>> >
>> >
>> >This is the saving every 10 steps mdp file, the hundred step is identical
>> >except steps is a 1000 and nstxout is 100.
>> >
>> >; VARIOUS PREPROCESSING OPTIONS =
>> >title                    = Martini
>> >cpp                      = /usr/bin/cpp
>> >
>> >; RUN CONTROL PARAMETERS =
>> >integrator               = md
>> >; start time and timestep in ps =
>> >tinit                    = 0.0
>> >
> This is your problem.  You're no longer at t = 0.  The checkpoint file
> specifies some time in the future, so grompp thinks you're going backwards
> in time and thinks you're going to generate infinite data by doing so.
> Then mdrun gets confused, because it thinks you're trying to go backwards
> as well, from step 1000 to step 100.  Proper use of tinit and init_step in
> the .mdp file should alleviate this problem.
>
> -Justin