Extending Simulaton

User discussions
, ,
1

GROMACS version:
GROMACS modification: Yes/No
Here post your question
I have simulated 1 microseconds of a protein and now I wold like to extend it. This is how I did simulation:
First, in the md.mdp file I set nsteps to 50000000 and dt 0.002 picosendos (simulating 100 nanoseconds in each run). Then I wrote the code to execute mdrun for 100 nanoseconds. When the 100ns was completes, I extended it using convert-tpr. For example, the code for extending simulation from 900ns to 1000ns would be as shown below:

###Extend Simulation
gmx convert-tpr -s md_900.tpr -extend 100000 -o md_1000.tpr
###Run
mpirun mdrun_mpi -deffnm md_1000 -s md_1000.tpr -cpi md_900.cpt -noappend

Now I want to extend the simulation another 1 microseconds but I want to do it in a different way because this approach wastes a lot of time queuing to run.
I want to use nsteps -1 in the mdp file and then run this code

###Run
mpirun mdrun_mpi -deffnm md_1000 -s md_1000.tpr -cpi md_900.cpt -noappend -maxh 71.9

By doing this, I will make sure to end the simulation before it gets killed exceeding the 72 hour allocation time. However, I do not know how to extend the tpr file and what to use instead of md_1000.tpr in the code above.
Any help is so much appreciated
Thanks!

2

make a tpr for higher time, lets say 10 microsecond and extend after very 72 hrs using -cpi flag

3

Hi!

Although deprecated, you can still pass the -nsteps -1 flag to gmx mdrun for now. This is the easiest way to keep your .tpr running forever without having to recreate it by putting nsteps = -1 in the .mdp file.