Start a new simulation from a previous state and coordinates

User discussions
1

GROMACS version:2022.4
GROMACS modification: No
Here post your question
Good afternoon,
I would like to start a brand new simulation using the last coordinates from a previous run as a starting point.

I don’t understand the difference between:
gmx grompp -f new_run.mdp -c previous.gro -t previous.cpt -p topol.top -o new_tpr.tpr
(which explicitly refers to the last generated .gro and .cpt)

and
gmx convert-tpr -s previous.tpr extend 200 -o new.tpr
(which uses the last generated .tpr, extending the duration).

Say I want to start a brand new simulation from the last checkpoint/frame, which one should I use?
Given that every time I reassign velocities (as specified in the .mdp) I don’t understand which command I should use.

It seems that -extend increases the duration of a simulation, maintaining all the data from the previous one (being a continuation) while creating the tpr from “scratch” every time might what I am looking for, but I’d rather ask more experienced users.

Thanks!

2

But after you have the new extended tpr, you have to start the simulation with that tpr and the last checkpoint file using mdrun.
I think the first one should be used when you are using a new mdp file with some parameter change, but still want to do a continuation. In case of the convert tpr , you are using the same parameters from the old mdp and just extending the simulation time.
Based on your requirements: Given that you probably have gen vel on in the mdp, I think you should re grompp everytime you restart a simulation.

1 Like
3

Thank you, I just was not sure about what to do!

1 Like