Correct way to restart a secong MD simulation?

User discussions
1

GROMACS version:
GROMACS modification: Yes/No
Here post your question

Hello Everyone,

What is the exact way to elongate a simulation after finishing 1st run?
I have run a simulation for 5ns and then used the .cpt file to run for the next 5ns.
The md.mdp file contains,

continuation = yes
gen_vel = no

The gmx commands I have used are as follows,
gmx grompp -f md.mdp -c *NPT.gro -r *NPT.gro -n *.ndx -p *.top -t *NPT.cpt -o *MD1.tpr
gmx grompp -f md.mdp -c *MD1.gro -r *MD1.gro -n *.ndx -p *.top -t *MD1.cpt -o *MD2.tpr

When I was catenating two .xtc files, I found the start time of the second run was at 0. It was solved by using -settime and got a 10 ns trajectory.

I am curious if the velocities are read correctly from NPT to MD1 and then MD1 to MD2.
And what is the correct way to start the second trajectory (MD2) that should not start from t0 = 0.0 ps.

Waiting for your comments and suggestions.
Thank you in advance.

Regards
Ranajit

2

You do not need to invoke grompp to restart a simulation (you do if you want to change some aspect of the parameters, but this is not a typical use case). See Managing long simulations — GROMACS 2022.1 documentation for how to extend/restart a simulation.

3

Thank you, Justin,

I saw those pieces of information. It uses convert-tpr followed by *-cpi .cpt. I will use it nexttime.

But, I have already completed the second simulation using grompp (and -t cpt). Then I connected the geometries using trjcat (with -settime flag). I wonder if it is the correct way to read x and v (using -t cpt and continuation) from the last MD1 and then catenate them.

Regards
Ranajit

4

Yes, that should be fine.

5

Thank you.
It was helpful.