CHARMM Development Project Forums User Discussion & Questions Molecular Dynamics Can I get rst file from dcd?

Dear Charmmers,

Suppose I forgot to write rst file.
Can I get one from the dcd?

Typical input will be as follows (shown the DYNA part only)

------------------------------

open write name traj.dcd unit 32

dyna cpt leap star time 0.002 nstep 20000 iseed 111111 nprint 5000 nsavc 5000 -
pconst pmxx 100. pmyy 100. pmzz 100. pref 1. -
hoover reft 300. tmass 1000. -
nsavv 0 inbfrq -1 iprfrq 100 -
iunrea -1 iunwri -1 iuncrd 32 iunvel -1 kunit -1 -
iasors 0 iasvel 1 iscvel 0 ichecw 0 echeck 1000

Thank you in advance -

Just rerun it. It will be quicker than discussing possible work-arounds in these forums.

Dear Lennart and others,

Out of curiosity,

Does not the dcd have all the information to retrieve the rst?
Is there any way to get rst after the dynamics run?

dcd only contains coordinate information and it is not possible to reconstruct rst out of dcd alone.

I think a "dcd" file can be used, provided that it is a restart file and not some other kind of file. For example, a coordinate trajectory file (or a blo.ip file as it could be called) only contains coordinates. Since F=ma is a second order differential equation more information is required for a proper restart; this information is contained in the restart file (the spru.nk file).

Specifically, the restart file contains coordinates, delta coordinates, velocities, integrator state information, etc. A binary trajectory file, sometimes called a 'dcd' file by people with limited CHARMM experience (I use .trj extension), contains only coordinates.

You can get a restart file (I use .res, but the extension is irrelevant) after the dynamics run since that's when it is normally written, but the proper DYNAmics setup is required; OPEN the file, and point to the integer unit number via the 'IUNWRI integer' option.

In summary, it cannot be done from just coordinates, and the best advice is still to repeat the initial run with the correct setup.

Thank you for the responses

All that said, in an NPT simulation, it usually isn't disastrous to use the last coordinate set of the trajectory and re-initialize the velocities, provided that this doesn't happen too frequently. However, your simulation seems to be only 40ps; reinitializing the velocities every 40ps would definitely classify as "too frequently" in my opinion.

And as Lennart mentioned in his first answer, I would be hard pressed to explain in detail how to extract the last coordinate set from a trajectory and start a new simulation in less time than it takes the computer to run 40ps worth of simulation, unless the simulations is incredibly slow.