[gmx-users] extending simulation with grompp incorporating velocities

[Mark Abraham]

On 21/10/2011 3:25 PM, Chandan Choudhury wrote:
>
> Dear gmx users,
>
> A simulation was performed for 50 ns with
>
> ; RUN CONTROL
> integrator               = md
> nsteps                   = 50000000
> dt                       = 0.001
> ; OUTPUT CONTROL
> nstxout                  = 1000
> nstvout                  = 1000
> nstxtcout                = 0
> nstlog                   = 100
> nstenergy                = 100
>
> The output produced were 0-50.edr, 0-50.trr, 0-50.log, 0-50.pdb 
> state.cpt, state_prev.cpt. The simulation completed normally.
>
> After 50 ns, I intend to extend for 4 more ns, write the trajectory 
> file frequently with

Use tpbconv -extend. See 
http://www.gromacs.org/Documentation/How-tos/Extending_Simulations

> ; RUN CONTROL
> integrator               = md
> nsteps                   = 4000000
> dt                       = 0.001
> ; OUTPUT CONTROL
> nstxout                  = 00
> nstvout                  = 00
> nstxtcout                = 10
> nstlog                   = 500
> nstenergy                = 10
>
> $grompp -f md.mdp -p topol.top -o md50-54.tpr -c 0-50.pdb  -n index.ndx
>
> $gmxdump -s md50-54.tpr | more
>
> md50-54.tpr:
> inputrec:
>    integrator           = md
>    nsteps               = 4000000
>    init_step            = 0
>    ns_type              = Grid
>    nstlist              = 10
>    ndelta               = 2
>    nstcomm              = 1
>    comm_mode            = Linear
>    nstlog               = 500
>    nstxout              = 0
>    nstvout              = 0
>    nstfout              = 0
>    nstenergy            = 10
>    nstxtcout            = 10
>    init_t               = 0
>    delta_t              = 0.001
>    xtcprec              = 1000
>    nkx                  = 22
>    nky                  = 22
>    nkz                  = 40
>    pme_order            = 4
>    ewald_rtol           = 1e-05
>    ewald_geometry       = 0
>    epsilon_surface      = 0
>    optimize_fft         = FALSE
>    ePBC                 = xyz
>    bPeriodicMols        = FALSE
>    bContinuation        = FALSE
>    bShakeSOR            = FALSE
>    etc                  = V-rescale
>    epc                  = Berendsen
>    epctype              = Isotropic
>    tau_p                = 1
>
> $mdrun_mpi-4.5.5 -s md50-54.tpr -c 50-54.pdb -x 50-54.xtc -e 50-54.edr 
> -g 50-54.log -cpi state.cpt -nice 0
>
> state.cpt is the output of the initial 50ns run. I have used state.cpt 
> to incorporate the velocities frm the initial run.
> So, I would expect my 2nd simulation to get complete after running for 
> 4 ns. But, that was not the case. I have checked the log file. The 
> time clearly exceeds 4ns.
>
> $ tail -15 50-54.log
> DD  step 54500989 load imb.: force  5.6%  pme mesh/force 0.429
>
> DD  step 54500999 load imb.: force  6.0%  pme mesh/force 0.428
>
>            Step           Time         Lambda
>        54501000    54501.00259        0.00000
>
>    Energies (kJ/mol)
>            Bond          Angle    Proper Dih. Ryckaert-Bell.          
> LJ-14
>     1.82780e+02    3.37147e+02    8.77788e+01    1.92877e+01    
> 3.90071e+01
>      Coulomb-14        LJ (SR)   Coulomb (SR)   Coul. recip.      
> Potential
>     5.96624e+02    7.91553e+04   -5.73437e+05   -2.73830e+04   
> -5.20402e+05
>     Kinetic En.   Total Energy    Temperature Pressure (bar)
>     9.75908e+04   -4.22811e+05    2.97855e+02   -5.82198e+01
>
> Kindly suggest what wrong am I doing. and how do I incorporate the 
> velocity from my earlier run.

On the information given, I can't explain this observation.

Mark