[gmx-users] rerun forces
Attilio
vargiu at sissa.it
Tue Feb 21 00:40:53 CET 2006
Hi,
thanks for answering my question!
Here are reported the whole mdp file, and the differences with that used
for the rerun.
I do not remember the reference, but the things I read are those
reported in the previous mail.
For what concerns the errors in the mdp file, they are errors due to
"cut & paste" from the screen, I didn't put any
crashes
or others flags.
Anyway, I didn't modify the flags you suggest, i.e. gen vel and
unconstrained start, maybe this is my error... I will check!!!
Thanks,
Attilio
MDP
mdp file
title = imine+DNA ; a string
cpp = /lib/cpp ; c-preprocessor
; RUN CONTROL
integrator = md
dt = 0.0015
nsteps = 7000000 ; number of steps
init_step = 0 ;if there is need to
restart, see tpbconv program
comm_mode = angular ; removal of
translational and rotational CM motion
nstcomm = 25 ;frequency of the removal
comm_grps = solute solvent ;group(s) for CM motion
removal, default is all system
; OUTPUT CONTROL
nstcheckpoint = 1000 ; checkpointing helps
you continue after crashes
;nstxtcout = 0 ; output frequency and
precisionfor xtc file
;xtc-precision = 1000
nstxout = 1000 ; write coords
nstvout = 1000
;nstfout
nstlog = 1000 ; print to logfile
nstenergy = 1000 ; print energies
energygrps = solute solvent
; NEIGHBOR SEARCHING
nstlist = 15 ; update pairlist
ns_type = grid ; pairlist method
pbc = xyz
rlist = 1.2 ; cut-off for ns
; ELECTROSTATIC AND VDWAALS
coulombtype = PME
;rcoulomb_switch = 0
rcoulomb = 1.2 ; cut-off for coulomb
epsilon_r = 1
;vdwtype
;rvdw_switch = 0
rvdw = 1.2 ; cut-off for vdw
DispCorr = EnerPres ; Apply long range dispersion
corrections for Energy and Pressure
fourierspacing = 0.12
fourier_nx = 64
fourier_ny = 60
fourier_nz = 66
ewald_rtol = 1e-5
pme_order = 4
optimize_fft = yes
; T COUPLING
tcoupl = Nose-Hoover
tc_grps = solvent solute
ref_t = 300. 300.
tau_t = 1.0 2.0
; P COUPLING
pcoupl = Parrinello-Rahman
compressibility = 4.5e-5
tau_p = 1.0
ref_p = 1.0
pcoupltype = isotropic
; VEL GEN
gen_vel = yes
;gen_temp = 300
;gen_seed = 173529
; BONDS
constraints = h-bonds
constraint_algorithm = lincs
lincs_order = 4
__________________
Here are the differences between the two files:
sdiff -s ../md_imine+DNA.mdp md_rerun.mdp
<
nstcheckpoint = 1000 ; checkpointi |
nstcheckpoint = 100000000000 ; checkpointi
;xtc-precision = 1000 |
;xtc-precision = 1000
nstxout = 1000 ; write coords |
nstxout = 100000000000 ; write coord
nstvout = 1000 |
nstvout = 100000000000
;nstfout |
nstfout = 1
nstlog = 1000 ; print to log |
nstlog = 100000000000 ; print to lo
nstenergy = 1000 ; print energi |
nstenergy = 100000000000 ; print energ
nstlist = 15 ; update pair |
nstlist = 1 ; update pair
Mark Abraham wrote:
> Please include the whole .mdp file. The simplest explanations for what you
> describe are that you are attempting to restart with an inappropriate
> combination of inputs, or that your .mdp file specifies generation of new
> velocities upon restart. There's a few other comments below.
>
> Mark
>
>
>> Hi all,
>> I have a little problem with the rerun of the trajectories to get out
>> forces acting on atoms.
>> To calculate the forces I followed the instructions I read in the
>> mailing list.
>>
>
> If you have a reference, please give a link to it - maybe they were wrong :-)
>
>
>> I changed the mdp file, as below, to get the forces every frame, without
>> re-calculating the neighbor list....
>>
>> -----------------
>> mdp ORIGINAL
>> ....
>> ; OUTPUT CONTROL
>> nstcheckpoint = 1000 ; checkpointing helps you continue after
>> crashes
>>
>
> If this option exists, it isn't documented!
>
>
>> nstxtcout = 1000 ; output frequency and
>> precisionfor xtc file
>>
>
> This is only the output frequency, not precision
>
>
>> nstxout = 1000 ; write coords
>> nstvout = 1000
>> nstfout = 1000
>> nstlog = 1000 ; print to logfile
>> nstenergy = 1000 ; print energies
>> ; NEIGHBOR SEARCHING
>> nstlist = 15 ; update pairlist
>> ns_type = grid ; pairlist method
>> pbc = xyz
>> rlist = 1.2 ; cut-off for ns
>>
>> -----------------------
>> mdp RERUN
>> ....
>> ; OUTPUT CONTROL
>> nstcheckpoint = 1000000000 ; checkpointing helps you continue
>> after crashes
>> nstxtcout = 1000000000 ; output frequency and
>> precisionfor xtc file
>> nstxout = 1000000000 ; write coords
>> nstvout = 1000000000
>> nstfout = 1
>> nstlog = 1000000000 ; print to logfile
>> nstenergy = 1000000000 ; print energies
>> ; NEIGHBOR SEARCHING
>> nstlist = 0 ; update pairlist
>> ns_type = grid ; pairlist method
>> pbc = xyz
>> rlist = 1.2 ; cut-off for ns
>>
>> I obtain forces of the order of 10e+30 Kj mol-1 nm-1 (are these the
>> correct units, to divide by f~139 to get forces in (e/nm)^2 ?), that
>> sounds quite strange.
>>
>
> There's a section on units in the start of the gromacs manual. I suggest
> you have a read of it. I can't think where forces this large are coming
> from without more information.
>
>
>> In fact I run a new short (100ps) simulation in which forces are printed
>> out, and I obtain much smaller forces, between 10e+0 and 10e+5.
>>
>
> They sound more correct.
>
>
>> Then I tried putting _nstlist = 1_, updating the forces each step during
>> the rerun (yes?) of the trajectory, and I obtain values of the same
>> order of magnitude, but with different values of the forces.
>> I would expect to find the same forces at least for the frame number 0.
>> What's wrong? I need to do some imaging of the trajectory, or to leave
>> unchanged also the nslist flag? In my case, the results do not vary by
>> changing neither the flag nstfout from 1 to 1000, neither the nstlist
>> from 1 to 1000.
>>
>
> Restarts (and reruns of restarts) require that you supply the previous
> .trr and .edr files to grompp so that all the coordinates, box parameters,
> etc. are correct. You will also want unconstrained_start = yes and gen_vel
> = no to avoid performing geometric constraints before the first step
> (since they were applied after the last step of the previous simulation)
> and to use the velocities in the .tpr file rather than invent new ones.
>
> Mark
>
> _______________________________________________
> gmx-users mailing list gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>
>
--
Attilio Vittorio Vargiu
PhD Student in Statistical and Biological Physics
SISSA/ISAS, via Beirut 4 I-34014 Trieste, Italy
tel +390403787335 fax +390403787528
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20060221/91e5bd02/attachment.html>
More information about the gromacs.org_gmx-users
mailing list