[gmx-users] calculation speed of GROMACS and cut off parameters

[Justin Lemkul]

On 4/21/14, 5:48 PM, mircial at sjtu.edu.cn wrote:
> Hi everyone,
>
> I am using GROMACS to simulate bilayers, and I am testing two sets of parameters at present. However, I found significant differences of simulation speed of these two parameters sets.
> The mainly differences between these parameters are the cut off parameters or of the VDW and coulomb interactions, in the first parameter set, a cut off of 1.0 nm was used for both of the VDW and the coulomb interactions. In the second parameter set, a cut off 1.4 was used for the coulomb interactions, whereas a switch from 1.4 nm to 1.5 nm was used for the VDW parameters.  Under the first parameter set, my system could run about 20 ns per day, whereas for the second parameter set, my system only run 7-8 ns per day.  So I am wondering if this is only raised because of the cut off parameters? And is this a normal pheonomen?
>

Yes, it's normal.  The most important thing to know is you need to use the 
parameters for which the force field was designed.  Lipids are extremely 
sensitive to changes in the cutoff.  Varying these values is very hazardous, so 
if you're doing something non-standard, you'd better have a lot of good proof 
that what you're doing isn't going to lead to artifacts.

-Justin

> I am listing the mdp file at the end of this email. Please note the only relevarent differences between these two parameter sets are the cut off differences.
>
> With my best regards,
> rxgu
> --------------------------------------------
> the first parameter set:
>
> title          = NPT Equilibration
>
>
> integrator     = md
> nsteps         = 25000000
> dt             = 0.002
> nstxout        = 0
> nstvout        = 0
> nstenergy      = 1000
> nstlog         = 2000
> nstxtcout      = 2000
> xtc-precision  = 10000
>
> constraint_algorithm = lincs
> constraints    = all-bonds
> lincs_iter     = 1
> lincs_order    = 4
>
> ;cutoff-scheme  = Verlet
> ns_type        = grid
> nstlist        = 10
> rlist           = 1.0
>
> coulombtype    = PME
> rcoulomb       = 1.0
>
> vdwtype           = cut-off
> rvdw           = 1.0
> DispCorr       = EnerPres
> pme_order      = 4
> fourierspacing = 0.12
> optimize_fft   = yes
>
> pbc            = xyz
>
> tcoupl         = V-rescale
> tc-grps        = POPC SOL
> tau_t           = 0.5   0.5
> ref_t           = 323  323
>
> pcoupl         = Berendsen
> pcoupltype     = semiisotropic
> tau_p           = 10.0 10.0
> ref_p           = 1.0  1.0
> compressibility = 4.5e-5 4.5e-5
>
> gen_vel         = yes
> gen_temp        = 323
> gen_seed        = 82907543
>
> comm-mode       = Linear
> comm-grps     = system
> ----------------------------------------------------------------
> the second parameter set:
>
> title          = NPT Equilibration
>
>
> integrator     = md
> nsteps         = 25000000
> dt             = 0.002
> nstxout        = 0
> nstvout        = 0
> nstenergy      = 1000
> nstlog         = 2000
> nstxtcout      = 2000
> xtc-precision  = 10000
>
> constraint_algorithm = lincs
> constraints    = all-bonds
> lincs_iter     = 1
> lincs_order    = 4
>
> ;cutoff-scheme  = Verlet
> ns_type        = grid
> nstlist        = 10
> rlist           = 1.4
> rlistlong      = 1.7
>
> coulombtype    = PME
> rcoulomb       = 1.4
>
> vdwtype           = Switch
> rvdw           = 1.5
> rvdw-switch    = 1.4
> DispCorr       = EnerPres
> pme_order      = 4
> fourierspacing = 0.12
> optimize_fft   = yes
>
> pbc            = xyz
>
> tcoupl         = V-rescale
> tc-grps        = POPC SOL
> tau_t           = 0.5   0.5
> ref_t           = 323  323
>
> pcoupl         = Berendsen
> pcoupltype     = semiisotropic
> tau_p           = 10.0 10.0
> ref_p           = 1.0  1.0
> compressibility = 4.5e-5 4.5e-5
>
> gen_vel         = yes
> gen_temp        = 323
> gen_seed        = 82907543
>
> comm-mode       = Linear
> comm-grps       = System
> --------------------------------------------------------------------------------------
>

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================