[gmx-users] NVE Energy drop in Mixed Precision 4.6.7

[Guillaume Chevrot]

Hi,

You should maybe show us a plot of the energy for more than just a few 
picoseconds.
Nevertheless, from my experience the only way to obtain constant energy 
in the NVE ensemblewas to use double precision.


Guillaume

On 10/06/2014 09:34 PM, Johnny Lu wrote:
> The run was on 12 Xeon CPU with a GPU.
>
> On Mon, Oct 6, 2014 at 3:03 PM, Johnny Lu <johnny.lu128 at gmail.com> wrote:
>
>> here is the energy from nve1.edr
>> http://oi62.tinypic.com/2pqw70l.jpg
>>
>> On Mon, Oct 6, 2014 at 3:01 PM, Johnny Lu <johnny.lu128 at gmail.com> wrote:
>>
>>> Hi.
>>>
>>> The system has 9000 water and a ~160 aa amino acid and uses
>>> amber99SB-ildn.
>>>
>>> After extensive NPT (300ns, 2ns ts), and then NVT equilibrium (300 ns,
>>> 2ns ts), the NVE simulation still has energy drop, despite of the very
>>> small time step.
>>>
>>> Is double precision gromacs the only solution?
>>>
>>> The mdp file is:
>>>
>>> title           = NVT equilibration
>>> ; Run parameters
>>> integrator      = md            ; leap-frog integrator
>>> nsteps          = 29500000      ; 1 ns
>>> dt              = 0.0002        ; 0.5 fs
>>>
>>> ;NVE
>>> ;nstcomm         = 500 ; remove center of mass motion to reduce energy
>>> and temp. drift.
>>> shake-tol       = 0.0000001     ; shake tolerance.
>>> verlet-buffer-drift = -1
>>>
>>>
>>> - Output control
>>> ;nstxout         = 20           ; save coordinates every 1.0 ps
>>> ;nstvout         = 4           ; save velocities every 1.0 ps
>>> ;nstfout                = 20
>>> nstenergy       = 20           ; save energies every 1.0 ps
>>>
>>> nstlog          = 1000          ; update log file every 0.2 ps
>>> ; Bond parameters
>>> continuation    = yes           ; Restarting after NVT
>>> constraint_algorithm = lincs    ; holonomic constraints
>>> constraints     = all-bonds     ; all bonds (even heavy atom-H bonds)
>>> constrained
>>> lincs_iter      = 4             ; accuracy of LINCS
>>> lincs_order     = 4             ; also related to accuracy
>>> ; Neighborsearching
>>> ns_type         = grid          ; search neighboring grid cells
>>> nstlist         = 40            ; 10 fs
>>> rlist           = 1.3           ; short-range neighborlist cutoff (in nm)
>>> rcoulomb        = 1.0           ; short-range electrostatic cutoff (in nm)
>>> rvdw            = 1.0           ; short-range van der Waals cutoff (in nm)
>>> ; Electrostatics
>>> coulombtype     = PME           ; Particle Mesh Ewald for long-range
>>> electrostatics
>>> pme_order       = 6             ; cubic interpolation
>>> fourierspacing  = 0.1           ; grid spacing for FFT
>>>
>>> ; Temperature coupling is off
>>> tcoupl          = no    ; modified Berendsen thermostat
>>> ; Pressure coupling is off
>>> pcoupl          = no
>>>
>>> refcoord_scaling = com
>>> ; Periodic boundary conditions
>>> pbc             = xyz           ; 3-D PBC
>>> ; Dispersion correction
>>> DispCorr        = EnerPres      ; account for cut-off vdW scheme
>>> ; Velocity generation
>>> gen_vel         = no            ; Velocity generation is off
>>> cutoff-scheme = Verlet
>>>
>>> Thank you in advance.
>>>
>>