I tried to perform NPT equilibriation on a protein with nose-hoover thermostat and MTTK barostat with the following .mdp file
; VARIOUS PREPROCESSING OPTIONS
title = NPT simulation (constant number, pressure and temperature)
cpp = /lib/cpp
; RUN CONTROL PARAMETERS
integrator = md-vv
dt = 0.001
nsteps = 150000
refcoord_scaling = com
; OUTPUT CONTROL OPTIONS
nstvout = 0
nstfout = 0
nstlog = 1
nstenergy = 10
nstxout = 50 ; save coordinates every 1.0 ps
nstxtcout = 50
xtc_precision = 1000
xtc-grps = System
energygrps = Protein Non-protein
; NEIGHBORSEARCHING PARAMETERS
nstlist = 5
ns-type = Grid
pbc = xyz
rlist = 0.9
; OPTIONS FOR ELECTROSTATICS AND VDW
coulombtype = Reaction-Field
rcoulomb = 1.4
epsilon_rf = 78
vdw-type = Cut-off
rvdw = 1.4
; Temperature coupling
Tcoupl = nose-hoover ; If you use pcouple berendsen tcouple nose-hoover will not work , nose-hoover Parinello-Rahman combination also didnot work
tc-grps = system
tau_t = 5
ref_t = 300
; Pressure coupling
Pcoupl = MTTK
Pcoupltype = Isotropic
tau_p = 1.0
compressibility = 4.5e-5
ref_p = 1.0
; GENERATE VELOCITIES FOR STARTUP RUN
gen_vel = yes ; Assign velocities to particles by taking them randomly from a Maxwell distribution
gen_temp = 300.0 ; Temperature to generate corresponding Maxwell distribution
gen_seed = 9999 ; Seed for (semi) random number generation.
; OPTIONS FOR BONDS
constraints = all-bonds
constraint-algorithm = SHAKE
With this .mdp file I was able to generate .tpr file using grompp command but when I was using mdrun command to start the simulation I found the following error
'SHAKE is not supported with domain decomposition and constraint that cross charge group boundaries, use LINCS
For more information and tips for troubleshooting, please check the GROMACS'
I tried with LINCS but it was complaining that LINCS is not compatible with MTTK
What will be the possible solution for this problem ?
The issue is not so much about the ensemble as it is the algorithms you chose. When generating velocities, the random initial state can lead to instability if using Parrinello-Rahman, as the message suggests. But you're certainly not going to make anything better using MTTK in this regard. For the initial equilibration (i.e. the only time you generate velocities) you can use the Berendsen method, as it converges quickly and is very robust. Then move on to more correct thermostat/barostat combinations when doing actual data collection.
You didn't mention which version of GROMACS you are using. How are you choosing your thermostat and barostats ?
Hi Sayantan ,
I am using GROMACS 5.0 .
I used these thermostat and barostat previously in namd and I found no problem . Nose-hoover is better thermostat than Berendsen because it is known that Berendsen can't produce canonical sampling .
The ERROR you are saying -- 'SHAKE is not supported with domain decomposition and constraint that cross charge group boundaries, use LINCS For more information and tips for troubleshooting'-- I think it is not due to NH and MTTK mismatching.
The keyword "SHAKE" doesn't allow constrains between atoms on different nodes, thus it can not be used with domain decomposition when inter charge-group constraints are present.
So, You can't do domain decomposition with SHAKE in parallel environment. If you want to use SHAKE, run in serial (i.e, single processor).
"LINCS is not supported with MTTK" - Is a different issue. Needs to be cultivated.
Not every combination of input options is supported. There are issues with how MTTK iteratively solves constraints and therefore LINCS is not supported.
The solution is to use Parrinello-Rahman as the barostat and/or the regular md integrator, not md-vv.
Hi Sayantan ,
Thank you very much for your answer . I shall try to use 'SHAKE' in single processor .
Hi Justin ,
When I was using nose-hoover thermostat and Parrinello-rahman barostat with regular md integrator I was getting warning that -
'You are generating velocities so I am assuming you are equilibrating a
system. You are using Parrinello-Rahman pressure coupling, but this can
be unstable for equilibration. If your system crashes, try equilibrating
first with Berendsen pressure coupling. If you are not equilibrating the
system, you can probably ignore this warning.'
I was performing NPT equilibriation step. What thermostat and barostat should I use in a NPT equilibration step ?
The issue is not so much about the ensemble as it is the algorithms you chose. When generating velocities, the random initial state can lead to instability if using Parrinello-Rahman, as the message suggests. But you're certainly not going to make anything better using MTTK in this regard. For the initial equilibration (i.e. the only time you generate velocities) you can use the Berendsen method, as it converges quickly and is very robust. Then move on to more correct thermostat/barostat combinations when doing actual data collection.
- Badges
- Science topic