I wish to know the simple relationship between 'numstep' and simulation time. I mean, if I wish to run a one nanosecond simulation what should be the value that I should put in front of the numstep line of my configuration file? Secondly I want to know if the 'numstep' value is dependent on the 'timestep' value.
For example, if in my configuration file my timestep value is set to 1.0, then would setting the 'numstep' value to 1500000 suffice to yield a 1.5 Ns simulation?
In such a case if I change the timestep value to 2.0, is it necessary to double the numstep value to 3000000? I just want to be sure that by putting these values in a configuration file I am indeed doing the simulation on the desired time scale.
For convenience here is my configuration file details that I think is for 1.5 NS run. MD cons are humbly requested to see if I am doing it right.
##############################################
#### input topology and initial structure ####
##############################################
structure ionized.psf
coordinates topo_equil010.coor # Reading the final structure from
bincoordinates topo_restart_equil010.coor # equilibration stage in a binary format
#binvelocities protein_restart_equil010.vel # Initial velocities from restart file
###########################
#### force field block ####
###########################
paratypecharmm on
parameters par_all27_prot_na.prm
exclude scaled1-4
1-4scaling 1.0
dielectric 1.0
##############################################
#### dealing with long-range interactions ####
##############################################
switching on
switchdist 8.0
cutoff 12.0
pairlistdist 13.5
margin 2.5
stepspercycle 20
rigidBonds all
rigidTolerance 0.00001
rigidIterations 100
#######################################################
#### this block specifies the Ewald electrostatics ####
#######################################################
PME on
PMETolerance 0.000001
PMEGridSizeX 96
PMEGridSizeY 120
PMEGridSizeZ 98
####################################################################
#### this block specifies the parameters for integrator and MTS ####
####################################################################
timestep 1.0
fullElectFrequency 4
####################################################################
######## this block determines the output ##########################
####################################################################
outputenergies 1000
outputtiming 1000
binaryoutput no
outputname topo_quench010 # The file name (without extension) to which final
# coordinates and velocities are to be saved
# (appended extensions are .coor or vel)
restartname topo_restart_quench010 # The file name (without extension), which holds
# the restart structure and velocities
# (appended extensions are *.coor or *.vel)
restartfreq 10000
binaryrestart yes
DCDfile topo_quench010.dcd # Trajectory filename (binary file)
dcdfreq 1500
####################################################################
####################### MD protocol block ##########################
####################################################################
seed 3010 # Random number seed used to generate initial
# Maxwell distribution of velocities
numsteps 1500000 # Number of integration steps during
# production simulations
temperature 300 # see also (b)
####################################################################
######## this block defines periodic boundary conditions ###########
####################################################################
cellBasisVector1 93.08 0.0 0.0
cellBasisVector2 0.0 118.56 0.0
cellBasisVector3 0.0 0.0 96.30
cellOrigin 35.17000961303711 -1.7790613174438477 -14.380327224731445
wrapWater on
Rohan,
The relationship between simulation time and the time step is linear:
timestep x numsteps = simulation time
timestep and numsteps are independent.
If you double the value of time step, you have to halve 'numsteps'.
Your configuration file has 1 fs timestep and 1500000 steps, meaning that the total simulation time will be:
1 fs * 1500000 steps = 1500000 fs = 1.5 ns
In NAMD configuration file, 'timestep' is in femtoseconds = 1e-6 ns. Usually, timestep is either 1 or 2 fs. If you choose to run simulations with 2 fs timestep, you have to make bonds to hydrogens rigid with 'rigidbonds all'. Don't use timestep greater than 2 fs.
Looking through your configuration file, I would recommend the following changes:
1. Remove 'dielectric 1.0'
2. Remove 'margin 2.5' - it is needed in NPT simulations only and decreases performance.
3. Remove 'rigidTolerance 0.00001' and 'rigidIterations 100'
4. Remove 'PMETolerance 0.000001', 'PMEGridSizeX 96', 'PMEGridSizeY 120', 'PMEGridSizeZ 98'
5. Add 'pmeGridSpacing 1.0'
6. Add 'useGroupPressure yes' (since you use rigid bonds)
7. Optional (recommended). Since you make bonds between heavy atoms and hydrogens rigid ('rigid bonds all'), you can also use 2 fs timestep ('timestep 2.0').
You can also improve performance by lowering values of cutoff and pairlistdist, e.g. 10 A cutoff ('cutoff 10') and 12 A pairlistdist ('pairlistdist 12.0').
I would also recommend specifying 'nonbondedfreq' parameter explicitly: 'nonbondedfreq 1'.
The time step in MD production runs is 2 fs (0.002 ps); if you want to run a 1ns (1000 ps) number of steps would be 500000, that correspond to 1000 ps, since:
500000 (step) * 0.002 time (ps/step) =1000 ps= 1 ns
Then the equation for calculation of number of step for X ps is:
X (ps)/0.002 (ps/step)= Y (step)
Rohan,
The relationship between simulation time and the time step is linear:
timestep x numsteps = simulation time
timestep and numsteps are independent.
If you double the value of time step, you have to halve 'numsteps'.
Your configuration file has 1 fs timestep and 1500000 steps, meaning that the total simulation time will be:
1 fs * 1500000 steps = 1500000 fs = 1.5 ns
In NAMD configuration file, 'timestep' is in femtoseconds = 1e-6 ns. Usually, timestep is either 1 or 2 fs. If you choose to run simulations with 2 fs timestep, you have to make bonds to hydrogens rigid with 'rigidbonds all'. Don't use timestep greater than 2 fs.
Looking through your configuration file, I would recommend the following changes:
1. Remove 'dielectric 1.0'
2. Remove 'margin 2.5' - it is needed in NPT simulations only and decreases performance.
3. Remove 'rigidTolerance 0.00001' and 'rigidIterations 100'
4. Remove 'PMETolerance 0.000001', 'PMEGridSizeX 96', 'PMEGridSizeY 120', 'PMEGridSizeZ 98'
5. Add 'pmeGridSpacing 1.0'
6. Add 'useGroupPressure yes' (since you use rigid bonds)
7. Optional (recommended). Since you make bonds between heavy atoms and hydrogens rigid ('rigid bonds all'), you can also use 2 fs timestep ('timestep 2.0').
You can also improve performance by lowering values of cutoff and pairlistdist, e.g. 10 A cutoff ('cutoff 10') and 12 A pairlistdist ('pairlistdist 12.0').
I would also recommend specifying 'nonbondedfreq' parameter explicitly: 'nonbondedfreq 1'.
- Badges
- Science topic
- Similar topics
- Chemistry
- Theoretical Chemistry
- Computational Chemistry
- Molecular Dynamics