Mod:Hunt Research Group/control basics

control file basics

title
io output my_output_file

ensemble nst hoover 0.1 1.0
integration        velocity verlet
temperature      300.0
pressure            0.001

timestep            0.005
steps                 2000
equilibration      500
#restart               noscale

rcut                          10.0
rpad                         0.1
ewald precision       1.0e-06
ewald evaluate        4

vdw mixing 	      Lorentz
rvdw                      8.0
vdw                       shift

print               100
stats               100
rdf                  10
binsize           0.05
print rdf
traj                 501 100 0
dump              1000


job time          3600
close time         100
finish

• block 1

ensemble nst hoover 0.1 1.0
integration velocity verlet
temperature      300.0
pressure            0.001

set the ensemble and any coupling constants

do not use berendsen, can use hoover if far from equilibration but not ideal, best is langevin

two numbers are the thermostat and barostate relaxation times in ps

berendsen is a simple scaling of the temperature dependent on the difference to the target temperature

it is good for getting quickly to the target temperature

nose-hoover includes an artificial variable s with mass which can have a velocity and thus serves to absorb and release kinetic energy (temperature), the coupling is via the mass of s

decide on the integrator VV=velocity verlet and LF=leapfrog

which is better!

temperature, in Kelvin pressure, in k-atmospheres so 0.001=1atm

• block 2 equilibration and run duration
• pre-equilibration

timestep            0.005 pico-seconds
steps                 2000
equilibration      500
scale        10
cap 		     10000 kBT/Å

• ready to run

timestep            0.005 pico-seconds
steps                 2000
equilibration      500
restart               noscale

timestep is in pico-seconds so 1fs is 0.001 ps so the above is a 5fs timestep steps, how many steps to do 5fs*100=1ps, 5fs*100,000=1ns equilibration, is how many timesteps to equilibrate default x=0 scale, scale temperature (velocities) every n steps in equilibration, used only to get the system into a good starting point, then a second equilibration should be carried out cap, is an option to cap forces during the equilibration procedure, f is a max and defaults to 10000 kBT/Å restart, restarts from the endpoint of a previous run with no scaling of the temperature (velocities) reading the REVOLD file

• block 3 system electrostatics

rcut                          10.0 Å
rpad                         0.1 Å
ewald precision       1.0E-06
ewald evaluate        4

rcut, sets the long range interactions cutoff

maximum half box size, advantages to making this smaller as simulation runs faster

a minimum value 3-4 lj radius, but ideally this should be larger ... one should also set this to the same value as that for which the potential was developed!

rpad, dlpoly works on nearest neighbour lists, but when two particles have moved more than rpad the nearest neighbour list is rebuilt, rule of thumb rpad=1-5%*rcut, rcut=10 then 1%=0.01, 5%=0.05 in dlpolyclassic this is the delr command, but delr x =rpad 4x, so if delr=0.5 then rpad=0.5/4=0.125 ewald evaluate, evaluates the k-space contributions to the ewald sum every n timesteps ewald precision, calculate electrostatic forces using ewald sum, automatic parameter optimisation, default is 10*-20

• block 4 system vdw

vdw mixing 		Lorentz
rvdw cutoff         	 8.0 Å
vdw shift

vdw mixing, apply the defined mixing rules for the vdw cross terms when they are not already specified rvdw cutoff, set short range interaction cutoff to x Å vdw shift, shift the vdw interactions (PES) so that they scale smoothly to zero at the cut-off r_cut

• block 5 saving and printing data

print 100
stats 100
rdf                 10
binsize           0.05 Å
print rdf
traj                 501 100 0
dump             1000

print, print every x steps stats, accumulate statistics every x steps

storage requirements are an issue, files can get very large, so set these 100-200 to start with

rdf / print rdf / binsize, collect and print the rdf, with bin size of xÅ traj, writes a HISTORY file, first number is when to start collecting, second number is the intervals to collect, third number is the data level 0=coords only, 1= coords and velocity, 2 =coords, velocity and forces

here we start recording after equilibration

this stores the trajectory and will influence the rdf evaluated by outside software

for water use 10 but water is dynamic has low viscosity and moves very quickly

for ionic liquids which are viscous use a much larger number say 100 or larger.

dump, sets the restart data dump interval to every x steps default is 1000

• block 6 set max time

job time          3600
close time         100
finish

close time, is the time a job has to finalise I/O before the job is killed job time, is the max time for the job, time in seconds 3600s=60min=1hr