equilibration process
- 1. steepest descent minimization (done)
- 2 NVT equilibration (700K) and maybe NVE check
- 3 NPT equilibration (700K)
- 4 NPT reduce temperature steps 50K
- 5 NPT finish equilibration (400K)
- 6 short (5ns) NVT reset after changes
- 7 NVT production run
- a good equilibration for pure water is 200-500ps
- but for ions in water, or ILs you need 2000-5000ps
- don't want the dynamic properties via NPT because of the forces applied to keep the pressure constant, these directly effect the dynamic properties
- use the NVT for properties
- NVE is possible, but NVT conserves the temp and will correspond to experimental const temp
NVT equilibration (700K)
- copy *.gro output from energy min to confemin.gro
- use gromacs file nvt_eq.mdp
- copy to nvt_eq.mdp
- temp=700, initial velocities
- use V-rescaling for T and Berendsen for P
- run for about 30ps
gmx grompp -f nvt_eq.mdp -c confemin.gro -p field.top
gmx mdrun -v
gmx energy -w
gmx trjconv -f traj_comp.xtc -o traj_out.gro
- copy confout.gro output from second md to nvt_eq.gro, you may see some contraction
- take a look at the final structure confout.gro in vmd, there should not be any strong changes from the minimized structure
- check total energy (12) is stabilising
- look at the temperature (14) it should oscillate around 400 looking for ±20K
- look at the pressure (16) it is unconstrained but should oscillate around 1, values ±1000bar are ok!
- we want a final average pressure of 1Atm=1Bar
- look at the pressure tensor components, Pxx=Pyy=Pzz should be true but there will be variations, size of the differences is an indication of the accuracy, also off-diagonal terms should be zero again small difference from zero indicates accuracy
- seen the Berendsen used for T and P next
- however Berendsen does not generate the correct kinetic energy distribution
- we use the V-scale thermostat
NVE check
- run ≈30ps with no contraints on pressure or temperature ie NVE
- no initial velocities
- the total energy should not drift
- use gromacs file nve_eq.mdp
- if you time step (eg 2fs from 1fs), the oscillations in E should increase
NPT equilibration (700K)
- now we want to reduce the cell to generate the experimental volume
- for gromacs use npt_eq.mdp
- copy to npt_eq.mdp
- restart from end of last run, ie use confout.gro
- run ≈30ps
- temp=700
- no initial velocities?? NOT SURE HERE
- still use V-rescaling for T and Berendsen for P
- you might need to run this a few times ≈3?
- may need to manually increase or decrease box volume, use small steps!
gmx grompp -f npt_eq.mdp -c confout.gro -p field.top
gmx mdrun -v
gmx energy -w
gmx editconf -f confout.gro -o confout.pdb
- check volume (20) should come down and then stabilize
- look at the pressure (15) it is now constrained and should oscillate around 1, values ±1000bar are ok!
- copy confout.gro output from second md to npt_eq.gro
- take a look at the final structure, check the volume ie cell parameters
NPT reduce temperature steps 50K
- change npt_eq.mdp file as required
- reduce from 700K to 400K at 1 Bar
- each run 30ps
- 700->650
- 650->600
- 600->550
- 550->500
- 500->450
- 450->400
NPT finish equilibration (400K)
- for gromacs use npt_eq2.mdp
- copy to npt_eq2.mdp
- restart from end of last run, ie use confout.gro
- run for 30ps
- increase nstlist
- increase coulomb and vdw cut-offs to 1.6
- switch to nose-hoover for T and Parrinello-Rahman for P
- nose and parrinello coupling 0.5ps
- no initial velocities?? NOT SURE HERE
gmx grompp -f npt_eq2.mdp -c confout.gro -p field.top
gmx mdrun -v
gmx energy -w
gmx editconf -f confout.gro -o confout.pdb
- cannot use md (leapfrog) with nose-hoover for T and Parrinello-Rahman for P
- you will see a note: leapfrog does not yet support Nose-Hoover chains, nhchainlength reset to 1
- this not is not important?? WHY??
short NVT to prepare for production (400K)
- use gromacs file nvt_prep.mdp
- 5ns to set up for production run
- turn off pressure coupling
- no initalisation velocities
gmx grompp -f nvt_prep.mdp -c confout.gro -p field.top
gmx mdrun -v
gmx energy -w
NVT production
- extend the time steps from nvt_prep.mdp to make nvt_prod.mdp
- increase time between recorded coords, energy etc
- these should run for at least 20ns
- these will run for a long time!!
gmx grompp -f nvt_prod.mdp -c confout.gro -p field.top
gmx mdrun -v
gmx energy -w
- things to check in the production run
- want to check the temperature and pressure are all stable
- the temperature should oscillate +/- 5-10K around the set target value
- the pressure can fluctuate widely
- the total energy should be conserved, you should see no drift after initial equilibration
