[gmx-users] initial configuration problems and energy minimization

[Mark Abraham]

Arun kumar wrote:
> Dear gmx-users,
>                        Here I have a some problem regarding initial
> configuration....  My system is composed of  cationic surfactants,
> co-surfactants, water. I am able to minimize the energy of my
> system(random), if it contains 15(surfactants)+ 15(co-surf)+
> 1000(water)+
> 15 (counterions) and also am able to simulate.......
> 
>                       But when I increase the number of surfactants
> or co-surfactants..... Even the energy of the system initially is a
> big negative..... When I start true simulation, Potential energy is
> becoming positive...and ultimately box is exploding [xtc error or some
> times ci error]...... still I tried to minimize the energy of initial
> configuration, but I am getting the same error.
> 
> Below I describe the steps i took to do simulations
> 
> 1)minimize the energy of surfactnat(btm.gro)(includes counterion) and
> co-surfactant(str.gro) in Vaccum
> 
> 2)Generate a system of 100 surfactnats(includes counterions), 20
> co-surfactants and 1000 waters by using  packmol(random)
> 
> 3)Minimize the whole system until potential will become a reasonable 
> negaive
> 
> 4)With that minimized structure I run simulation but always the box is 
> exploding
> 
> My simulation is successful when  the no of surf and co-surf are 15+15
> 
> For this successful simulation Initially i am conducting a NVT
> simulation with 0.01 fs timestep  at 100K temp and then another NVT at
> desired temp(300K) and then NPT with 2fs timestep

This looks pretty reasonable. You should check visually that packmol is 
not generating a conformation with atoms that are too close together. 
You can try using position restraints during your equilibration steps, 
and equilibrating for longer. The explosion is almost certainly 
happening because atoms are too close at some point, and these leads to 
huge energies. If your density is wrong for your NVT, then relaxing to 
NpT too quickly can cause this sort of problem too - use a large tau_t, 
or generate a box closer to the right size initially.

Mark