Re: [ESPResSo-users] failed to tune P3M paremeters to required accuracy

[Stefan Kesselheim]

Hi,

On Feb 24, 2014, at 4:18 PM, roya moghaddasi <address@hidden> wrote:

> Dear Stefan,
> What do you mean by saying "In every increment, you should simulate a bit." 
> and "By feeling I'd say 30 increments of 0.1 and 100 MD steps in between 
> could be enough."? Do you mean I should integrate after each coulomb 
> interaction like this:
> set bj 0.001
> for {set b 1} {$b<3000} {incr b} {
> set bj [expr $bj+0.001]
> puts [inter coulomb $bj p3m tune accuracy $accuracy]
> integrate 100
> }
> Could you please explain it more?

Exactly. In every step you make the electrostatic interactions somewhat 
stronger, and allow for an adjustment to that change. This way you (hopefully) 
do not rip your chains apart by a sudden onset of a strong repulsive Coulomb 
interaction, but they can gradually stretch, while the interactions are 
switched on.

My guess was that 30 steps where lb is increased by 0.1 could already be enough.

Note that you definitely need to check the equilibration, because it is likely 
that your system needs quite some time to reach thermal equilibrium.

Cheers
Stefan