[gmx-users] Heating schemes

[X.Periole]

Hi Mark,

> MD protocols in papers show a wide variety of heating 
>schemes, including heating from zero K in stepwise 
>fashion with a certain amount of MD time at various 
>temperatures, and just jumping straight in the deep end 
>at the simulation temperature and running for a while to 
>hopefully enter the equilibrated region of phase space.
> 
> While I can see the intellectual attraction of a gradual 
>heating scheme, the chaotic nature of the integrators 
>makes me suspect that it doesn't make much of an impact. 
>Is anybody aware of meaningful differences induced by 
>these schemes? Are there any published studies comparing 
>them?
> 

I don't think anybody has ever published such study. It is
not a real problem. The heating schemes were used in the
early days to avoid "large" deformation of the protein 
when
it was given 300K velocities. The smulations at hat time
were 100-300 ps. It was thus necessary to minimize its
deformation because it was then impossible to recover 
within
the simulation time.
The position restrains is another, and to me equivalent,
scheme which allows the initial set of velocities to relax
within the context of the system (and vis versa) and avoid 
the loss of important interactions in the system.

Best
XAvier