[gmx-users] flexible water vs. rigid water

[Erik Lindahl]

Hi Rui,


On Thursday, September 11, 2003, at 08:28 PM, qiao rui wrote:

> Hi, Gromacs-er:
>
> 	I am curious whether the differnt water model can give different
> dynamic properties, especially viscosity? In particular, which is 
> better
> in terms of providing good viscosity - flexible water or rigid water?
>
> 	In addition, what is the major advanges of using flexible water
> rather than rigid water molecules? (rigid one is easier to implement,
> right?)
>
> 	Is there a flexible SPC/E water model? If there is, how to include
> it in the Gromacs simulation?
>

Water is difficult, and I do not claim to be a world-leading authority 
on the subject - let's see if other people fill in :-)

Some comments, though:

1. Normally, bond constraints are good since they are a much better 
approximation of the quantum mechanical ground state than a classical 
harmonic oscillator (flexible bond).
2. Water is a bit special, partly because we also keep the angle fixed, 
but also because flexible bonds can help give the molecule 
polarizability.
3. Models are usually parameterized for a specific case: either 
flexible or fixed. While you might improve one of the properties by 
using the other setup, the overall properties can often be worse.
4. It is almost impossible to minimize energy efficiently with fixed 
waters.
5. You can just add defines = -DFLEX_SPC to get the flexible versions 
of SPC or SPC/E, but it is mainly intented for energy minimization.
6. You might need to use a smaller timestep (1fs) with flexible waters.

Cheers,

Erik