Re: [AMBER] 3D-RISM in NAB AmberTools: what is tolerance and initial guess

From: Tyler Luchko <>
Date: Fri, 2 Jul 2010 09:39:18 -0400

On 2010-07-01, at 12:09 PM, Andrey Frolov wrote:

> Dear NAB and 3D-RISM Developers,
> I would like to run some simulation with 3D-RISM, and after reading the
> Manual for the AmberTools and playing with the program, several
> questions appeared:
> 1.
> How to run one point calculation of 3D-RISM in NAB so that to have the
> solvent distribution functions saved on the disk?
> I have to run 1 step of MD simulation with 0 K temperature and zero
> velocities on atoms to get 3D-distributions - so actually i calculate
> the same conformation twice. The only way to save 3D functions, that is
> written in AmberTools manual, is to use ntwrism key:
> mm_options("ntwrism=1");
> But i have to run at least one step of MD or energy minimization (2
> energy evaluations) to save functions.

Unfortunately, single point calculations as you describe cannot currently be done in NAB. 3D-RISM is set up to output distributions the same way NAB outputs trajectories. So, no distributions are output on the zeroth step.

> 2.
> Does the 3D-RISM take the solution of the previous iteration as an
> initial guess for next iteration of MD or energy maximization?

Yes. This is controlled by 'npropagate' which defaults to using up to five previous solutions.

> If so, why when i run MD where the coordinates of a molecule do not
> change at all, the initial error (tolerance) of 3D RISM is about 10-3 in
> my case, but the previous iteration converged to 10-11?
> See the NAB input at the end of the message.

Even though you have the 3D-RISM forces turned off forces from the force field are still applied and the atoms do move. Very small movements in atoms can lead to relatively large increases in the residual. There is a very large solvent density gradient in the solvent accessible regions of the solute, 'moving' this to the new atom positions results in a large residual.

Hope this helps,

AMBER mailing list
Received on Fri Jul 02 2010 - 07:00:04 PDT
Custom Search