Re: how should I set those PMEGridSizeX,Y,Z?

From: yxiong_at_mail.ccnu.edu.cn
Date: Wed Dec 10 2003 - 18:51:51 CST

Dear Tim,
  Thank you for your explanation and direction!
  I can understand " the number of grid points", but I can't understand the
meaning of "Grid size". Because PMEGridSizeX,Y,Z sets the number of grid points
along cellBasisVector1,2,3 respectively, why can these points decide the size of
grid? Maybe the distance between each point is fixed (for example, 1 angstrom ).
That is to say, For a cellbasisvector of 100, the length of cellbasisvector1 is
100 angstrom, if PMEGridSizeX is 30, then the size of grid is 30 angstrom along
cellBasisVector1, the grid only covers part of the cell. if PMEGridSizeX is 120,
then the size of grid is 120 angstrom along cellBasisVector1, the grid covers more
than all the cell.
  Is my understanding right?

YingXiong

>From: Tim Isgro <timisgro_at_ks.uiuc.edu>
>Reply-To:
>To: <yxiong_at_mail.ccnu.edu.cn>
>Subject: Re: namd-l: how should I set those PMEGridSizeX,Y,Z?
>
>PMEGridSizeX,Y,Z sets the number of grid points along cellBasisVector1,2,3
> respectively. The online NAMD manual says each should have small integer
> factors (2, 3, and 5) for best running speed (but I have no idea why).
> In choosing the grid size, you'll want to balance computational effort with
> accurately representing the charge in your system.
>
> For a cellbasisvector of 100, a reasonable grid size could run from 30
> (less accurate, but faster) to 90 or 120 (more points for charge, slower
> running).
>
> -Tim
>
> On Wed, 10 Dec 2003, wrote:
>
> > Dear sir,
> > When I use PME method in MD, how should I set those
PMEGridSizeX,PMEGridSizeY,
> > PMEGridSizeZ? Whether need these values be larger than cellBasisVector1,
> > cellBasisVector2,cellBasisVector3 respectively ?
> > I am looking forward to your help!
> >
> >
>
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:37:13 CST