Re: collective variable component -- eigenvector usage

From: Giacomo Fiorin (
Date: Tue Mar 27 2012 - 21:13:00 CDT

Hello Chang, the extreme values depend on the way you computed the vector
that you're providing.

Say that you computed the vector V as the difference between two structures
A and B, and you normalized it so that |V| = 1, and chose A as the
reference structure. If we call C the simulated structure, the value is (C
- A) * V. It will give you 0 if C is near A, and 1 if C is near B. Mind
that there is also a correction because V gets centered around zero by the
code, so in this particular example the actual range is -0.5 to 0.5. If
|V| is different from one, you can do the conversion easily.

If you're in doubt, try setting up a test calculation using A as reference
coordinates, and one using B as reference coordinates. The two values of
the eigenvector colvar should give you a good estimate of the actual range
of values.


On Tue, Mar 27, 2012 at 8:46 PM, chang <> wrote:

> Dear Sir/Madam,
> I have been trying to use eigenvector as the coordinate of ABF simulation
> (to sample the space between a starting molecule and an ending molecule).
> But I have trouble on deciding the upperWall/lowerWall and the force:
> upperWallConstant/**lowerWallConstant. Information in the user's guide is
> far from enough. Could you give me some suggestion on choosing these
> parameters?
> (BTW. The molecule in question has approximately 4800 atoms and 66000
> atoms underwater. )
> Your answer would be greatly appreciated.
> Best regards,
> Chang Lu

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:22 CST