[gmx-users] Re: problem of tabulated bonded potential

[Z.Xiao]

XAvier, thanks for your reply.
 the numberical derivative -f'(n)=dy/dx=-(y(n+1)-y(n-1))/(n+1)-(n-1).
so -f'(n)=-(y(n+1)-y(n-1))/2. Is it wrong?


On Nov 9, 2010, at 7:12 AM, Z.Xiao wrote:

> Dear all gmxers,
> I meet some problems when I use the tabulated bonded potential.
> My original function is a sum of An*cos(x)^n (n=1-8).For short here 
> I replaced it by cos(x).
> if f(x)=cos(x),then -f'(x)=sin(x).and the numberical derivative -
> f'(x)=-(y(n+1)-y(n-1))/2.
here you should have dy/dx, re you really doing this?
> but there are great discrepancy between the two -f'(x).
> which is right?
> and if mdrun with the derivative of original function I would met a 
> warning:the forces deviate
> on average 207% from minus the numerical derivative of the 
> potential.but mdrun with the the
> numberical derivative there is no warning.
>
>
> --


 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20101109/20d13bbe/attachment.html>