回复： [gmx-users] Re: Some questions on Tabulated Dihedral Potential

[hong bingbing]

Hi, Ran,

The potential f(x) and force -f'(x) in the table are calculated by myself before constructing the table. The potential can be written in an analytical form and the force is calculated as the analytical negative derivative of the potential at point x. There should not be so large deviation betw. the value I supplied and the value calculated by GROMACS. Wait. Is there something wrong in my interpretation? What's  your method to get the force? Is there a way to see the numerical derivative generated by GROMACS?

Thanks
CH





________________________________
发件人： Ran Friedman <r.friedman at bioc.uzh.ch>
收件人： Discussion list for GROMACS users <gmx-users at gromacs.org>
已发送： 2009/7/21(周二), 上午8:17:11
主题： Re: [gmx-users] Re: Some questions on Tabulated Dihedral Potential

Hi,

Johannes Kamp wrote: 
Hi Cynthia,

I'm also working on including some tabulated functions but I don't have
any simulation yet. Thus I'm not a 'specialist'  in this topic, but I
hope I can help you a little.


Dear
all,  

I tried to include 2 tabulated dihedral potential functions
into
my simulation. But it seems to be not able to generate correct results.
The system just exploded. I defined 3601 points in each table (from
-180 to 180 with an increment 0.1). After 'mdrun', GOMACS generates two
warning information:

WARNING: For the 3598 non-zero entries for table 0 in
table_d1.xvg the forces deviate on average 193% from minus the
numerical derivative of the potential
WARNING: For the 3598 non-zero entries for table 0 in
table_d2.xvg the forces deviate on average 193% from minus the
numerical derivative of the potential

 
Do these two warnings matter very much? I checked the values
for
x,f(x),-f'(x) in the tables and don't think there're mistakes. has
anyone of you met with such problems? 

In the tables you should supply potentials and forces. The forces are
also calculated numerically, and a warning is emitted if the numerical
derivatives deviate too much from the forces supplied by the user.
Since the forces are interpolated, if the numerical derivative is too
far from the input what you'd get will be different then expected. You
can calculate the derivatives yourself and see where there are
deviations.

Good luck,
Ran



      ___________________________________________________________ 
  好玩贺卡等你发，邮箱贺卡全新上线！ 
http://card.mail.cn.yahoo.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20090721/8d5d086c/attachment.html>