[gmx-users] spacing distance in tabulated angle interaction function

[Mark Abraham]

wuxiao wrote:
> Dear GMXers,
>   I want to use tabulated interaction functions for bonds and angles. 
> According to the manual, f(x) and f'(x) should be computed as functions 
> of x in spacing distance of dx. For bonds, the spacing distance can be 
> set as 0.002 (in single precision) at default. But for angles, what 
> shoud be set for the spacing distance? Please give me some hints. Thanks 
> a lot for your attention!

It barely matters :-) GROMACS is probably using the difference between
adjacent values of the derivatives to estimate the second derivative for
a second-order Taylor expansion, and so as long as the spacing is small
enough that this estimate is valid, you're fine.

Mark