[gmx-users] f(x) g(x) h(x)in the user defined potential functions

[Mark Abraham]

He, Yang wrote:
> Hi all users,
> 
> When I am defining the user potential functions using the table, I encountered a problem that there are several unstable parameters  in the separate f(x),g(x),h(x) .For example, in the g(x), there is a parameter "epsilon" whose value will depend on different pairs.In this situation, I can not get a specific value in the table file . I don't know how to solve this problem.
> 
> Can anyone of you give me some suggestions about that?

So your nonbonded interaction depends on your atom types. You could 
write a table for each interaction type, modify GROMACS to read them all 
in, and then modify the routine that calls the kernels to use the 
correct one. This additional memory usage would grow as the square of 
the number of atom types. You would also see some performance loss which 
you could minimize by arranging to evaluate all of one type of nonbonded 
interaction close together in the nonbonded routines, to minimize cache 
misses.

Better would be if some of these functions (conveniently) differed only 
by a multiplicative or additive constant so that you could re-use the 
same table and then apply a correction function, but you'll have to look 
at your own maths for that.

Mark