Re: [ESPResSo-users] Angle potential 3-body vs 2-body (?) forces

[Jean-Noël Grad]

Dear Edvin,

They are the same. One function returns three vectors, the other returns 
two vectors (the third vector is deduced by addition: 
https://github.com/espressomd/espresso/blob/bd1cce48aef5634fca29db584664760703bdceff/src/core/forces_inline.hpp#L697-L699).

Function calc_angle_cosine_force() is called by forces_inline.hpp, while 
function calc_angle_cosine_3body_forces() is called by 
pressure_inline.hpp. Whenever you define a bond in Python with 
espressomd.interactions.AngleCosine, both functions get called during 
integration (the forces and pressure .hpp files are walked through).

Both functions carry out the same calculation, even though the coding 
style differs. We're currently refactoring bond angles to remove code 
duplication and improve readability. You can see the progress on GitHub 
with #2497, #2542 and #1224. The latest version of the file you attached 
can be found here: 
https://github.com/jngrad/espresso/blob/bb5b3d9bf650b4d556361c2bd4f6a50cf9c64ae0/src/core/bonded_interactions/angle_cosine.hpp

If you need to edit these files, just keep in mind that intermediate 
variables related to angles follow the external angle convention (i.e. 
pi - theta, so the cosine changes sign, but the sine is unaffected). 
This will change in the future.

Hope this helps.

Best,
JN

On 3/26/19 7:21 AM, Memet, Edvin wrote:
> Dear all,
>
> In a file like angle_cosine.hpp, for instance, what is the difference
> between the following functions?
>
> inline int calc_angle_cosine_force(...)
> inline void calc_angle_cosine_3body_forces(...)
>
> The first one seems like it computes the forces on only two of the
> particles rather than on all three particles composing the angle bond.
> If that is the case, then why/how would that work?
>
> Also, I've tried to check which of the two functions is actually being
> called when I define an angle potential. Surprisingly, it seemed like
> only the first function was being called. Is that correct? Is the force
> on the third particle zero, then, or is it being computed elsewhere?
>
> If it's better to ask this in the developer's list, please let me know.
>
> Best,
> Edvin