In the Sturat paper the Vtors has half angle of the cos but looking at the source code of airebo.cpp line number 1035 it is given as: cw2 = (.5*(1.0-cw)). But in order to find the half angle square of the cos shouldnt there be positive sign? Or I am missing something.
The AIREBO code in LAMMPS has been validated against independent implementations multiple times, so we are quite confident that the forces it computes are correct.
I am not very familiar with the details of the implementation, perhaps @athomps has an idea.