8 differences between the equations in the ReaXFF reference listed in the manual ( Supporting information for the manuscript “A ReaxFF Reactive Force Field for MolecularDynamics Simulations of Hydrocarbon Oxidation” by Kimberly Chenoweth, Adri C.T. van Duin and William A. Goddard, III) and the corresponding equation in the ReaXFF source files, I think those differences are clerical errors, the differences are listed in the attachment. I think all these equations and the derivatives of them should be corrected.
Are you sure what you are implying here? You are basically stating that all simulations with ReaxFF in LAMMPS that have been performed so far are wrong (and there have been quite a few). Do you believe that nobody would have noticed it, and that you are the first one in so many years to find it? Wouldn’t it be more likely that those differences in the equations are just incorrectly transcribed, that the parameter files use different sign conventions than the paper, or that some details have been revised since the original publication?
The key point is that the definition of ReaxFF in LAMMPS follows the reference implementation of ReaxFF in Fortran. As the pair_style reaxff documentation states, the LAMMPS implementation “matches the version of the reference ReaxFF implementation from Summer 2010”. Every published ReaxFF parameter file was fitted using the reference implementation, so the parameters only have meaning together with the functional forms exactly as implemented there. All available ReaxFF codes (the original Fortran version in the git history, PuReMD, LAMMPS’ reaxff and reaxff/kk) agree with each other, not with the printed equations, wherever the two differ. “Correcting” the LAMMPS source to match the printed equations would make LAMMPS disagree with every other ReaxFF implementation and would invalidate every published parameter file.