Why uncorrected sigma bond order is enhanced with coefficient `cutoff`?

[shishaochen]

According to equations of ReaxFF, the weight of sigma bond order is 1. However, JAX-ReaxFF add a coefficient cutoff.

ehulp = (1 + cutoff) * np.exp(bop1 * rh2)

Related codes:

• https://github.com/cagrikymk/JAX-ReaxFF/blob/master/jaxreaxff/helper.py#L164
• https://github.com/cagrikymk/JAX-ReaxFF/blob/master/jaxreaxff/reaxffpotential.py#L1080
• https://github.com/cagrikymk/JAX-ReaxFF/blob/master/jaxreaxff/reaxffpotential.py#L1133

Could you please tell the reason?

[cagrikymk]

My main references are the Fortran implementation (written by Dr. Adri van Duin) and the PuReMD code. Since these implementations use that formula, I stick to it as well.

Here is a reference to compare (from the Fortran code): https://github.com/CFDEMproject/LAMMPS/blob/master/lib/reax/reax_connect.F#L742

[shishaochen]

@cagrikymk Got it. Thx.