1. The phrases in the manual you mentioned implies that in the cases of C Si
C and Si C Si, the Tersoff code will ignore whatever the values are in the
place for two-body interactions?
Yes
2. For such entries as C Si C and Si C Si, the parameters for two-body
interaction are same with corresponding values in C Si Si or Si C C because
the bond between 1st and 2nd atom are C-Si, equivalently Si-C? If so, the
required values for the two-body interaction bring from other entries such
as C Si Si or Si C C?
The parameters following C Si C specify the values of the threebody
parameters m, gamma, lambda3, c, d, and costheta0, and the cut-off paramters
R and D which are needed. All these parameters appear in the sum over k that
defines zeta_ij. Hence, they are depend on the types of atoms i, j and k.
Conversely, the twobody parameters n, beta, lambda2, B, lambda1, and A only
depend on the types of atoms i and j. For historic reasons, we chose to
handle this by arbitrarily defining the two-body parameters in the line
where j = k.
It would have been less confusing to the new user if we had defined all the
twobody parameters in lines starting with two elements, say C Si, followed
by values for n, beta, lambda2, B, lambda1, A, R and D, and all the twobody
parameters in line starting with three elements, say C Si Si, followed by
values for the threebody parameters m, gamma, lambda3, c, d, costheta0, R
and D. But the existing scheme is more straight-forward to code up and is
actually simpler to use as well, once you understand it.
Aidan