Dear Pankaj,
there is no a priori way to compute the optimal values of SIZE & PROB.
The most efficient values are the ones that maximizes the product of
diffusion times the acceptance probability (the latter is an observable,
not PROB, which is the probability this particular move is selected).
Since these are highly dependent on your system, the kind of moves
and the interplay between them, a general rule has been that decent
sampling speed is obtained for an acceptance probability of ~50%. In
my experience, however, this is rarely optimal and for the current
situation without local potential energy updates, an acceptance ~25%
is often more efficient.
With CHARGE the total charge of your system is specified, which
is compensated by a uniform background. As such, this can be
directly „calculated“ ;)
Cheers,
Thomas