This class implements a local conditional constraint with the form of "R(C|L)",
where "L" (a set) contains at least one parent variables of variable C from
the Bayesian Belief Network.
This class implements a local marginal constraint with the form of "R(C)" or "R(C, L)", where "L" (a set) contains
at least one parent variables of variable "C" from the Bayesian Belief Network.
Takes an int which represents the offset in the one-dimensional arrary as argument and
returns the corresponding 'indices' in the multi-dimensional array, by checking the hashtable.