“…In current implementations, auxiliary variables are not handled differently by SAT solvers. However, the special handling of these variables may be considered in the future, as it has been recently shown that not branching on these variables, as well as not learning constraints with these variables, leads in general to more robust performances [15]. Besides the basic model outlined above, SAT-based haplotyping requires the inclusion of a number of effective techniques, including lower bounds and identification of symmetries [11] (see Section 4.3).…”