updated preprocessing and rewriting input equalities into inequalities for LRA