“…Alternatively, at the expense of sacrificing performance and accelerating the solution, (V-B) can be sub-optimally tackled [44] with the following iterative approach. Keeping Ļ fixed ā , the optimum w , is obtained ā as w opt , = v max (C , , Ļ Ļ C , , ), while for fixed w , ā , , the optimum Ļ is computed as Ļ opt = exp ( arg(C , , w , )).…”