“…Early theories were aimed to single out the physical origin of the exchange bias by using over simplified AFM spin structures [4,5,6,7]. These models were soon replaced by more sophisticate models which are believed to better represent realistic experimental systems [8,9,10]. In experiment, most of the early measurements are on the FM layer hysteresis loops to study the pinning effect [11,12,13], training effect [14,15,16], and the finite size effect [17,18], etc.…”