\nNote: You can try other complex boundary condition but we do not \nrecommend to do so. The main reason is that FWI is to recover \nthe low-frequency information of the earth model. Low-freq \nmeans that exact absorbing is not necessarilly needed. The \nresult will be improved with the optimization precedure. \nFurthermore, complex boundary condition (such as sponge ABC or \nPML) implies more computational cost, which will degrade the \nefficiency of FWI. \n\n