A maximum principle for partially observed optimal control of forward-backward stochastic control systems · Study