Deep Reinforcement Learning (DRL) techniques have gained substantial attention in recent years for future wireless networks. They can overcome the ever-increasing challenges of wireless environment and limitations of traditional techniques. Some applications of DRL for future wireless and mobile networks are network optimization, network control, power management, mobile edge computation (MEC), offloading, edge cachings, network slicing and many others. This paper mainly addresses network optimization specifically energy optimization, and secondly resource allocation. A recent literature review is presented by looking at research so far achieved, DRL methods used, and future research directions.