a reinforcement learning algorithm for cryptocurrency mining

In this article, we introduce a stochastic differential equation for cryptocurrency mining and the rewards each miner can receive is modelled by winer process. then we solve stochastic differential equation in continuous and discrete form, which is introduced the control(action) as hash rate (computing power)and the wealth as state of the system. our goal is to find the policy that maximizes the expected return. using reinforcement learning and neural networks, we approximate the weights of the state-action value function. the obtained numerical results show that the weights of the state-action value function in analytical and approximate mode have a small error.