Investigating the Effects of Hyperparameters in Quantum-Enhanced Deep Reinforcement Learning

Algorithm 1

Quantum enhanced deep Q-learning.

	Set replay memory M to state size N
	Initialize action-value function quantum circuit Q with arbitrary parameters θ
	For episode e = 1, 2, 3, 4, ……. E do
	Initialize State s1 from the set state S and encode it into
	the quantum state using basis encoding
	for the time step t = 1, 2, 3, …. T do
	With probability ε, select a random action at
	otherwise, select the optimal action at from the result of quantum circuit
	Execute the selected action at and see the reward rt and the next state st+1
	Store transition in replay memory M
	Sample a random minibatch of transitions from the replay memory M
	Perform a gradient descent step on
	end for
	end for