Capstone project for Minerva Schools, Yoav Rabinovich, March 2020.
A Conditional Stacked Recurrent Neural Network with GRU cells is used to generate quantum circuits based on desired target quantum states. The conditional network is trained on randomly sampled circuits and their simulated output states as conditions that are introduced into the internal memory state of the initial GRU cell. The network fails to achieve the desired target states, possibly due to the synthesized dataset which lacks correlations in gate placement within each circuit. An analysis of the method and the results is provided, as well as a discussion about possible avenues for refinement, and an overview of the subjects of conditional RNNs and quantum circuits.