/Supervised-learning-of-quantum-circuits

Relevant scripts and data for the paper "Supervised learning of random quantum circuits via scalable neural networks"

Primary LanguagePython

ML_QC

Relevant scripts and data for the paper entitled "Supervised learning of random quantum circuits via scalable neural networks"

Table of contents

Python scripts

  • "mlqc.py" contains all the important functions. The "main" part of this script can be used to train and test the neural network on quantum circuits of different sizes.

  • "Extrapolation_train_test.py" is used to train a neural network on quantum circuit of a certain size. We can use this neural network to predict expectation values of quantum circuits of different sizes.

  • "Extrapolation_test.py" is used to make prediction on quantum circuits given a pre-train model.

  • "circ_gen.py" generates random quantum circuits with the gates belonging to the set [T, H, CX].

"Dataset" Folder

It contains some of the data used to get the results shown in the paper. One can generate more data by using "circ_gen.py". Input data contains integers in the range [0, 3]. Each integer is related to a quantum gate. In the python scripts, we use the one-hot encoding technique to convert each integer into an array.

"Weights" Folder

This folder can be used to store the weights of the neural network that we want to use again. We have already uploaded a model of a neural network trained on quantum circuits with 10 qubit and 6 gates per qubit. This neural network has been trained using an "EarlyStopping" with "patience = 5". One can decide to use different training protocols.