EnQode is a novel amplitude encoding framework that employs symbolic optimization with warm-start centroid learning.

1. In your terminal, run git clone https://github.com/positivetechnologylab/enqode.git to clone the repository.
2. Run cd enqode.
3. Create a virtual environment if necessary (our code uses Python 3.12.3), and run pip install -r requirements.txt to install the requirements.
4. Run either baseline_script.py or enqode_script.py to run the Baseline or EnQode technique on MNIST, Fashion-MNIST, and CIFAR-10 datasets.
   • The variable NUM_QUBITS can be changed to indicate the number of qubits to perform embedding.
5. The results will be stored in either baseline_data/baseline_metrics_sim_{NUM_QUBITS}qubits.csv or enqode_data/enqode_metrics_{NUM_QUBITS}qubits.csv, for the baseline and EnQode technique, respectively.

The requirements and specific versions are provided in requirements.txt. For loading in an IBM Quantum account, the script finds a .env file located at ENV_FILE_PATH and finds the value associated with key IBMQ_TOKEN as the IBM Quantum Token.

The scripts will create folders {DATA_FOLDER_NAME}/{DENSITY_MAT_FOLDER_NAME}, {DATA_FOLDER_NAME}/{GEN_CIRCS_FOLDER_NAME}, and {DATA_FOLDER_NAME}/{RESULTS_FOLDER_NAME} for storing density matrices, generated circuits, and the results from optimization, respectively.

• README.md: Repository readme with setup and execution instructions.
• baseline_script.py: Python script for executing the baseline technique on datasets.
• enqode_script.py: Python script for executing the EnQode technique on datasets.
• requirements.txt: Requirements to be installed before running the Python scripts.
• noise_model_ibm_brisbane.pkl: The noise model used in the scripts for noisy simulation.
• sampled_dataset_indices.csv: The indices of the datapoints used when evaluating the baseline and EnQode technique in the paper.
• utils.py: Python file for utility functions that are shared between the baseline and EnQode techniques.

Copyright © 2025 Positive Technology Lab. All rights reserved. For permissions, contact ptl@rice.edu.