IQP Initializations

Studying how initialization strategies affect the trainability of IQP (Instantaneous Quantum Polynomial) circuits, measured via MMD (Maximum Mean Discrepancy) loss variance and training dynamics.

Structure

Path	Description
mmd_variance_calc.py	Compute MMD² variance across init strategies, scales, and qubit counts
mmd_variance_render.py	Plot results from precomputed variance data
mmd_variance_plots/training.py	Train IQP circuits and record loss curves
correlator_assumption.py	Verify the correlator assumption on genomic data
datasets/genomic/download_data.py	Download genomic SNP data
mmd_variance_config.yaml	Central configuration file
common.py	Shared utilities

Setup

Install dependencies:

pip install numpy jax jaxlib pandas matplotlib scipy pyyaml scikit-learn pennylane iqpopt qml_benchmarks

Usage

# 1. Download genomic data
python datasets/genomic/download_data.py

# 2. Compute MMD² variance
python mmd_variance_calc.py

# 3. (Optional) Run training curves
python mmd_variance_plots/training.py

# 4. Render plots
python mmd_variance_render.py

All scripts should be run from the project root.