qaipf

Reproducibility + analysis code for “Quantum Annealing Algorithms for Estimating Ising Partition Functions”.

Programs

qaipf/make_all.py

Generates all figures in the paper (main text + supplement) by running the individual figure scripts and collecting outputs.

qaipf/make_fig2.py

Generates Fig. 2 from instance-level statistics (computed by analyze_instance.py) and/or cached intermediate results.

qaipf/make_fig3.py

Generates Fig. 3 from instance-level statistics (computed by analyze_instance.py) and/or cached intermediate results.

qaipf/analyze_instance.py

Computes analysis quantities for a single problem instance, including:

• instance-dependent statistics (e.g., μ_m and related quantities)
• partition-function estimates Z used in the paper
• weighting/mixing parameters (e.g., MV/KL variants used in the paper)
• divergence metrics (e.g., KL terms used in the paper)
• normalized variance / variance diagnostics used in plots

Figure scripts typically call this logic repeatedly over many instances.

qaipf/build_pnmmat.py

Regenerates and saves Pnm / Pnmmat matrices as .mat datasets (optional; only needed if you want to rebuild the intermediate data rather than using existing .mat files).

Data

The code expects precomputed .mat datasets under a consistent directory layout for Pnmmat (SK and 3SAT variants).