lattice_data_tools

Lattice Data Tools (LDT) is a Python library for lattice gauge theories. It contains programs to produce and analyse lattice data.

Developer: Simone Romiti. Feel free to contact me for bug reports, desired features, etc.

What can I do with LDT?

lattice_data_tools provides building blocks for the full analysis pipeline of lattice Quantum Chromodynamics (QCD) simulations. Starting from raw Monte Carlo configurations, the library covers:

• Statistical analysis of Monte Carlo simulations histories (autocorrelation, binning, etc.)
• Resampling: bootstrap samples and jackknifes
• Extracting information from lattice QCD correlators:
  • Effective masses, Amplitudes
  • Generalized Eigenvalue Problem (GEVP) for multi-state correlator matrices
• Non-linear least squares fits (with errors on both x and y)
• Nested Sampling algorithm (production [ TODO, available on this library ], analysis)
• Model Averaging (Akaike Information Criterion, error budget of systematic effects)
• Frequently made plots in lattice QCD works
• Gauge configurations, Canonical Momenta associated to the links
• Lattice Convolutional Neural Networks (L-CNNs)

Documentation

The documentation can be automatically generated in ./doc by running the following command:

python generate_doc.py \
	--src ./ \
	--out ./doc \
	--github https://github.com/simone-romiti/lattice_data_tools/

The information is taken from each folder and subfolder, looking at their __init__.py files, as well as in the docstrings of the various .py modules and functions.

Tests

When I add a new feature to the library I usually write also a test script to test the feature using synthetic data. For those, please refer to the ./test/ folder. This is however not possible with complicated analyses like for the $g-2$ HVP and HLbL, where one needs also the data.