gvm-combinations 0.1.0

Gamma Variance Model for combining correlated measurements

GVM Combinations

gvm-combinations provides tools for combining correlated measurements with errors-on-errors using the Gamma Variance Model.

It supports:

• statistical covariance matrices and correlated systematic uncertainties;
• dependent and independent errors-on-errors;
• likelihood fits with iminuit;
• Bartlett-corrected confidence intervals and goodness-of-fit statistics;
• YAML configuration files and combination summary plots.

Installation

The package requires Python 3.11 or newer.

python -m pip install gvm-combinations

Example

from gvm_combinations import GVMCombination, build_input_data

data = build_input_data("combination.yaml")
combination = GVMCombination(data)
fit = combination.fit()
low, high, _ = combination.confidence_interval()

print(f"mu = {fit.mu:.4f}")
print(f"68.3% CI = ({low:.4f}, {high:.4f})")

The public Python namespace is gvm_combinations. Combination inputs are defined in YAML files containing the measurements, statistical covariance and systematic uncertainty model.

Documentation and examples

The GitHub repository contains the full mathematical overview, configuration reference, example runs and tutorials:

• Toy GVM tutorial
• Top-mass combination tutorial

Methodological details are available in arXiv:2407.05322.

Citation

Citation metadata are provided in CITATION.cff.

License

GVM Combinations is distributed under the MIT License.