Bootstrap-determined <math display="inline"><mrow><mi>p</mi></mrow></math> values in lattice QCD

INSPIRE: 2829225
arXiv: 2409.11379
DOI: 10.1103/PhysRevD.111.074514

Authors: Christ, Norman, Eranki, Rajiv, Kelly, Christopher

Submitted: 17 September 2024

Subjects:

Journal reference: Phys.Rev.D 111 (7) 074514 (2025)

Citations: 4

Abstract

We present a general method to determine the probability that stochastic Monte Carlo data, in particular those generated in a lattice QCD calculation, would have been obtained were that data drawn from the distribution predicted by a given theoretical hypothesis. Such a probability, or <math display="inline"><mi>p</mi></math>-value, is often used as an important heuristic measure of the validity of that hypothesis. The proposed method offers the benefit that it remains usable in cases where the standard Hotelling <math display="inline"><msup><mi>T</mi><mn>2</mn></msup></math> methods based on the conventional <math display="inline"><msup><mi>χ</mi><mn>2</mn></msup></math> statistic do not apply, such as for uncorrelated fits. Specifically, we analyze <math display="inline"><msup><mi>q</mi><mn>2</mn></msup></math>, defined as the correlated <math display="inline"><msup><mi>χ</mi><mn>2</mn></msup></math> statistic obtained using an arbitrary covariance matrix estimator, and show how to use the bootstrap as a data-driven method to determine the expected distribution of <math display="inline"><msup><mi>q</mi><mn>2</mn></msup></math> for a given hypothesis with minimal assumptions. This distribution can then be used to determine the <math display="inline"><mi>p</mi></math>-value for a fit to the data. We also describe a bootstrap approach for quantifying the impact upon this <math display="inline"><mi>p</mi></math>-value of estimating population parameters from a single ensemble of <math display="inline"><mi>N</mi></math> samples. The overall method is accurate up to a <math display="inline"><mn>1</mn><mo>/</mo><mi>N</mi></math> bias which we do not attempt to quantify.