Speaker
Description
Bayesian inference efficiently explores high-dimensional theoretical parameter spaces with direct comparisons to experimental data. We present progress towards an analysis framework integrating established tools to enable reproducible, precision calibration of theoretical models in order to ease these comparisons. We use the Rivet toolkit to evaluate theoretical predictions at design points across parameter space, ensuring direct comparability with published experimental measurements. These comparisons provide training data for Gaussian process emulators built with surmise (from the BAND framework), enabling rapid interpolation across the full parameter space. The emulators are then incorporated into Bayesian calibration workflows using the bilby inference library, facilitating robust parameter estimation and uncertainty quantification. We demonstrate the framework's capabilities using PYTHIA 8 for proton-proton collisions, validating both functionality and computational performance. We compare our results to the Detroit Tune. The modular design enables straightforward extension to heavy-ion collision models, additional observables, and alternative theoretical frameworks. This work establishes a flexible foundation for systematic Bayesian studies, with future applications targeting precision extraction of heavy-ion collision properties and the fundamental characteristics of nuclear matter under extreme conditions.
| Is this an experimental talk? | No |
|---|---|
| Is this on behalf of a collaboration? | No |
| Are you willing to present as a poster if it is not selected for oral presentation? | Yes |