Bayesian estimation of the specific shear and bulk viscosity of the quark-gluon plasma

This repository contains all data necessary to reproduce Bayesian parameter estimates of a computational model of relativistic heavy-ion collisions.

The computational collision model which generated the raw data is available at https://github.com/Duke-QCD/hic-eventgen.  The analysis code which postprocessed the raw data and performed the parameter estimation is available at https://github.com/jbernhard/hic-param-est.

For more information, see the author's dissertation, Bayesian parameter estimation for relativistic heavy-ion collisions (arXiv:1804.06469 [nucl-th]).  In particular, chapter 3 describes the computational collision model, chapter 4 describes the parameter estimation method, and section 5.3 presents the results based on this dataset.

Usage

hic-param-est-cache.tar.gz is designed to be used with the analysis code (https://github.com/jbernhard/hic-param-est).
See the documentation at http://qcd.phy.duke.edu/hic-param-est.  After downloading the code, extract the data archive in the hic-param-est folder.  This will create the cache directory containing the parameter design files, model calculations, and experimental data.  Note that this data archive includes Python pickle files created by joblib, which is included with scikit-learn, a dependency of the analysis code.

alternate-format.hdf contains the same data in HDF5 format.  This is not intended for use directly with the analysis code, but may be more convenient for other purposes.

chain.hdf contains the posterior distribution sample (the "chain") generated by MCMC sampling in HDF5 format.  It may be used with the analysis code by placing it in a subfolder "mcmc" in the hic-param-est folder.  It can also be regenerated by the analysis code, although this can take a long time.