RESEARCH SUMMARY

hcp_combination_126_v_solidThe Nobel Prize in Physics 2013 was awarded jointly to François Englert and Peter W. Higgs “for the theoretical discovery of a mechanism that contributes to our understanding of the origin of mass of subatomic particles, and which recently was confirmed through the discovery of the predicted fundamental particle, by the ATLAS and CMS experiments at CERN’s Large Hadron Collider”

Science Higgs Breakthrough of the yearThe discovery of the Higgs boson required sorting through more than a quadrillion high-energy proton-proton collisions at the Large Hadron Collider.  The ATLAS and CMS experiments developed a number of classification algorithms to identify collisions that were consistent with the production and decay of a Higgs boson. The discovery required bringing together several of these individual different categories of events into a single coherent and comprehensive statistical analysis.  This was achieved through a collaborative statistical modeling paradigm realized via the RooFit and RooStats software tools.

Teams of 20–200 scientists distributed around the world addressed each of these production & decay modes.  They used a theory of fundamental particle interactions and a detailed detector simulation together with parametric and non-parametric modeling techniques.  The tools allow us to combine the statistical models developed for these individual different categories of events into a coherent and comprehensive statistical model (visualized in the figure).  Of course, systematic uncertainties of our complicated detectors must be modeled, too. Because we use the same detector for each of the individual searches, we must take special care of correlated systematic uncertainties.

Since introducing the RooFit/RooStats collaborative statistical modeling tools, the growth of model complexity has grown roughly exponentially in several metrics (see figure).  The growth in model complexity is leading to new statistical and computational challenges.

This technology provides the ability to publish complex statistical models; enabling reproducibility and a broader range of reuse by the community.

Simplified forms of the ATLAS Higgs likelihoods have been published, assigned DOIs, and are now being used and cited by others.

These published likelihoods have vastly improved the ability for theorists outside of the collaborations to reproduce our high-level inference on the properties of this particle and enhanced the scientific discourse around profound discovery.

 

 

 

 

EvolutionOfModels

4l-FixedScale-NoMuProf2

Moore Sloan Poster Research Project