Skip to content

Multichannel Double Recursive Frequentist-Bayesian Particle Identification algorithm in HEP

License

Notifications You must be signed in to change notification settings

mieskolainen/DREM-PID

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

28 Commits
 
 
 
 
 
 

Repository files navigation

DREM-PID

License: MIT

Multichannel Double Recursive Frequentist-Bayesian Particle Identification (pion, kaon, proton, electron...) in High Energy Physics

A fully working proof of concept C++/ROOT code, to be written into a C++ class.

The code utilizes preprocessed ionization dE/dx and TOF (time of flight) information in terms of independent Gaussian nsigmas -> independent Gaussian likelihoods for different particle hypotheses per track, but is extendable for any kind of (multidimensional) likelihood based input. The Expectation Maximization (EM) algorithm is first used for individual particle abundancies and then in second phase for final state channel abundancies (pi+pi-, K+K-, ppbar etc.), differentially in transverse momentum (or total 3-momentum). This is a frequentist fit approach, but also a Bayesian approach is possible via prior distributions.

Finally, event by event decay channel probabilities can be used with different schemes such as Maximum (Posteriori) Probability classification, event by event weights or as cuts, with suitable frequentist ROC-working (false positive, true positive) points.


mikael.mieskolainen@cern.ch, 2017

About

Multichannel Double Recursive Frequentist-Bayesian Particle Identification algorithm in HEP

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages