Composition Classification of Ultra-High Energy Cosmic Rays.

Area of Science:

• High-energy astrophysics and particle physics.
• Cosmic ray physics and detector instrumentation.

Background:

• Determining the primary particle type for cosmic ray events is a long-standing challenge in physics.
• Direct detection of high-energy primary cosmic rays is impossible, necessitating the use of simulations.

Purpose of the Study:

• To evaluate the effectiveness of machine learning classifiers for determining cosmic ray primary types.
• To compare the computational cost and performance of different machine learning models.
• To identify key features for accurate primary cosmic ray event classification.

Main Methods:

• Utilized a simulated dataset generated by the CORSIKA Monte Carlo code.
• Designed, trained, and compared various machine learning classifiers.
• Employed a feature selection algorithm to determine feature relevance.

Main Results:

• Machine learning classifiers demonstrated effectiveness in classifying primary cosmic ray types under ideal conditions.
• The study highlighted the significance of separating electromagnetic and muonic components in the data.
• Feature selection identified crucial features for improved classification accuracy.

Conclusions:

• Machine learning offers a promising approach to solving the cosmic ray primary identification problem.
• Further research with more refined simulations is warranted.
• The electromagnetic-muonic separation is a key factor for future advancements.